1 INFO-VAX	Thu, 28 Feb 2002	Volume 2002 : Issue 116       Contents:6 Re: Anyone See the full page ad letter to Mr. Hewlett?& Re: Command line length limit and LINK+ Re: Compaq adopts microsoft approach to VMS + Re: Compaq adopts microsoft approach to VMS + Re: Compaq adopts microsoft approach to VMS 6 Re: Compaq hardware and software maintenance providers copying system disks Re: copying system disks Re: decnet phase IV -> phase V Re: decnet phase IV -> phase V Re: decnet phase IV -> phase V Re: decnet phase IV -> phase V Re: decnet phase IV -> phase V Re: decnet phase IV -> phase V" DECWindows - A lot of problems ...* define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.. Re: define "group-logicals" at system startup.& Re: DEQNA/DELQA/DESQA - More questions DTE Unsynchronised Re: expensive procedure calls  Re: expensive procedure calls  Re: expensive procedure calls  Re: expensive procedure calls  Re: expensive procedure calls 2 HP Sees Charges Up To $1.4 bln After Compaq Merger6 Re: HP Sees Charges Up To $1.4 bln After Compaq Merger( HSD30 licensing with mirroring and Cache Re: I: PAKGEN Software for ISVs  Re: I: PAKGEN Software for ISVs  Re: I: PAKGEN Software for ISVs = Re: IBM says they have best real time mainframe, where's VMS?  Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles Re: Itanium troubles' Microsoft Curries Favor With Undergrads - Re: Need help on Alpha:  Calling C from MACRO - Re: Need help on Alpha:  Calling C from MACRO A OpenVMS get together in Houston - thought you might be interested . Oracle Rdb Technical Update in NY - InvitationL Re: Prior Version Support (was: Re: Compaq adopts microsoft approach to VMS)L Re: Prior Version Support (was: Re: Compaq adopts microsoft approach to VMS)B Q: Why is XP-1000 RAM so expensive from Compaq vis-a-vis Kingston?J Question on a Deskpro EP`s motherbaord and possibily of changing it out :) restricting access to ports  Re: restricting access to ports  Re: restricting access to ports  Re: restricting access to ports  Re: restricting access to ports  RMS: rop: RAB$M_LIM  Re: RMS: rop: RAB$M_LIM  Re: Solution for the merger  Re: Sound on alpha/VMS? # Steps for installation OVMS 6.2 1H3 ' Re: Steps for installation OVMS 6.2 1H3 ' Re: Steps for installation OVMS 6.2 1H3 ' Re: Steps for installation OVMS 6.2 1H3 ' Re: Steps for installation OVMS 6.2 1H3  Use of PCA on Alpha % Re: vax/vms documentation - psi / x29  Re: Very slow ftp  Re: Very slow ftp  Re: Very slow ftp ) Re: VMS single Login with NDS e-directory # Re: What return codes mean success? # Re: What return codes mean success? # Re: What return codes mean success? # Re: What return codes mean success? & Re: Why does ERRFMT give me RSZ error?& Re: Why does ERRFMT give me RSZ error?& Re: Why does ERRFMT give me RSZ error?& Re: Why does ERRFMT give me RSZ error?C Re: Why is XP-1000 RAM so expensive from Compaq vis-a-vis Kingston? C Re: Why is XP-1000 RAM so expensive from Compaq vis-a-vis Kingston?  YA Amusing Message from DEC. Re: [Q] internet and VMS  F ----------------------------------------------------------------------  % Date: Thu, 28 Feb 2002 08:28:28 -0800 ' From: David Mathog <mathog@caltech.edu> ? Subject: Re: Anyone See the full page ad letter to Mr. Hewlett? + Message-ID: <3C7E5AAC.890D4769@caltech.edu>    Bill Todd wrote: > M > Interesting - especially as IIRC the personal attacks were started by those 0 > (Carly and the BoD) *supporting* the merger... >   F Carly and the rest of the HP board are definitely big believers in the# "shoot the messenger" philosophy.     A Hewlett's suggestion to break up/spin off HP has a lot more merit F than Carly and the HP board would like us to believe.  If nothing elseE it's much easier to spin off units into separate companies than it is K to mash together different companies into one coherent whole.  There's very B little synergy between the different business units in HP. Kind ofA like Compaq, where the PC division could easily be split off from ? the enterprise side - which in itself does hang together pretty M well as high margin/heavy iron/heavy storage/megabuck support.  For instance, B all ipaq devices could suddenly cease to exist and the only effectN this would have on the VMS division would be a decrease in the amount of moneyA that would need to be siphoned off to support that other product.    Regards,   David Mathog mathog@caltech.edu   ------------------------------   Date: 28 Feb 2002 09:18 CST ' From: carl@gerg.tamu.edu (Carl Perkins) / Subject: Re: Command line length limit and LINK - Message-ID: <28FEB200209182957@gerg.tamu.edu>   g In article <Hubf8.29360$7a1.2003704@bin5.nnrp.aus1.giganews.com>, Ernest <wmozart5@yahoo.com> writes... K }I'm struggling to find a way around the tiny command line length limit on  J }OpenVMS/Alpha 7.2-1, and would greatly appreciate any tips on how to get  }around it.  } L }The commands I am wrestling with are LINK and CXXLINK, while creating both L }executable programs and shareable images.  It boils down to having quite a J }large number of object files, which pushes the total line length to over  }the 1024 limit. } K }Is there any way around this?  Like, for example, response files (Windows  B }compilers generally support this sort of thing, as does the Java L }compiler)?  Or creating the library/program incrementally (which is what I F }have to do to create object libraries, and thankfully LIBRARY can be K }called multiple times with /INSERT to incrementally build the library one  % }object at a time).  Any other ideas?  }  }Thanks, }Ernest   ) One or more libraries are the usual way.    9 Insert (or /replace) each object file into a library upon  successful compilation.   E Alternatively, you can produce fewer object files by compiling groups C of source files to produce a single object file for the group (this C can also get you some additional optimizations between parts of the B program in different source files, the exact details of doing thisB depend on the compiler - C has a /PLUS_LIST_OPTIMIZE qaulifier and@ Fortran always does it when the input files are separated by "+"8 instead of ",", but I see nothing for the C++ compiler).   --- Carl   ------------------------------    Date: 28 Feb 2002 02:52:46 -0800) From: P.Young@unsw.EDU.AU (Patrick Young) 4 Subject: Re: Compaq adopts microsoft approach to VMS< Message-ID: <55f85d77.0202280252.6b87953@posting.google.com>  w antony.wardle@optusnet.com.au (Antony Wardle) wrote in message news:<fe52053.0202261622.4183b2e7@posting.google.com>...  > G > How many people upgrade the minute a new version is out? The hobbiest F > probably, but what about the business customers who are running 24*7A > on critical systems. The upgrade planning can take a few months  > not to mention the testing.  >   A Being an .EDU.AU type site - *not during session*. Middle of year C and end of year are "open season" :-) time permitting. Then again I C take a copy of the running system, upgrade it, test it, and slot it , back into production some lazy Friday night.  ? I don't have any old OpenVMS version dependant software though.   O       *  December 21, 2001. OpenVMS Upgraded to Version 7.3 & product upgrades.   ;          Digital BASIC                V1.3 upgraded to V1.4 ?          Digital BLISS                V1.8 upgraded to V1.10-30 
          .
          .
          .          etc, etc.  8 At home, OTOH, any time is Open(VMS) upgrade season :-).  E That said, I'm behind with CSWS (some local mods to the source are in ? place on the old version) and PMDF updates, which is not really  excusable on my part.   B Then again I'm behind with a number of other projects (some vmstar new features included).    ------------------------------    Date: 28 Feb 2002 03:15:54 -0800) From: P.Young@unsw.EDU.AU (Patrick Young) 4 Subject: Re: Compaq adopts microsoft approach to VMS= Message-ID: <55f85d77.0202280315.2d7da451@posting.google.com>   w antony.wardle@optusnet.com.au (Antony Wardle) wrote in message news:<fe52053.0202261622.4183b2e7@posting.google.com>...  > A > on critical systems. The upgrade planning can take a few months  > not to mention the testing.  >   = Forgot to mention, from my previous reply "At home, OTOH, any * time is Open(VMS) upgrade season :-)" ....  @ A home/hobby OpenVMS system is important for any OpenVMS Systems9 Manager because it tends to get the "latest and greatest" A immediately it comes out (even before you officially get your own ! copy of the software at work :-).   @ I usually find the upgrade at work _very_ smooth because by that? time I have already installed, and played at home to my heart's  content.   ------------------------------    Date: 28 Feb 2002 05:40:36 -0800( From: bob@instantwhip.com (Bob Ceculski)4 Subject: Re: Compaq adopts microsoft approach to VMS= Message-ID: <d7791aa1.0202280540.541ed216@posting.google.com>   ] John Santos <JOHN@egh.com> wrote in message news:<1020227185421.3430B-100000@Ives.egh.com>... % > On 27 Feb 2002, Bob Ceculski wrote:  > r > > "Fred Kleinsorge" <kleinsorge@star.zko.dec.com> wrote in message news:<D0Ye8.237$fL6.5571@news.cpqcorp.net>...R > > > Beyond a certain point, it becomes very costly to debug and fix a problem inM > > > an old release.  Prior version support allows you to get the fix if you N > > > really need it, and want to pay for the engineering.  If you want normalK > > > maintenance, then you need to be running a release that is relatively N > > > current.  Note that we support V7.2-2, and will for some time to come...Q > > > how many years did we keep a V7.1x version current - quite a few.  It isn't K > > > like we are obsoleting releases every other day.  V7.2-2 was built to Q > > > consolidate all the fixes and hardware support in the V7.2 releases, and to D > > > *not* force people into upgrading to a new functional release. > > > R > > > We *don't* expect people to upgrade immediately, which is why we support theR > > > current release, and an earlier release (7.3 and 7.2-2 today).   But we willQ > > > eventually release a new functional release, and then V7.3x and the current - > > > release will become the supported pair.  > > >  > > K > > that is not true ... we would like to get the decwindows cert patch for I > > 7.1-1H2 and is only available for 7.1-2 ... that is a forced upgrade!  > ) > What is 7.1-1H2????  Never heard of it.  > H > The DECwindows ECO is available for Alpha V6.2 (all variants), V7.1-2,C > V7.2-1, V7.2-1H1, V7.2-2, V7.3 and VAX V6.2, V7.1, V7.2, V7.3 and " > SEVMS V6.2 (both Alpha and VAX).  4 there are versions 7.1-1, 7.1-1H1, 7.1-1H2 out here!   ------------------------------  % Date: Thu, 28 Feb 2002 09:45:56 +0000 - From: Roy Omond <Roy.Omond@BlueBubble.UK.Com> ? Subject: Re: Compaq hardware and software maintenance providers 1 Message-ID: <3C7DFC54.67620140@BlueBubble.UK.Com>    Doug Sharp wrote:   N > Does anyone in the U.K. have any experience of  Maindec or Perseus providingL > hardware/software maintenance support on their OpenVMS systems. Any input,$ > good or bad, would be appreciated.  J Well, yes, I do (I do quite a bit of contract work for Perseus, so my view% contains a certain amount of bias :-)   	 Roy Omond  Blue Bubble Ltd.   ------------------------------  + Date: Thu, 28 Feb 2002 15:56:37 +0100 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>  Subject: copying system disks ; Message-ID: <01KET9GQMAMA8ZL57Q@sysdev.deutsche-boerse.com>   F It seems to me to be a waste of time to maintain more than one system F disk if they are more or less similar (same VMS version, same layered G products installed etc).  Especially if one wants to remain up-to-date  F on patches etc, some of which require a reboot after installation.  A I better approach would seem to be to have a "master system disk" which is  A maintained.  This could then be copied to the other system disks  I whenever they need to be updated.  (If the system disks are shadow sets,  I one could dismount the master system disk and remove one member and then  I use this as the new system disk for another system; after it is mounted,  A the remaining disk(s) in the shadow set on the second system are  ; mounted, and the extra disk goes back to the first system.)   E There are basically five things to keep in mind to transform a fresh  H copy of the master system disk to an updated copy of the system disk on 3 the target system (or have I forgotten something?):    1:H The node name needs to be changed.  Unfortunately, it shows up in a lot E of places; the FAQ has some information on this.  Still, I'd like to  ) have a working CHANGE_NODE_NAME.COM.  :-)    2:8 Network information (DECnet, TCPIP) needs to be changed.   3:A If the system parameters are different, an AUTOGEN should be run.    4:) Files like SYSUAF.DAT need to be changed.    5:F The log files should be the ones appropriate for the target disk, not  the master system disk.   
 3 is trivial.   E 4 is not a problem if these files (I think the FAQ has a list), i.e.  F files other than log files which change after VMS is installed and in H some sense make up the "personality" of the system, are located off the I system disk (with the appropriate logicals defined), which is often done    in a cluster environment anyway.  I 1 and 2 are a pain, but in principle could be automated, if not as a DCL  > procedure at least as a step-by-step list of tasks to be done.  F 5 seems to be the most difficult.  (ACCOUNTING.DAT and the like would I also belong to this group.)  I suppose one could back them up before the  I upgrade and then restore then afterwards, but this is not a very elegant  	 solution.   F Does anyone actually maintain system disks like this?  In particular,  what about point 5?   A Since the files in 5 are all(?) located in directories which are  C normally referred to by logicals which translate to search list, I  > suppose one could move them to another disk and re-define the E appropriate logicals.  This is certainly not supported and I have no  % idea if it would work (in all cases).   I All of this would be no problem, of course, if VMS distinguished between  G static and dynamic files at a really low level, the dynamic ones being  E the "personality" files which grow with the system.  Of course, a VMS G upgrade doesn't do a fresh install but just updates some files, that is G not the problem.  The problem is to avoid the work of applying the same H patches to many machines.  Also, there could be a third category, namelyE user-specific static files.  These would be, for example, things run  F during startup which are common to all the system disks one maintains.I (User-defined dynamic files, it seems to me, should not be on the system  
 disk at all.)    ------------------------------   Date: 28 FEB 2002 16:56:11 GMT4 From: karcher@thuria.waisman.wisc.edu (Carl Karcher)! Subject: Re: copying system disks 6 Message-ID: <28FEB02.16561178@thuria.waisman.wisc.edu>  K If you had a separate system root (i.e. sys1,sys2...syse) for each system's 0 node specific stuff it might make things easier.  H I maintain two systems disks in a two node scsi cluster with HSZ storageE in order to avoid a cluster shutdown during upgrades. I've toyed with C using a common shadowed system disk on the HSZ but then I'd have to D perform extra steps during upgrades (remove a member from the shadowE set, boot the node to be upgraded from the former shadow member, then F more musical disks to boot the other node from the upgraded disk). TooH many reboots for me. I don't think maintaining two systems disks is that* bad. More than two and I might reconsider.  Q In a previous article, Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> wrote:pH ->It seems to me to be a waste of time to maintain more than one system H ->disk if they are more or less similar (same VMS version, same layered I ->products installed etc).  Especially if one wants to remain up-to-date tH ->on patches etc, some of which require a reboot after installation.  A K ->better approach would seem to be to have a "master system disk" which is aC ->maintained.  This could then be copied to the other system disks oK ->whenever they need to be updated.  (If the system disks are shadow sets, -K ->one could dismount the master system disk and remove one member and then  K ->use this as the new system disk for another system; after it is mounted, tC ->the remaining disk(s) in the shadow set on the second system are  = ->mounted, and the extra disk goes back to the first system.)o -> sG ->There are basically five things to keep in mind to transform a fresh  J ->copy of the master system disk to an updated copy of the system disk on 5 ->the target system (or have I forgotten something?):t -> . ->1:J ->The node name needs to be changed.  Unfortunately, it shows up in a lot G ->of places; the FAQ has some information on this.  Still, I'd like to d+ ->have a working CHANGE_NODE_NAME.COM.  :-)v -> c ->2:: ->Network information (DECnet, TCPIP) needs to be changed. -> a ->3:C ->If the system parameters are different, an AUTOGEN should be run.C ->   ->4:+ ->Files like SYSUAF.DAT need to be changed.m -> e ->5:H ->The log files should be the ones appropriate for the target disk, not  ->the master system disk.r ->   ->3 is trivial.o -> eG ->4 is not a problem if these files (I think the FAQ has a list), i.e.  H ->files other than log files which change after VMS is installed and in J ->some sense make up the "personality" of the system, are located off the K ->system disk (with the appropriate logicals defined), which is often done t" ->in a cluster environment anyway. -> rK ->1 and 2 are a pain, but in principle could be automated, if not as a DCL  @ ->procedure at least as a step-by-step list of tasks to be done. -> eH ->5 seems to be the most difficult.  (ACCOUNTING.DAT and the like would K ->also belong to this group.)  I suppose one could back them up before the iK ->upgrade and then restore then afterwards, but this is not a very elegant i ->solution.a ->  H ->Does anyone actually maintain system disks like this?  In particular,  ->what about point 5?P -> tC ->Since the files in 5 are all(?) located in directories which are oE ->normally referred to by logicals which translate to search list, I  @ ->suppose one could move them to another disk and re-define the G ->appropriate logicals.  This is certainly not supported and I have no m' ->idea if it would work (in all cases).o -> tK ->All of this would be no problem, of course, if VMS distinguished between oI ->static and dynamic files at a really low level, the dynamic ones being  G ->the "personality" files which grow with the system.  Of course, a VMSoI ->upgrade doesn't do a fresh install but just updates some files, that issI ->not the problem.  The problem is to avoid the work of applying the same J ->patches to many machines.  Also, there could be a third category, namelyG ->user-specific static files.  These would be, for example, things run  H ->during startup which are common to all the system disks one maintains.K ->(User-defined dynamic files, it seems to me, should not be on the system X ->disk at all.)e ->     --G -- Carl Karcher, Waisman Computing Services, Waisman Center, UW-Madisona2 --                      karcher@waisman.wisc.edu     ------------------------------    Date: 28 Feb 2002 01:55:07 -0800% From: mb301@hotmail.com (Mark Bowman) ' Subject: Re: decnet phase IV -> phase Va= Message-ID: <1d08b916.0202280155.50605e01@posting.google.com>u   Jim   2 The command set node n.n name xzy is done by using $ decnet_register  -2 Register.  : Then you must flush the cache with the following command:-  5 $ mc ncl flush session control naming cache entry "*"t  @ If it still dosn't work work try running the nodename to address@ lookup program on another screen and find out what is happening.   $ mc CDI$TRACE.EXE;n [CDI$TRACE V2.2 - rev. b]s   10:49:38 DECnet/CDI V2.2 10:49:38% 10:49:38    Lookups processed:      8 % 10:49:38   Successful lookups:      6o% 10:49:38        Syntax checks:      0t% 10:49:38       Cache bypasses:      0i% 10:49:38   CDI trace sessions:      1r% 10:49:38 Requests in progress:      0t% 10:49:38      CDI misz errors:      0m= 10:49:38 Cache hit ratio for successful lookups(4/6) = 66.67%  10:49:38     ----# NCP> COPY KNOWN NODES FROM nodenamet*    USING PERMANENT TO PERMANENT WITH PURGE     Try  $ mc decnet_register 7- export to a filec Then use option 8 to import    -----e   Help with v4 to v5 commands.   TryS* $ mc SYS$COMMON:[SYSUPD]DECNET_MIGRATE.EXE1 DECNET_MIGRATE> convert command "show known node"p  2 ! *** Converting the following NCP command to NCL: !     SHOW KNOWN NODE-; SHOW NODE 0 ROUTING CIRCUIT CSMACD-0 ADJACENCY * ALL STATUS                                  |1 NB: Your circuit name maybe different, check withL# $ mc ncl show node 0 rout cir * allg   Regardsu Mark   ------------------------------  % Date: Thu, 28 Feb 2002 11:21:23 +0100m( From: Paul Sture <paul.sture@bluewin.ch>' Subject: Re: decnet phase IV -> phase V ) Message-ID: <3C7E04A3.6030203@bluewin.ch>1   Jim Strehlow wrote:r  - > How do you get from there (DECnet Phase IV) 2 > to here (DECnet Phase V/DECnet-Plus/DECnet-OSI)? > * > My favorite MCR NCP commands have become2 > menu items via sys$manager:net$configure.com and > mcr decnet_registerj) > (NCP> SET NODE n.n NAME nodename failedt3 > with %NCP-W-UNRCMP, unrecognized component, node)r > F > I used MCR decnet_register 's option 2 to configure a Phase IV node. >    After which you should issue:s  6 $ mcr ncl flush session control naming cache entry "*"    5 > It did not yet allow me to do a $ SET HOST nodenamet > to that Phase IV nodename. > = > I want to get to the point where I can do the equivalent ofb% > NCP> COPY KNOWN NODES FROM nodenamem, >    USING PERMANENT TO PERMANENT WITH PURGE >  > Any help is appreciated. > Is that an FAQ somewhere?/= > Is there a "guide to migrate Phase IV to Phase V/-Plus/OSI"i > (whatever its name is)?y > L You will find various folks trying to persuade you not to use Phase V, based  I on their initial experiences with it. It really isn't so bad once you geto   used to it.t   ------------------------------    Date: 28 Feb 2002 05:42:25 -0800( From: bob@instantwhip.com (Bob Ceculski)' Subject: Re: decnet phase IV -> phase Vd< Message-ID: <d7791aa1.0202280542.e256504@posting.google.com>  e "Jim Strehlow" <jimSnoSpam@data911.com> wrote in message news:<a5jq4g$mmn@dispatch.concentric.net>...-- > How do you get from there (DECnet Phase IV).2 > to here (DECnet Phase V/DECnet-Plus/DECnet-OSI)? > * > My favorite MCR NCP commands have become2 > menu items via sys$manager:net$configure.com and > mcr decnet_registert) > (NCP> SET NODE n.n NAME nodename failedw3 > with %NCP-W-UNRCMP, unrecognized component, node)  > F > I used MCR decnet_register 's option 2 to configure a Phase IV node. > 5 > It did not yet allow me to do a $ SET HOST nodenamer > to that Phase IV nodename. > = > I want to get to the point where I can do the equivalent ofi% > NCP> COPY KNOWN NODES FROM nodenameU, >    USING PERMANENT TO PERMANENT WITH PURGE >  > Any help is appreciated. > Is that an FAQ somewhere?T= > Is there a "guide to migrate Phase IV to Phase V/-Plus/OSI"s > (whatever its name is)?t >  > Thank you. > ) > Jim Strehlow, Data911, jimS@data911.comV > Alameda, CA, USA  < big mistake to migrate to phase v ... big learning curve ...   ------------------------------  # Date: Thu, 28 Feb 2002 14:55:40 GMT # From: "John N." <JNixon@cfl.rr.com> ' Subject: Re: decnet phase IV -> phase Ve< Message-ID: <Mxrf8.6667$1p6.1542930@typhoon.tampabay.rr.com>  5 "Paul Sture" <paul.sture@bluewin.ch> wrote in messages# news:3C7E04A3.6030203@bluewin.ch...i > Jim Strehlow wrote:s >   >H > You will find various folks trying to persuade you not to use Phase V, based  >uK > on their initial experiences with it. It really isn't so bad once you gets > 
 > used to it.t >sI The same can be said for creamed spinach or Unix or rape.  But why should4K one go through the pain and agony of "getting used to it" unless there is asF pressing need.  For most people, phase IV will do the job reliably and easily.g   ------------------------------    Date: 28 Feb 2002 10:09:44 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen)g' Subject: Re: decnet phase IV -> phase V.3 Message-ID: <Bf2TYf982R4g@eisner.encompasserve.org>o  b In article <Mxrf8.6667$1p6.1542930@typhoon.tampabay.rr.com>, "John N." <JNixon@cfl.rr.com> writes: > 7 > "Paul Sture" <paul.sture@bluewin.ch> wrote in messages% > news:3C7E04A3.6030203@bluewin.ch...s >> Jim Strehlow wrote: >> >  >I >> You will find various folks trying to persuade you not to use Phase V,i > basedy >>L >> on their initial experiences with it. It really isn't so bad once you get >> >> used to it. >>K > The same can be said for creamed spinach or Unix or rape.  But why shouldtM > one go through the pain and agony of "getting used to it" unless there is amH > pressing need.  For most people, phase IV will do the job reliably and	 > easily.o  @ Some people like to prepare themselves for future possibilities,< such as the need to run some software that requires Phase V.   ------------------------------   Date: 28 Feb 2002 17:11:55 GMT- From: "Jim Strehlow" <jimSnoSpam@data911.com>@' Subject: Re: decnet phase IV -> phase VD0 Message-ID: <a5locr$8sp@dispatch.concentric.net>  * 1) thanks for the phase IV to phase V tips  D 2) we are probably going to install some FDDI cluster failover cable0   which requires phase V onto one remote system.E   So if I am going to install that soon, I need to "get up to speed".   @ 3) On a test node I first installed phase IV and got it working.+   I uninstalled that and installed phase V.-  6 4) I can not import node information from another node7   without first changing that phase IV node to phase V,sE   running the conversion, and then exporting the phase V information.s   (can not get there from here)   8   Instead, I will manually register the node information   onto my new test system.  * 5) I am happily going to maintain Phase IV!   on twenty+ other nodes for now.e   Again, thanks for sharing.   Jim Strehlow, Data911  Alameda, CA, USA   ------------------------------  + Date: Thu, 28 Feb 2002 10:53:45 -0800 (PST)r. From: Fabio Cardoso <fabiopenvms@yahoo.com.br>+ Subject: DECWindows - A lot of problems ...h? Message-ID: <20020228185345.2166.qmail@web20210.mail.yahoo.com>l  % I'm having some problems to configureo' the DECWindows with KEA X software from  Attachmate.t  * a) I can't run the DECW$STARTLOGIN program" at the Application Manager module.& I must open a telnet session first and run it. So it works fine.   3 b) After this, the user cannot open any application.- clicking in the CDE icons. I see the programs.6 being started by $ SHOW USERS, but the MBA (mailboxes) are "disconnected".-    2 c) Where can I configure a normal user to login at1 the DECWindows? For example, an account with OPERj- is being denied - at least I understand this:c( Appears a message after the username and password login: invalid login.    2 d) Where can I download DECWindows documentation ?, I have the OpenVMS Documentation CD-ROM, but4 I didnt find anything useful (configurations) in it.# I search the Wizard and same thing.d   Regardsa   FC     =====, ========================== Fbio dos Santos Cardoso OpenVMS System Manager Rio de Janeiro - Brazil  fabiopenvms@yahoo.com.br ==========================  2 __________________________________________________ Do You Yahoo!?8 Yahoo! Greetings - Send FREE e-cards for every occasion! http://greetings.yahoo.com   ------------------------------  % Date: Thu, 28 Feb 2002 08:44:03 +0100e- From: Jouk Jansen <joukj@hrem.stm.tudelft.nl>e3 Subject: define "group-logicals" at system startup. 3 Message-ID: <3C7DEDD3.71E37188@hrem.stm.tudelft.nl>    Hi all,i  G I would like to define some logicals for a group during system startup. ) In order to tdo this I put something likeo   $  create/name LNM$GROUP_000303t8 $  assign/table=LNM$GROUP_000303 anything group$anything  ! in sys$manager:sysstartup_vms.com G However this fails, since the first time one of the group members logon ) on the system the table is reinitialized. A  How do I tell the system to maintain the table created at systemo startup?                   Jouk   ------------------------------  % Date: Thu, 28 Feb 2002 03:16:04 -0500 - From: JF Mezei <jfmezei.spamnot@videotron.ca>r7 Subject: Re: define "group-logicals" at system startup.n, Message-ID: <3C7DE742.87A8F732@videotron.ca>   Jouk Jansen wrote:! > $  create/name LNM$GROUP_000303 : > $  assign/table=LNM$GROUP_000303 anything group$anything > # > in sys$manager:sysstartup_vms.comt    # $SHOW LOG/FULL LNM$SYSTEM_DIRECTORY;. and HELP CREATE/NAME will give you good hints.  - I *think* it would have to be something like:.  G CREATE/NAME/PARENT=LNM$SYSTEM_DIRECTORY LNM$GROUP_000303/attrib=noalias:  M I am not sure because the HELP on CREATE/NAME doesn't quite specify this, but V you would also need to specify /KERNEL (if available), a /PROT=(RWCD,R,R,)/OWNER=[5,*]  B $SHOW LOG/TABLE=LNM$GROUP_000001/FULL  will give you an idea also.   ------------------------------  % Date: Thu, 28 Feb 2002 09:12:29 +0100 9 From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com>l7 Subject: Re: define "group-logicals" at system startup.y' Message-ID: <3C7DE66D.F0A7B736@aaa.com>w  ? Take a look at the following swithes for the cre/name command :c  ( /exec /parent_table=LNM$SYSTEM_DIRECTORY  - Not sure that it helps, but I'd try anyway...    Jan-Erik Sderholm.t     Jouk Jansen wrote: > 	 > Hi all,t > I > I would like to define some logicals for a group during system startup.o+ > In order to tdo this I put something likeo > ! > $  create/name LNM$GROUP_000303r: > $  assign/table=LNM$GROUP_000303 anything group$anything > # > in sys$manager:sysstartup_vms.comuI > However this fails, since the first time one of the group members logon + > on the system the table is reinitialized.nC >  How do I tell the system to maintain the table created at systems
 > startup? >  >                 Jouk   ------------------------------  % Date: Thu, 28 Feb 2002 09:09:19 +0100i2 From: "Ren Schelbaum" <rene.schelbaum@datakom.at>7 Subject: Re: define "group-logicals" at system startup.nG Message-ID: <3c7de5a8$0$36578$5039e797@newsreader01.highway.telekom.at>f  @ "Jouk Jansen" <joukj@hrem.stm.tudelft.nl> schrieb im Newsbeitrag- news:3C7DEDD3.71E37188@hrem.stm.tudelft.nl...e	 > Hi all,p >dI > I would like to define some logicals for a group during system startup. + > In order to tdo this I put something like  >l! > $  create/name LNM$GROUP_000303h: > $  assign/table=LNM$GROUP_000303 anything group$anything >i# > in sys$manager:sysstartup_vms.comrI > However this fails, since the first time one of the group members logoni+ > on the system the table is reinitialized.eC >  How do I tell the system to maintain the table created at system-
 > startup? >- >                 Jouk   Hi!0  F do a "run/uic=[UIC of that Group] / input=(command-file containing theL required group-logical definitions) sys$system:loginout.exe" during startup. that does the trick!   Ren   ------------------------------  % Date: Thu, 28 Feb 2002 10:30:58 +0200t' From: "Gabriel Sterk" <gabi@aipm.co.il>n7 Subject: Re: define "group-logicals" at system startup.Z2 Message-ID: <000a01c1c032$400d5d40$2c46bf10@manai>  9 We define group logicals at system startup by running thed9 a command file, maintained and stored by the group in theu following (laziest) way:  M $ RUN/UIC=useruic_having_grpprv/INPUT=the_group's_GROUP_DEFS.COM/OUTPUT=NL: -mK                                                         SYS$SYSTEM:LOGINOUT.   Works for 10 years.m  
 Gabriel Sterkm   ------------------------------   Date: 28 Feb 2002 07:59:28 GMT3 From: gartmann@immunbio.mpg.de (Christoph Gartmann) 7 Subject: Re: define "group-logicals" at system startup.t0 Message-ID: <a5ko10$g6k$2@n.ruf.uni-freiburg.de>  c In article <3C7DEDD3.71E37188@hrem.stm.tudelft.nl>, Jouk Jansen <joukj@hrem.stm.tudelft.nl> writes:lH >I would like to define some logicals for a group during system startup.* >In order to tdo this I put something like >t  >$  create/name LNM$GROUP_0003039 >$  assign/table=LNM$GROUP_000303 anything group$anythingn >n" >in sys$manager:sysstartup_vms.comH >However this fails, since the first time one of the group members logon* >on the system the table is reinitialized.B > How do I tell the system to maintain the table created at system	 >startup?   F $ ! Assign a group-logical for B. Herrmanns group and sequence projectD $ RUN/DETACHED/UIC=[2400,1]/INPUT=NL:/OUTPUT=NL: sys$system:loginout! $ !    => lnm$group_002400 existsr> $ ASSIGN/TABLE=lnm$group_002400 udsk1:[herrmann.common] labseq   Regards,    Christoph Gartmanny  H -- --------------------------------------------------------------------+H | Max-Planck-Institut fuer      Phone   : +49-761-5108-464   Fax: -452 |H | Immunbiologie                                                        |H | Postfach 1169                 Internet: gartmann@immunbio.mpg.de     |H | D-79011  Freiburg, FRG                                               |H +--------- http://www.immunbio.mpg.de/home/english/menue.html ---------+   ------------------------------  % Date: Thu, 28 Feb 2002 10:52:58 +0100A- From: Jouk Jansen <joukj@hrem.stm.tudelft.nl>i7 Subject: Re: define "group-logicals" at system startup.o3 Message-ID: <3C7E0C0A.219EF721@hrem.stm.tudelft.nl>o   Christoph Gartmann wrote:i > e > In article <3C7DEDD3.71E37188@hrem.stm.tudelft.nl>, Jouk Jansen <joukj@hrem.stm.tudelft.nl> writes:iJ > >I would like to define some logicals for a group during system startup., > >In order to tdo this I put something like > >t" > >$  create/name LNM$GROUP_000303; > >$  assign/table=LNM$GROUP_000303 anything group$anythingr > >a$ > >in sys$manager:sysstartup_vms.comJ > >However this fails, since the first time one of the group members logon, > >on the system the table is reinitialized.D > > How do I tell the system to maintain the table created at system > >startup?o > H > $ ! Assign a group-logical for B. Herrmanns group and sequence projectF > $ RUN/DETACHED/UIC=[2400,1]/INPUT=NL:/OUTPUT=NL: sys$system:loginout# > $ !    => lnm$group_002400 exists @ > $ ASSIGN/TABLE=lnm$group_002400 udsk1:[herrmann.common] labseq > 
 > Regards, >    Christoph Gartmanni >  > -- e    Thanks, I'll try this one.o                   Jouk   ------------------------------  % Date: Thu, 28 Feb 2002 11:58:43 +0100f( From: Paul Sture <paul.sture@bluewin.ch>7 Subject: Re: define "group-logicals" at system startup.u) Message-ID: <3C7E0D63.7060906@bluewin.ch>i   Jouk Jansen wrote:  	 > Hi all,f > I > I would like to define some logicals for a group during system startup. + > In order to tdo this I put something like  > ! > $  create/name LNM$GROUP_000303o: > $  assign/table=LNM$GROUP_000303 anything group$anything > # > in sys$manager:sysstartup_vms.comrI > However this fails, since the first time one of the group members logono+ > on the system the table is reinitialized. C >  How do I tell the system to maintain the table created at systeme
 > startup? > S I tried to do similar when I first encountered the "new style" logicals in VMS 4.0.e    I The neatest solution was to do a SUBMIT/USER= in a suitable startup file aE (make sure any relevant disks are online first) for a member of that mB group who possesses the GRPNAM privilege. Create a "group leader" H account if necessary just to perform the function of creating the group # table and populating it at startup.m  H Other group members no longer require GRPNAM, since you now know who is ! the first to log in after a boot.     
 Paul Sture Switzerlande   ------------------------------  % Date: Thu, 28 Feb 2002 12:21:03 +0100u( From: Paul Sture <paul.sture@bluewin.ch>7 Subject: Re: define "group-logicals" at system startup.n) Message-ID: <3C7E129F.9000306@bluewin.ch>y   Gabriel Sterk wrote:  ; > We define group logicals at system startup by running the ; > a command file, maintained and stored by the group in thes > following (laziest) way: > O > $ RUN/UIC=useruic_having_grpprv/INPUT=the_group's_GROUP_DEFS.COM/OUTPUT=NL: - M >                                                         SYS$SYSTEM:LOGINOUT  >  > Works for 10 years.s >  > Gabriel Sterk4 > - That should surely read useruic_having_grpnamD   ------------------------------    Date: 28 Feb 2002 03:50:24 -0800) From: P.Young@unsw.EDU.AU (Patrick Young)C7 Subject: Re: define "group-logicals" at system startup.M= Message-ID: <55f85d77.0202280350.4df062d3@posting.google.com>a  h Jouk Jansen <joukj@hrem.stm.tudelft.nl> wrote in message news:<3C7DEDD3.71E37188@hrem.stm.tudelft.nl>... > I > I would like to define some logicals for a group during system startup.  >   A If there is an account in the affected group that has GRPNAM (for E example a "Group Manager" account), ie: ['GRP_NUMBER',0] or whatever,  you can do something like:  A $ RUN/UIC=['GRP_NUMBER',0]/INPUT=SYS$STARTUP:DEFONEGRP/OUTPUT=NL:r  SYS$SYSTEM:LOGINOUT  A For each of your groups during system startup where DEFONEGRP.COM2, could do the appropriate DEFINE/GROUP stuff.   ------------------------------  % Date: Thu, 28 Feb 2002 12:57:10 +0100A9 From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com>y7 Subject: Re: define "group-logicals" at system startup.h' Message-ID: <3C7E1B16.C013E7A3@aaa.com>f  > There have been a number of post on this issue now, all giving0 RUN/UIC=xxx or SUBMIT/USER=yyy as the way to go.  ; Isn't there a way for SYSTEM to directly create this table  4 without using one of the users in the actual group ?  	 Such as :    $ create /name_table -          /exec --          /parent_table=LNM$SYSTEM_DIRECTORY -l          LNM$GROUP_zzzzzzc  6 Is there some problem with the "owner" of this table ?   Jan-Erik Sderholm.t   ------------------------------    Date: 28 Feb 2002 07:17:56 -06002 From: cochrane@encompasserve.org (Arthur Cochrane)7 Subject: Re: define "group-logicals" at system startup. 3 Message-ID: <Goh4toB2UF5x@eisner.encompasserve.org>e  0     Execute the following from SYSTARTUP_VMS.COM  K     Note the RUN command uses the group UIC number with a member UIC numberr     of zero. This works great.  
 $ set noon2 $ file=f$element(0,";",f$environment("PROCEDURE"))F $ group_file=f$parse("decnet_group_logicals.com",file,,,"SYNTAX_ONLY")" $ if f$search(group_file) .nes. "" $ then $	group:=decneti $	run sys$system:loginout.exe -: 		/input='group_file' -  		/output=nla0: -7 		/error=nla0: - 		/noauthorize - 		/noaccounting -r 		/privileges=(grpnam,sysnam) -r" 		/process_name="Group Logicals" - 		/uic=['group',0] $ endifo $ exit  J     This is the command procedure run above to define group logical names.  < $! Define in DECnet group logical name table for FAL$SERVER . $ define/group/nolog	decus_cd4_disk	decus_cd4:. $ define/group/nolog	decus_cd5_disk	decus_cd5:. $ define/group/nolog	decus_cd6_disk	decus_cd6:. $ define/group/nolog	decus_cd7_disk	decus_cd7:. $ define/group/nolog	decus_cd8_disk	decus_cd8:. $ define/group/nolog	decus_cd9_disk	decus_cd9:0 $ define/group/nolog	decus_cd10_disk	decus_cd10:0 $ define/group/nolog	decus_cd11_disk	decus_cd11:0 $ define/group/nolog	decus_cd12_disk	decus_cd13:0 $ define/group/nolog	decus_cd13_disk	decus_cd13:0 $ define/group/nolog	decus_cd14_disk	decus_cd14: $ exit   ------------------------------    Date: 28 Feb 2002 07:31:27 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) 7 Subject: Re: define "group-logicals" at system startup.,3 Message-ID: <7p9b8KjweUJY@eisner.encompasserve.org>r  c In article <3C7E1B16.C013E7A3@aaa.com>, Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com> writes:.@ > There have been a number of post on this issue now, all giving2 > RUN/UIC=xxx or SUBMIT/USER=yyy as the way to go. > = > Isn't there a way for SYSTEM to directly create this table  6 > without using one of the users in the actual group ? >  > Such as :o >  > $ create /name_table - >          /exec -/ >          /parent_table=LNM$SYSTEM_DIRECTORY -l >          LNM$GROUP_zzzzzzl  B Normally group name tables are in kernel mode.  You can experimentC to find out what might go wrong when LOGINOUT creates a kernel moderB table when an exec mode table already exists.  Probably I tried it? at one time in the deep dark past, but even if I remembered they9 symptoms might be different with current versions of VMS.8  1 What I have used since 1986 in SYSTARTUP*.COM is:   K $ RUN SYS$SYSTEM:LOGINOUT/INPUT=NL:/OUTPUT=NL:/ERROR=NL:/UIC=[37774,177776]Y$ <unrelated stuff that takes up time>: $ SET ACL/OBJECT=LOGICAL/ACL=((ID=[37774,*],ACCESS=READ),-H                                 (ID=[*,*],ACCESS=NONE)) LNM$GROUP_037774@ $ DEFINE/TABLE=LNM$GROUP_037774 HLP$LIBRARY DISK$XYZ:[000000]XYZ  @ Technically the ACL definition is only required for those of you  planning on running VMS V4.2 :-)   ------------------------------    Date: 28 Feb 2002 17:46:13 +0100* From: eplan@kapsch.net (Peter LANGSTOEGER)7 Subject: Re: define "group-logicals" at system startup. * Message-ID: <3c7e5ed5$1@news.kapsch.co.at>  c In article <7p9b8KjweUJY@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes:pd >In article <3C7E1B16.C013E7A3@aaa.com>, Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com> writes:A >> There have been a number of post on this issue now, all giving 3 >> RUN/UIC=xxx or SUBMIT/USER=yyy as the way to go.8 >> 2> >> Isn't there a way for SYSTEM to directly create this table 7 >> without using one of the users in the actual group ?a >> $ >> Such as : >> t >> $ create /name_table -h >>          /exec - 0 >>          /parent_table=LNM$SYSTEM_DIRECTORY - >>          LNM$GROUP_zzzzzz >c/ >Normally group name tables are in kernel mode.i  ; Yup. That's the problem with CREATE/NAME_TABLE in this case G unless Q finally manages it to add a /KERNEL_MODE to CREATE/NAME_TABLE.   L This discussion/problem pops up quite regularly, so I assume it's in the FAQI (but the answer why DEQ didn't include /KERNEL_MODE so far is surely not)s   -Peter   -- o< Peter "EPLAN" LANGSTOEGER           Tel.    +43 1 81111 2651; Network and OpenVMS system manager  Fax.    +43 1 81111 888s< KAPSCH AG      Wagenseilgasse 1     E-mail  eplan@kapsch.netH A-1121 VIENNA  AUSTRIA              "I'm not a pessimist, I'm a realist"   ------------------------------  % Date: Thu, 28 Feb 2002 16:19:41 +0100  From: "B. Eckstein" <be@cli.de> / Subject: Re: DEQNA/DELQA/DESQA - More questionsm% Message-ID: <3C7E4A8D.4050707@cli.de>b   sword7@speakeasy.org aeusserte:p    H > First layer looks like PDP-11 code, others looks like different codes,H > etc.  Does anyone know about that?  I do not know which VAX code is inE > layers yet.  Will DESQA rom work in DEQNA or DELQA or do DEQNA and  % > DELQA require different ROM images?u  / DELQA has a MC68000, 10MHz as Controller on it.o. I don't have DESQA or DEQNA to look at, sorry.   -- p  B.Eckstein, - mailto:be@epost.de5 FAQ zu de.comp.hardware.netzwerke: http://how.to/dchn,   ------------------------------    Date: 28 Feb 2002 08:22:13 -08005 From: pat.saunders@sis.securicor.co.uk (pat saunders)x Subject: DTE Unsynchronised = Message-ID: <bc0e3bd8.0202280822.38590801@posting.google.com>g   Hi, &   I am using DECNET/OSI on a MicroVax,$   I am using a LAPB circuit on DTE-0D   When I look at state of DTE , I find it is "Unsynchronised" , What1 is the cause of this message and how do I fix it,n  :    I find that the LAPB link goes up and down like a yo-yo    :%    Message from user SYSTEM on KEWG99 B Event: Link Setup Failed from: Node LOCAL:.KEWG99 LAPB Link DTE-0,-         at: 2002-02-28-17:16:00.693+00:00Iinf=7         eventUid   D645CF84-2C6E-11D6-8008-AA0004009E2A_7         entityUid  3718B880-2C6E-11D6-8007-AA0004009E2Af7         streamUid  61C47CE0-2C6E-11D6-8008-AA0004009E2A-     NCL>8 %%%%%%%%%%%  OPCOM  28-FEB-2002 17:16:00.75  %%%%%%%%%%%" Message from user SYSTEM on KEWG99B Event: Link Initialising from: Node LOCAL:.KEWG99 LAPB Link DTE-0,-         at: 2002-02-28-17:16:00.693+00:00Iinf .         Link Initialising Reason=Maximum Retry7         eventUid   D645CF85-2C6E-11D6-8008-AA0004009E2A 7         entityUid  3718B880-2C6E-11D6-8007-AA0004009E2A07         streamUid  61C47CE0-2C6E-11D6-8008-AA0004009E2At  
 Any ideas? Patg   ------------------------------  % Date: Thu, 28 Feb 2002 09:45:57 +0100 3 From: Terje Mathisen <terje.mathisen@hda.hydro.com>i& Subject: Re: expensive procedure calls- Message-ID: <3C7DEE45.C1F8A4BA@hda.hydro.com>?   Bruce Hoult wrote: > B > In article <rdqp7u408ish7rimmmmu61alst8971cl33@4ax.com>, Alberto* > Moreira <junkmail@moreira.mv.com> wrote:D > > So, yes, there can be such a thing as a procedure call overhead. > H > Yes, but fortunately the cases where it is a significant proportion ofH > the total runtime are usually exactly those where it can be reduced or > eliminated :-)  5 I've done something almost unheard of in this regard:M  B I've actually tried multiple different implementation methods, allH written in both C and asm to get full control, and then timed them, just to find the fastest method.3  G Yeah, I know that this is bad 'cause it can stop a nice discussion, but 7 I wanted to win a competition, so please excuse me. :-)/  F What I had was a classic recursive function, with a known maximum call2 depth (11 or 12, depending upon how it was setup).  G The recursive function was working with a set of 4 parameters, the last E of them was the call depth counter to determine when to exit, and the:/ three first was combined into a 96-bit bitmask.s  F I tried full unrolling, which got rid of the fourth parameter, as well< as manual stack handling, but the fastest approach was plainE CALL/PUSHA/.../POPA/RET recursion, saving _all_ registers on entry tor0 the function, and restoring them on the way out.  H Later on I improved this by a few percent by splitting the function into two:  F One that worked like the one above, but without any checking to see ifG the innermost level was reached: Instead it called a smaller version ofgF itself when it was possible to determine that only 64 or less bits was still active in that bitmask.   H A few years later the C version of this code, which was written to be asH portable as possible, was entered (not by me, I only learned about it inB a Google search a short while ago! :-) in a similar competition inD France, where it was also twice as fast as the second-fastest entry.  E Anyway, the important part is that plain recursion can indeed be verysH fast, in my case it was almost certainly due to keeping the code working? set really small, since the data was the same for all versions.e   TerjeD  C PS. Here's the original (simplified) C version, notice the enormouse amount of comments! :-)g# Anyone wants to guess what it does?d  . void bTry(ul b0, ul b1, ul b2, int piecesleft) {/ 	int p;G
 	ul *pbm, bm;    	while (((long) b0) < 0) { 		b0 += b0 + (b1 >> 31); 		b1 += b1 + (b2 >> 31); 		b2 += b2;o 	}   	p = --piecesleft; 	do {h 		pbm = bPieces[p];h# 		bPieces[p] = bPieces[piecesleft];e 		bPieces[piecesleft] = pbm;   		bm = *pbm++; 		do { 			if ((bm & b0) == 0) {
 				b0 |= bm;e 				bsoltab[piecesleft] = bm;   + 				if (piecesleft == 0) goto haveSolution;t  ! 				bTry(b0, b1, b2, piecesleft);i
 				b0 ^= bm;d 			} 			bm = *pbm++;o 		} while (bm);g   		pbm = bPieces[p];u# 		bPieces[p] = bPieces[piecesleft];- 		bPieces[piecesleft] = pbm;   	} while (--p >= 0);   	return;  
 haveSolution:i 	bsaveSol(); 	return; }n     Here's -- f  - <Terje.Mathisen@hda.hydro.com>@ "almost all programming can be viewed as an exercise in caching"   ------------------------------  % Date: Thu, 28 Feb 2002 09:47:59 -0500n/ From: Alberto Moreira <junkmail@moreira.mv.com> & Subject: Re: expensive procedure calls8 Message-ID: <7lgs7uco27h35bur5mmo00ihatal1hdmlv@4ax.com>  0 Sander Vesik <sander@haldjas.folklore.ee>  said:    YA >No, you can in principle apply it to vanilla C - see for example-) >	http://www.sics.se/~psm/sparcstack.html2 > ; >it just depends on how your architecture looks like, what aB >assumptions are being made and similar. Also, in many processors,B >there is no such thing as a "procedure call" there are just jumpsD >and som conventions on how registers are treated and where you jump> >to when finished. Which definately allows you to do tail-call= >elimination, unles sthe conventions specifcly disallow that.e  tC Can't convert to tail form without doing what's in effect rewriting E the source. That applies to any language, C included. Hence the issuev@ with debugging. Try this: write a Quicksort algorithm in C, thenF convert it to tail form, and put the two pieces of source code side by side.      Alberto.   ------------------------------  % Date: Thu, 28 Feb 2002 10:12:30 -0500n/ From: Alberto Moreira <junkmail@moreira.mv.com>[& Subject: Re: expensive procedure calls8 Message-ID: <vpgs7ucdmqjoq7j1a2ga4724f4ii5hkj73@4ax.com>  4 Terje Mathisen <terje.mathisen@hda.hydro.com>  said:   >Alberto Moreira wrote:nG >> Also, the semantics of your programming language must be amenable to-F >> tail-form conversion. If the semantics of the language demands thatI >> you keep multiply nested stack frames with linkages to both statically I >> and dynamically nested contexts, like Pascal for example, I'm not even H >> sure conversion to tail form is feasible. You may also get in trouble > H >There's absolutely nothing that stops a Pascal compiler from generatingG >C-style code as long as the source code doesn't use any variables fromA >linked contexts, right?  A If a Pascal program is written to stay within the limits of the Cm8 semantics, sure, no problem. But try, for example, this:  " PROCEDURE ReadOneWord ( ........ )   	VAR c : char ;C   	PROCEDURE GetFirstNonBlankt   	  PROCEDURE IsItBlank                     BEGINi                       ......                     ENDu   	  BEGIN! 		if (ItsBlank) GetFirstNonBlank;  		c =  ........e 	  END     BEGIN	 	GetFirstNonBlank ( )  		.....i   ENDe   ENDO  C And you're already doing things that C doesn't do; all of a sudden,v? the semantics of that maligned ENTER/LEAVE instructions becomeso
 relevant !     Alberto.   ------------------------------  % Date: Thu, 28 Feb 2002 10:30:41 -0500a/ From: Alberto Moreira <junkmail@moreira.mv.com> & Subject: Re: expensive procedure calls8 Message-ID: <d9is7uok1puvqf2s13b2k0eocskk01cnhm@4ax.com>  $ Bruce Hoult <bruce@hoult.org>  said:    tI >Note that a program compiled with tail-call elimination is *exactly* as dF >hard (or easy) to debug as a program written with loops in the first 	 >place.     E Yes, of course. But the problem still remains, that we sometimes want-A to debug on a stepping basis at source line level. Transforming al5 program into tail form makes that nearly impossible. o  lI >With a compiler doing tail-call elimination these two functions produce  G >*identical* machine code.  The debugability is the same in both cases.a  D Not if I'm stepping at source level and not looking at object code.   G >But, turn off the tail-call elimination and the first version becomes $F >*more* debuggable, because you can examine the entire history of the 6 >computation by stepping up and down the stack frames.  B That's another aspect: the semantics of the object program follows: that of the source code, and that makes debugging easier.   tG >Pascal is identical to Scheme, Dylan and Common Lisp in this regard.  t >All are lexically scoped.  A I forget which flavor of Lisp is dynamically scoped, that changesa/ things. And you don't have callcc in Pascal ! 	    H >BTW, modern lexically-scoped language implementations don't use either G >of the methods commonly used when Pascal was developed (maintaining a fH >static link in each stack frame, or maintaining a "display").  Instead J >each function is passed an environment structure containing a pointer to B >each uplevel variable that it uses.  This is needed to implement E >closures (which Pascal doesn't have), but is more efficient overall r >anyway.  ? Yes, that's the Scheme way. But you see, a closure is packing aaF procedure definition with the environment in effect at the time of theB definition - because in Scheme, we must keep bindings that were inA effect at the time the procedure was created. So, a free variable E inside a procedure is resolved against the original environment(s) in/B effect at the time the procedure was created - but in Pascal, freeC variables are resolved with the environment(s) at the time they arek/ *used*. Hence, no need for closures in Pascal. s  lJ >Call-by-name is just the passing of a function (a closure actually) that H >accesses the appropriate value.  You can express this directly in many D >modern languages -- generally precisely those which have tail-call  >elimination, as it happens :-)o   Yes, that's true.  o  J@ >Dynamic binding is evil for many reasons :-)   The most common G >implementation these days (e.g. in Common Lisp's "special" variables, aJ >and Perl's "local" variables) is to use a single global varable and have F >each function save the original global value and restore it prior to  >returning.U >RH >This "restoring before returning" makes a function, by definition, not I >suitable for tail-call elimination, though if the function is otherwise hJ >self-recursive a compiler can collapse multiple levels of restoring into > >one, since most of the restored values are never used anwyay.   That's true too. 0  1D >Arbitrarily compilicated dispatch functions present no problem for H >tail-call elimination, provided that they don't need to do any cleanup H >after the called virtual function returns.  This is normally the case, D >with the only exceptions I know of being when Common Lisp's method $ >combination facility is being used.  ? I'm curious to see Tail Form conversion generally applied to OO A languages. It's been a few years I stopped reading papers in thatxE area, so I'm not familiar with the latest developments. But I look atPE my C++ programs, and I don't know, I somehow doubt that anybody coulda+ effectively convert them to tail form ! :-)-  -H >Yes, but fortunately the cases where it is a significant proportion of H >the total runtime are usually exactly those where it can be reduced or  >eliminated :-)h  BF I don't know, I don't know. There's still difficulties in the way. The/ debugging issue is, as I see it, a major one !       Alberto.   ------------------------------    Date: 28 Feb 2002 17:23:49 +00002 From: Maneki Neko <{spamtrap}@erewhon.demon.co.uk>& Subject: Re: expensive procedure calls0 Message-ID: <878z9dr04a.fsf@erewhon.demon.co.uk>  1 Alberto Moreira <junkmail@moreira.mv.com> writes:u [snip]C > Read, for example, Andrew Appel's "Compiling with Continuations".	 [snip]F > Also, the semantics of your programming language must be amenable toE > tail-form conversion. If the semantics of the language demands thatoH > you keep multiply nested stack frames with linkages to both staticallyH > and dynamically nested contexts, like Pascal for example, I'm not even+ > sure conversion to tail form is feasible._  A IIRC Appel's book has a reference to a continuation passing stylen$ Pascal compiler in its bibliography.   -- p% Battery chickens are not rechargable.i   ------------------------------   Date: 28 Feb 2002 14:00:05 GMT) From: leslie@clio.rice.edu (Jerry Leslie)d; Subject: HP Sees Charges Up To $1.4 bln After Compaq Mergeri' Message-ID: <a5ld55$2r3$2@joe.rice.edu>s Keywords: hp,merger,chargesi  >     http://www.siliconvalley.com/mld/siliconvalley/2756636.htm     Mercury News | 02/27/2002 |t6     HP sees charges up to $1.4 bln after Compaq merger  4 --Jerry Leslie     (my opinions are strictly my own)   ------------------------------  # Date: Thu, 28 Feb 2002 18:10:57 GMT-* From: "Bill Todd" <billtodd@metrocast.net>? Subject: Re: HP Sees Charges Up To $1.4 bln After Compaq Merger?C Message-ID: <Rouf8.231321$d34.17177038@bin8.nnrp.aus1.giganews.com>   6 "Jerry Leslie" <leslie@clio.rice.edu> wrote in message! news:a5ld55$2r3$2@joe.rice.edu... @ >     http://www.siliconvalley.com/mld/siliconvalley/2756636.htm! >     Mercury News | 02/27/2002 | 8 >     HP sees charges up to $1.4 bln after Compaq merger >-6 > --Jerry Leslie     (my opinions are strictly my own)  G Note also the rosy assumptions about Compaq's performance in the cominga= year.  I think I'll address that separately in a few minutes.g   - bill   ------------------------------    Date: 28 Feb 2002 04:21:54 -0800- From: adrian_ogden@hotmail.com (Scumbag Adie)n1 Subject: HSD30 licensing with mirroring and Cache ; Message-ID: <91445c94.0202280421.fba04d@posting.google.com>    Hi,h  E I thought one could do mirroring with a Cache module (no battery) andNA just the mirroring license. But in practice it seems to object ift? there is no battery present.  Is this true, as according to theCA options catalogue you only get a battery if you purchase the WBCA) option (license and battery)?B     Ta mucho   Adie   ------------------------------  # Date: Thu, 28 Feb 2002 11:33:39 GMTr From: system@SendSpamHere.ORGi( Subject: Re: I: PAKGEN Software for ISVs0 Message-ID: <00A0A3AA.49C63D1D@SendSpamHere.ORG>  p In article <3c7d927c.524144494f47414741@radiogaga.harz.de>, martin@radiogaga.harz.de (Martin Vorlaender) writes: >system@SendSpamHere.ORG wrote:>) >>  "Tom Linden" <tom@kednos.com> writes:s >> > once installed@ >> > $ lic gen >> >  >> > and then it is menu driven. >> >> Not on my machine...E > . >From skimming over the docs, I thought it was >  >  $ LICENCE REGISTER /GENERATEs   Yeap!    --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMI            wJ   "And of course, I'm a genius, so people are naturally drawn to my fiery I   intellect.  Their admiration overwhelms their envy!" -- Calvin & Hobbesl   ------------------------------   Date: 28 Feb 2002 06:57:29 CDT= From: wayne@tachysoft.xxx.120519.killspam.00c7 (Wayne Sewell)e( Subject: Re: I: PAKGEN Software for ISVs. Message-ID: <cVzC8K+49bsV@tachxxsoftxxconsult>  [ In article <3C7DA1D7.B9A63756@fsi.net>, "David J. Dachtera" <djesys.nospam@fsi.net> writes:/ > Wayne Sewell wrote:e >> l   [stuff deleted]    >> FP >> I agree.  Admittedly, it should have always been this way, but at least it isQ >> finally fixed.  Maybe we will actually start using LMF for our products, afterrP >> all these years, instead of our own key format.  I'm sure the system managers. >> at our customer sites will appreciate that. > J > I've worked in shops where non-LMF software (other license methods) were > verboten.- >   O Really?  What's the reasoning on that?  Too much trouble dealing with alternateRI key formats?  In our case, it's just a text file residing in a particulareM product directory.  The information is just as easy to enter as that of LMF. eJ In fact, there are far fewer fields to enter.   In most cases, the key canF simply be extracted from a mail message sent by SP32 customer support.  F Now if you are talking about tying the key to the hardware as has beenN discussed many times in the past, bound to the cpu id or the ethernet address O or whatever, I can understand the objection to that.  I think nearly all of therM vms vendors have learned not to do that, due to the public outcry.   Software I Partners never made that mistake, at least not in the ten years I've beenn involved with them.U  K Would these sites of yours still ban alternate key systems if they were not C hardware based, just a different key format and checksum algorithm?i  M I have not heard of sales lost simply because of the lack of LMF keys, thoughhN of course it could have happened.  If we had been losing a lot of business forN that reason, I assume management would have been forced to bite the bullet and8 pay the (formerly) huge price for the LMF key generator.     -- -O ===============================================================================CM Wayne Sewell, Tachyon Software Consulting  (281)812-0738  wayne@tachysoft.xxx9: http://www.tachysoft.xxx/www/tachyon.html and wayne.html  K change .xxx to .com in addresses above, assuming you are not a spambot  :-)gO ===============================================================================A= Society Lady:  Are you familiar with the Great Wall of China? 5        Curly:  No, but I know a big fence in Chicago!    ------------------------------    Date: 28 Feb 2002 12:20:01 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen)t( Subject: Re: I: PAKGEN Software for ISVs3 Message-ID: <hmfHFBO8GbpX@eisner.encompasserve.org>-  n In article <cVzC8K+49bsV@tachxxsoftxxconsult>, wayne@tachysoft.xxx.120519.killspam.00c7 (Wayne Sewell) writes:] > In article <3C7DA1D7.B9A63756@fsi.net>, "David J. Dachtera" <djesys.nospam@fsi.net> writes::  K >> I've worked in shops where non-LMF software (other license methods) weree >> verboten. >> r > Q > Really?  What's the reasoning on that?  Too much trouble dealing with alternate  > key formats?  , I think standardization is a popular reason.  A The LMF guarantees regarding readability of the license, etc. arecA controlled by Compaq.  As such, they are unlikely to change, evenh if only due to inertia.   > Answering questions regarding expiration dates, etc. for _all_A products is not much more work than answering it for one product.h   ------------------------------   Date: 28 Feb 2002 07:01:21 CDT= From: wayne@tachysoft.xxx.120519.killspam.00c7 (Wayne Sewell) F Subject: Re: IBM says they have best real time mainframe, where's VMS?. Message-ID: <1YuRha6gJGrM@tachxxsoftxxconsult>  [ In article <3C7DA01A.561E334E@fsi.net>, "David J. Dachtera" <djesys.nospam@fsi.net> writes:  > Bob Ceculski wrote:s >> fO >> IBM has nothing close to VMS!  The only way we will switch to IBM is if they  >> buy VMS!i > 4 > ....which I still think would be a 1st-class idea!  L Yeah, at least they have a clue as to what the enterprise is all about.  And* they do know how to market and sell stuff.     -- oO ===============================================================================aM Wayne Sewell, Tachyon Software Consulting  (281)812-0738  wayne@tachysoft.xxxn: http://www.tachysoft.xxx/www/tachyon.html and wayne.html  K change .xxx to .com in addresses above, assuming you are not a spambot  :-)sO =============================================================================== = Society Lady:  Are you familiar with the Great Wall of China?h5        Curly:  No, but I know a big fence in Chicago!o   ------------------------------  % Date: Thu, 28 Feb 2002 00:32:53 -0700o+ From: "Dennis O'Connor" <dmoc@primenet.com>a Subject: Re: Itanium troubles 5 Message-ID: <a5kmef$19gu$1@node21.cwnet.roc.gblx.net>   . "Bill Todd" <billtodd@metrocast.net> wrote ...N > In this case, AMD gained share during the first 9 months of 2001 (continuingM > a trend it established solidly in 2000), then lost some of that gain in thesJ > last quarter (but still ended up +20% in share percentage for the year).M > There's at least some specific reason (beyond the fact that the quarter wasdA > anomolous in the first place) - the Celeron dumping I mentioned   @ What "Celeron dumping" ?  Profit margins on Intel(tm) Celeron(R)D processors are quite nice, thank you, although obviously not as nice2 as the profit margins on Intel Xeon(R) processors.  C Intel is very good at making large dies and doing so inexpensively.m --3 Dennis O'Connor                   dmoc@primenet.com 3 We don't become a rabid dog to destroy a rabid dog.    ------------------------------  % Date: Thu, 28 Feb 2002 09:48:34 -0000o9 From: "Peter Dickerson" <peter.dickerson@uk.online.co.uk>b Subject: Re: Itanium troubles[C Message-ID: <W1nf8.19170$Hg1.2785157@news6-win.server.ntlworld.com>n  3 "Peter da Silva" <peter@abbnm.com> wrote in messager news:a5jfq1$al@web.nmti.com...- > In article <a5hc45$6mr$3@milo.mcs.anl.gov>,e; > Scandora, Anthony \(35048\) <Scandora@cmt.anl.gov> wrote:2I > > I was impressed with Hammer until I found out that there are no plans  for ac > > Windows 64 port. >eK > You know, I was seeing people claim that there were no plans for a 64-bits7 > IA32-base chip from Intel right up to the last month.n > F > And in any case, who cares whether there's a 64-bit Windows port forI > Hammer? It's not like there's any compelling reason to run Windows on a J > 64-bit chip: if there was Microsoft would have kept the Alpha port goingG > no matter what Compaq did... they are never reluctant to chase dollar  > signs. >e > --- >  `-_-'   In hoc signo hack, Peter da Silva.pG >   'U`    "A well-rounded geek should be able to geek about anything."a; >                                                        --a nicolai@esperi.org >          Disclaimer: WWFD?H There is one compelling reason for MS. Marketing - 64 is a larger number than 32.   Peterh   ------------------------------   Date: 28 Feb 2002 09:02 CSTv' From: carl@gerg.tamu.edu (Carl Perkins)e Subject: Re: Itanium troubles - Message-ID: <28FEB200209025085@gerg.tamu.edu>-  1 JF Mezei <jfmezei.spamnot@videotron.ca> writes...eN }If a 64 bit machine with a 64 bit 3d video rendering card makes new much more  B High end video cards are already using 128 bit processors (by someC ways of counting bits - I doubt that they have more than 32 bits ofnA addressing since even high end consumer graphics cards don't comebD with more than 128MB of memory on them - or maybe 256MB, which mightE be an option on the new Nvidea GeForce4 cards). They are really quiteh> impressive, but the video card's "bitness" is unrelated to the system's CPU's bit count.>   --- Carl   ------------------------------  % Date: Wed, 27 Feb 2002 10:04:14 -0600s, From: "Tony Scandora" <Scandora@cmt.anl.gov> Subject: Re: Itanium troubless+ Message-ID: <a5lh7v$mnq$1@milo.mcs.anl.gov>a  5 "Nick Maclaren" <nmm1@cus.cam.ac.uk> wrote in messagea* news:a5ia4a$n6k$1@pegasus.csx.cam.ac.uk...- > In article <a5hc45$6mr$3@milo.mcs.anl.gov>, ; > Scandora, Anthony \(35048\) <Scandora@cmt.anl.gov> wrote:c >...I > >For what?  They have to run Windows, which means Pentium 4 at the high  end ' > >until something better is available.  > G > As Bill Todd said, no, they don't.  None of SGI, HP, IBM and NEC were D > ever planning to on their Itanic ranges which had to be withdrawn.G > Only SGI actually managed to deliver ANYTHING, and it certainly won't E > make a profit out of its current systems - they are there as a stepe< > to the 'real' system (which assumes the McKinley arrives).  L HP/UX has a viable hardware platform until the next one is ready, as do VMS,I Tru64, and NSK, and AIX has a long term viable platform.  I can't imagine1K why a new customer would want a vendor-specific non-Windows system from SGI  or NEC.   D > >> That is so.  But most of them (not Compaq) have now resusciatedG > >> their own alternative strategies - certainly NEC, HP and SGI have,a< > >> and IBM never announced the replacement of theirs.  ... > >mI > >HP has merely extended the life of PA, as can be done with EV7, for asn long. > >as it takes ... until then, Intel is in theL > >enviable position of having its customers stay right where they are until, > >Intel gets their 64 bit products working. >iD > I suggest that you talk to your colleagues at Argonne who are into. > high-end computing.  Most of that is not so.    J I am quite impressed with the work my colleagues at Argonne have done withE high perfomance computing.  They have pushed the leading edge of high L performance computing for as long as there has been computing, and like most< other high end customers, they are not doing it with Wintel.  K Just as in its day the MicroVAX II created a huge new mass market unrelatedmL to Cray's, Wintel created and owns a huge mass market today.  Unfortunately,I the next few versions of Windows will still not be as secure or robust ascJ MicroVMS V4.n was, but that's another story.  When design and fab cost forI new chips is measured in gigabucks, volume is important.  Wintel is whereoH the money is today, for better or for worse.  Argonne and other high endK customers will demand and pay for the high end.  The rest of the world williK go with Wintel, and VMS, Tru64 and NSK might as well go along for the ride.m  1 Tony Scandora, Argonne National Lab, 630-252-7541i scandora@cmt.anl.gov   ------------------------------   Date: 28 Feb 2002 15:43:01 GMT( From: nmm1@cus.cam.ac.uk (Nick Maclaren) Subject: Re: Itanium troublese0 Message-ID: <a5lj65$90i$1@pegasus.csx.cam.ac.uk>  + In article <a5lh7v$mnq$1@milo.mcs.anl.gov>,i. "Tony Scandora" <Scandora@cmt.anl.gov> writes:8 |> "Nick Maclaren" <nmm1@cus.cam.ac.uk> wrote in message- |> news:a5ia4a$n6k$1@pegasus.csx.cam.ac.uk...u |> >...eL |> > >For what?  They have to run Windows, which means Pentium 4 at the high |> end* |> > >until something better is available. |> >J |> > As Bill Todd said, no, they don't.  None of SGI, HP, IBM and NEC wereG |> > ever planning to on their Itanic ranges which had to be withdrawn.rJ |> > Only SGI actually managed to deliver ANYTHING, and it certainly won'tH |> > make a profit out of its current systems - they are there as a step? |> > to the 'real' system (which assumes the McKinley arrives).i |> eO |> HP/UX has a viable hardware platform until the next one is ready, as do VMS,cL |> Tru64, and NSK, and AIX has a long term viable platform.  I can't imagineN |> why a new customer would want a vendor-specific non-Windows system from SGI
 |> or NEC.  E Doubtless.  But I can assure you that there are a lot of good reasonsc
 for doing so.,  B SGI's MIPS has at least as much viability as Alpha (and probably aB great deal more), and IRIX is a very cost-effective solution for a' large number of important requirements.   C Not to say that SGI MIPS/IRIX is the ONLY SMP system scalable up topE large systems and has good high-end graphics support, and NEC remainst? one of the leading HPC vendors with some distinctive strengths.r     Regards, Nick Maclaren,* University of Cambridge Computing Service,> New Museums Site, Pembroke Street, Cambridge CB2 3QH, England. Email:  nmm1@cam.ac.uk/ Tel.:  +44 1223 334761    Fax:  +44 1223 334679a   ------------------------------  + Date: Thu, 28 Feb 2002 15:48:27 +0000 (UTC)t From: david20@alpha2.mdx.ac.uk Subject: Re: Itanium troublesp+ Message-ID: <a5ljgb$88d$1@aquila.mdx.ac.uk>a  Z In article <a5lh7v$mnq$1@milo.mcs.anl.gov>, "Tony Scandora" <Scandora@cmt.anl.gov> writes:6 >"Nick Maclaren" <nmm1@cus.cam.ac.uk> wrote in message+ >news:a5ia4a$n6k$1@pegasus.csx.cam.ac.uk... . >> In article <a5hc45$6mr$3@milo.mcs.anl.gov>,< >> Scandora, Anthony \(35048\) <Scandora@cmt.anl.gov> wrote: >>...u >nK >I am quite impressed with the work my colleagues at Argonne have done with-F >high perfomance computing.  They have pushed the leading edge of highM >performance computing for as long as there has been computing, and like most = >other high end customers, they are not doing it with Wintel.i >oL >Just as in its day the MicroVAX II created a huge new mass market unrelatedM >to Cray's, Wintel created and owns a huge mass market today.  Unfortunately,aJ >the next few versions of Windows will still not be as secure or robust asK >MicroVMS V4.n was, but that's another story.  When design and fab cost forb; >new chips is measured in gigabucks, volume is important.  l  L IA32 and compatible has the Volume. IA64 is unlikely to have any comparable 2 volume until near the end of the decade -if then.       >Wintel is whereI >the money is today, for better or for worse.  Argonne and other high end L >customers will demand and pay for the high end.  The rest of the world willL >go with Wintel, and VMS, Tru64 and NSK might as well go along for the ride. >p  A What do you mean by VMS, TRU64 and NSK going along for the ride ?rK For Tru64 going along on the IA64 / HP-COMPAQ merger ride means extinction.mM For VMS going along on the IA64 ride means replacing a high performance nichem) chip with a lower performance niche chip.nM NSK is probably the only one of these OS's where going along on the IA64 ride0L might lead to comparable performance. But that won't be much comfort if as IN expect Intel were to stop developing it in favour of their plan B x86-64 chip.  A As to the rest of the world going with Wintel I wouldn't discount. win-AMD or Lin-AMD.s  
 David Webb VMS and Unix team leader CCSS Middlesex University l  2 >Tony Scandora, Argonne National Lab, 630-252-7541 >scandora@cmt.anl.gov  >P >e   ------------------------------  % Date: Thu, 28 Feb 2002 09:48:43 -0600o, From: "Tony Scandora" <Scandora@cmt.anl.gov> Subject: Re: Itanium troubles + Message-ID: <a5ljpt$n2d$1@milo.mcs.anl.gov>l  5 "Bill Todd" <billtodd@metrocast.net> wrote in message = news:rn9f8.232019$Re2.17386184@bin6.nnrp.aus1.giganews.com...8G > > "Scandora, Anthony (35048)" <Scandora@cmt.anl.gov> wrote in messagei) > > news:a5hc45$6mr$3@milo.mcs.anl.gov...9 > .... > > > I wasaK > > > impressed with Hammer until I found out that there are no plans for aa > > > Windows 64 port. > > J > > Really?  Do you have a reference we can check to confirm this?  Last I > knew,rH > > XP code contained some conditionals for x86-64, though Microsoft has been< > > coy about making any public commitments to the platform. >BL > Just stumbled upon the following Tom's Hardware Guide article (following a > reference in comp.arch)y >m8 > http://www.tomshardware.com/cpu/02q1/020227/index.html >  > which states >AI > "Tom's Hardware Guide received the opportunity to take a closer look ati two.F > working test systems - each of which was equipped with the AMD "ClawE > Hammer". Suse Linux was installed on one of the PCs, and Windows XP> (64-bitn' > version) was installed on the other."t > K > Now, they may of course not have a clue what they're talking about, but I K > have to suspect that having visited AMD while researching the material in A > question they may know more about the situation than Tony does.E  K That bug must have been fixed.  It now says, "Tom's Hardware Guide receivedaK the opportunity to take a closer look at two working test systems - each of"F which was equipped with the AMD "Claw Hammer". Suse Linux (64-bit) wasJ installed on one of the PCs, and Windows XP (32-bit version) was installed on the other."  K Also on http://www.tomshardware.com/cpu/02q1/020227/hammer-03.html it says, K "AMD did not want to give us further details about the exact clock speed offH the CPU, nor did they give us details on the performance under Linux and Windows XP (32-bit version)."h  L There could very well be X86-64 conditionals in the code, as there are AlphaK and maybe still MIPS, PowerPC, and who knows what others.  AMD's own FAQ atnL http://www.amd.com/us-en/Processors/DevelopWithAMD/0,,30_2252_875_1024,00.htJ ml says, "Future processors derived from the 'hammer' architecture will beK able to use existing and future 32-bit operating systems such as Windows98,sJ Windows ME, Windows NT, Windows 2000, WindowsXP & Linux. Additionally, theK 'Hammer' products are planned to enjoy 64-bit operating system support from  the open source community."s  1 Tony Scandora, Argonne National Lab, 630-252-7541f scandora@cmt.anl.gov   ------------------------------   Date: 28 Feb 2002 16:32:40 GMT( From: nmm1@cus.cam.ac.uk (Nick Maclaren) Subject: Re: Itanium troublese0 Message-ID: <a5lm38$c4g$1@pegasus.csx.cam.ac.uk>  ' In article <3C7D9EE1.D90AFCE2@igs.net>,e% Paul DeMone <pdemone@igs.net> writes:u |> Bill Todd wrote:o |> tM |> > As a result, AMD's average selling price for its chips bounded from justtB |> > above $70 in the third quarter to $90 in the fourth quarter." |> aE |> Too bad AMD needs even higher ASPs to break even. Do you think AMD D |> can continue to raise its ASPs in 2002 in face of Intel's ramp inF |> Northwood shipments? Before you answer you might want to review theE |> roadmap AMD released today. K7 performance bumps have been furtherg |> scaled back and/or delayed.  @ Hence their interest in the Hammer range.  Yes, the stake is theD survival of the company, but they are aware of that and are tackling@ the right problem.  If they succeed completely, it is Intel that? will be in trouble; if they fail completely, they are finished.=$ My guess is something in between :-)     Regards, Nick Maclaren,* University of Cambridge Computing Service,> New Museums Site, Pembroke Street, Cambridge CB2 3QH, England. Email:  nmm1@cam.ac.uk/ Tel.:  +44 1223 334761    Fax:  +44 1223 334679=   ------------------------------  # Date: Thu, 28 Feb 2002 17:41:44 GMTy  From: cjt <cheljuba@prodigy.net> Subject: Re: Itanium troubles + Message-ID: <3C7E6BFE.9E46879E@prodigy.net>-   Peter Dickerson wrote: > 5 > "Peter da Silva" <peter@abbnm.com> wrote in messagem  > news:a5jfq1$al@web.nmti.com.../ > > In article <a5hc45$6mr$3@milo.mcs.anl.gov>,h= > > Scandora, Anthony \(35048\) <Scandora@cmt.anl.gov> wrote:bK > > > I was impressed with Hammer until I found out that there are no plansl > for a: > > > Windows 64 port. > >3M > > You know, I was seeing people claim that there were no plans for a 64-bito9 > > IA32-base chip from Intel right up to the last month.) > >eH > > And in any case, who cares whether there's a 64-bit Windows port forK > > Hammer? It's not like there's any compelling reason to run Windows on a L > > 64-bit chip: if there was Microsoft would have kept the Alpha port goingI > > no matter what Compaq did... they are never reluctant to chase dollare
 > > signs. > >  > > --/ > >  `-_-'   In hoc signo hack, Peter da Silva.dI > >   'U`    "A well-rounded geek should be able to geek about anything."o= > >                                                        --2 > nicolai@esperi.org > >          Disclaimer: WWFD?J > There is one compelling reason for MS. Marketing - 64 is a larger number
 > than 32. >  > Petere  0 Maybe they should port it to the PlayStation II.   ------------------------------  # Date: Thu, 28 Feb 2002 17:44:04 GMTi* From: "Bill Todd" <billtodd@metrocast.net> Subject: Re: Itanium troublesMC Message-ID: <E%tf8.214019$Aw2.16733077@bin7.nnrp.aus1.giganews.com>   7 "Tony Scandora" <Scandora@cmt.anl.gov> wrote in messageh% news:a5ljpt$n2d$1@milo.mcs.anl.gov...t   ...r  H > There could very well be X86-64 conditionals in the code, as there are Alpha=; > and maybe still MIPS, PowerPC, and who knows what others.=  J Er, Windows *ran* (and was sold) on all those other platforms you mention,H so if you're drawing a parallel it's one that doesn't exactly favor your
 viewpoint.     AMD's own FAQ at >aL http://www.amd.com/us-en/Processors/DevelopWithAMD/0,,30_2252_875_1024,00.htL > ml says, "Future processors derived from the 'hammer' architecture will beB > able to use existing and future 32-bit operating systems such as
 Windows98,: > Windows ME, Windows NT, Windows 2000, WindowsXP & Linux.  F Indeed.  However, the conditionals reportedly appeared to apply to the1 64-bit Windows code, not the 32-bit Windows code.9   - bill   ------------------------------  # Date: Thu, 28 Feb 2002 17:50:54 GMT * From: "Bill Todd" <billtodd@metrocast.net> Subject: Re: Itanium troublesiC Message-ID: <26uf8.238513$Re2.17986253@bin6.nnrp.aus1.giganews.com>   6 "Dennis O'Connor" <dmoc@primenet.com> wrote in message/ news:a5kmef$19gu$1@node21.cwnet.roc.gblx.net...t0 > "Bill Todd" <billtodd@metrocast.net> wrote ...D > > In this case, AMD gained share during the first 9 months of 2001 (continuinghK > > a trend it established solidly in 2000), then lost some of that gain ine theeL > > last quarter (but still ended up +20% in share percentage for the year).K > > There's at least some specific reason (beyond the fact that the quartero waspC > > anomolous in the first place) - the Celeron dumping I mentioned  >d > What "Celeron dumping" ?  D The major Asian Celeron sales in Q4 that held Intel's ASP flat in anJ expanding market, but did succeed in recovering some market share from AMDF in that quarter (and temporarily reverse the trend established in most  quarters of the past two years).  H I could go back to Google to find the references, but I'm sure you're as3 capable of doing that as I am if you're interested..   - bill   ------------------------------  % Date: Thu, 28 Feb 2002 13:48:27 -0500m- From: JF Mezei <jfmezei.spamnot@videotron.ca>e Subject: Re: Itanium troublesd, Message-ID: <3C7E7B7A.E3B39469@videotron.ca>   Tony Scandora wrote:N > HP/UX has a viable hardware platform until the next one is ready, as do VMS,K > Tru64, and NSK, and AIX has a long term viable platform.  I can't imaginerM > why a new customer would want a vendor-specific non-Windows system from SGI 	 > or NEC.:  M Why would someone buy SGI ? becayuse the software they need only runs on SGI.mM That is why. If the few software makers that make SGI-only software decide to_J switch to a different platform (lets say IBM-POWER), then SGI will be dead quite soon.t  K Why would software developpers switch platform from SGI to something else ? F Because they are not certain that SGI will be able to keep up and if aM competitor arises on a different platform that is better performing at better 0 price, then that software company is also toast.  J Uncertainty about a platform (hardware and OS) are deadly. SGI isn't awareN that one of its major software developpers is contemplating a platform switch,K and it won't know for quite some time until that company publicly announces G its product has been ported to platform X. By that time it is too late.i  0 I leave you the task of translating this to VMS.   ------------------------------   Date: 28 Feb 2002 13:57:54 GMT) From: leslie@clio.rice.edu (Jerry Leslie)c0 Subject: Microsoft Curries Favor With Undergrads' Message-ID: <a5ld12$2r3$1@joe.rice.edu> & Keywords: microsoft,undergrads,loyalty  ? $ 99 for each student and only $799/year for each university...a  ,    http://news.com.com/2100-1001-847298.htmlA    Microsoft curries favor with undergrads - Tech News - CNET.com4  *    Microsoft curries favor with undergrads    By Wylie Wong    Staff Writer, CNET News.com     February 28, 2002, 4:45 AM PT  F   "Microsoft hopes to win over software developers while they're still	    young.a  E    Microsoft on Thursday released software development tools aimed atcD    college-level computer science students, hoping it will produce aC    fresh crop of software programmers loyal to its Windows and .Net     technology.  H    The software maker has shipped a version of its new Visual Studio.NetI    tools with specific features for educational use. Visual Studio.Net is H    the latest version of Microsoft's development tools that allow peopleG    to write Web-based software and services. As part of its overarchingtH    .Net software strategy, the company earlier this month released three7    versions of the tools aimed at the corporate market.l  B    Michael Bronsdon, Microsoft's lead product manager for academicG    developer marketing, said win over software developers while they're F    still young, which competes against Sun Microsystems, IBM and otherH    Java supporters for talent, is offering its tools at a heavy discountF    to students and schools in hopes of enticing more people to use its    technology.  H    "We are doing an awful lot to get the next wave of developers workingG    with Web services and understanding and working with .Net, and we'reI3    focused quite a bit on academia," Bronsdon said.s  G    Visual Studio.Net Academic includes Web-based software that provides A    an online area where faculty and students can communicate. FornG    example, professors could put class assignments for students online.qH    The class, in turn, could download the assignments, complete them and$    upload them to Web-based servers.  G    The product also includes several wizards, which makes it easier forsH    programmers to write software by guiding them through the development    process.n  B    The cost of the Visual Studio.Net Academic for students is $99,G    Bronsdon said. Universities can gain access to the Visual Studio.NetaD    tools and install it in their computer labs through a $799 yearly/    subscription with Microsoft's MSDN Web site.t  E    Visual Studio.Net includes updates to programming languages Visual A    Basic, Visual C++, as well as the first version of Microsoft'st"    Java-like language called C#. "    4 --Jerry Leslie     (my opinions are strictly my own)   ------------------------------  % Date: Thu, 28 Feb 2002 03:18:41 -0500a( From: David Froble <davef@tsoft-inc.com>6 Subject: Re: Need help on Alpha:  Calling C from MACRO, Message-ID: <3C7DE7E1.1070207@tsoft-inc.com>   David Froble wrote:F   > Stanley F. Quayle wrote: > J >> I'm having a problem calling Compaq C from MACRO on Alpha.  This works H >> perfectly on VAX, calling VAX C from MACRO (I didn't try Compaq C on  >> VAX--yet!). >>" >> The MACRO code looks like this: >> >>         SUBL2   #20, SP) >>         MOVAW   TOTALON_N1(R11), 0(SP)n& >>         MOVAB   BIT1_I1(R11), 4(SP)& >>         MOVAB   BIT2_I1(R11), 8(SP)' >>         MOVAB   BIT3_I1(R11), 12(SP)> >>         MOVL    R11, 16(SP)" >>         CALLS   #5, G^BITON_C16 >  > I > I've not got an answer to your question, but, why are you messing with tB > SP instead of using the PUSHAW, PUSHAB, PUSHAL etc instructions? > K > Actually, I think this may be your problem.  On Alpha, I don't think you oK > have access to SP, at least the 'real' SP.  Remember, it's compiled, not e > assembled. >  > Dave >   . It's late, and I wasn't thinking real clearly.  Q This is the other side of the major change I had to make in MACRO-32 code to get nK it to work on Alpha.  A technique that worked well on VAX was to fetch the vH argument, and increment the AP, to be ready to fetch the next argument. Q Modifying the AP doesn't work on Alpha, you needed to include the offset of each S argument.  For example, on VAX:s  >          TSTL    (AP)+                   ; Skip argument count9          MOVL    @(AP)+,R2               ; Channel numbermI          MOVL    (AP)+,R3                ; Address of desc. of file spec.t.          MOVL    @(AP)+,R4               ; MBC4          MOVL    @(AP)+,R5               ; Open mode  5 Notice the increment of the AP.  Won't work on Alpha.k  9          MOVL    @4(AP),R2               ; Channel numberuI          MOVL    8(AP),R3                ; Address of desc. of file spec.a.          MOVL    @12(AP),R4              ; MBC4          MOVL    @16(AP),R5              ; Open mode  9 The second example is the mods required to work on Alpha.a  O The key thing to remember on Alpha is that you don't really have access to the  N machine, or at least some things like AP, SP, and such.  I can't remember the P details, it's about 3:15 AM and I'm not sure what I'm doing awake and answering  your post.  :-)x  O Note that the second method would also work on VAX, and might be considered by y, some to be a more 'proper' coding technique.  P Also remember that when you're PUSHing arguments onto the stack, start with the 1 last argument and work toward the first argument.    Dave   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com6 T-Soft, Inc.  170 Grimplin Road  Vanderbilt, PA  15486   ------------------------------  # Date: Thu, 28 Feb 2002 11:57:18 GMTu From: system@SendSpamHere.ORGs6 Subject: Re: Need help on Alpha:  Calling C from MACRO0 Message-ID: <00A0A3AD.975DD589@SendSpamHere.ORG>  [ In article <3C7D6967.23080.AE8A342@localhost>, "Stanley F. Quayle" <stan@stanq.com> writes:aB >I'm having a problem calling Compaq C from MACRO on Alpha.  This G >works perfectly on VAX, calling VAX C from MACRO (I didn't try Compaq w >C on VAX--yet!).  >a  >The MACRO code looks like this: >a >        SUBL2   #20, SP' >        MOVAW   TOTALON_N1(R11), 0(SP).$ >        MOVAB   BIT1_I1(R11), 4(SP)$ >        MOVAB   BIT2_I1(R11), 8(SP)% >        MOVAB   BIT3_I1(R11), 12(SP)h >        MOVL    R11, 16(SP)  >        CALLS   #5, G^BITON_C16 >o >The C code looks like this: >  >#include varargsd >BITON_C16(va_alist) >va_dclo >{ >    va_list list_incrementor; >i >[...] >e0 >    va_count(num_args);			/* this works okay */ >  >	/* these work okay, too */  >    va_start(list_incrementor);2 >    arg1_n_ptr = va_arg(list_incrementor, char*);2 >    arg2_n_ptr = va_arg(list_incrementor, char*); >    va_end(list_incrementor); >i8 >    for ( index = 0; index <= (num_args - 4); ++index ) >    {1 >		va_start_1(list_incrementor, (index + 2) * 4);M5 >	    arg3_n_i_ptr = va_arg(list_incrementor, char*); " >        va_end(list_incrementor); >	[...] etcr >n >aD >On the first pass through the loop, I get the address of the third $ >argument.  That's just what I need. > @ >BUT, on the second pass through the loop, I get zero!  If I go B >through the loop again, I get the address of the fourth argument. >lD >I've compiled the whole mess with /DEBUG/NOOPT, and the MACRO code D >writes 5 consecutive longwords of non-zero data.  When I step into F >the function, there are longwords of zero between each of the values  >that the MACRO code wrote.v  " LONGWORD => 32 bits... VAX 32 bits' QUADWORD => 64 bits... Alpha 64 bits...t    @ >I've tried various compiler options.  The behavior is constant F >between VMS 7.2 and VMS 7.3.  The behavior is the same between DEC C  >V5.5-002 and V6.0-001.e >n >Anyone have any ideas??  F Arguments on Alpha are not passed on the stack -- at least the first 6G are not -- they are passed via registers 16 through 21.  (Go on and tryr- Ben, it's public knowledge not a top secret).     G Anyway, it looks like your C code is walking through by longword steps.t  1 >		va_start_1(list_incrementor, (index + 2) * 4);i=                                              --------------^ c! This would need to be 8 on Alpha.D --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMi            )J   "And of course, I'm a genius, so people are naturally drawn to my fiery I   intellect.  Their admiration overwhelms their envy!" -- Calvin & Hobbes    ------------------------------  % Date: Thu, 28 Feb 2002 10:29:41 -0500 2 From: "Sue Skonetski" <susan.skonetski@compaq.com>J Subject: OpenVMS get together in Houston - thought you might be interested1 Message-ID: <X%rf8.331$fL6.6212@news.cpqcorp.net>   ( Compaq, PointSecure and Pioneer Standard- invite you to join us for a reintroduction to  the Houston OpenVMS community.   When: Wednesday March 13th From 5:30 PM - 7:30 PM  + Where: Benjy's Restaurant - in Rice Village2 (713) 522-7602 2424 Dunstan Rd.  : Why: Informal event to meet other local VMS users, discuss5 VMS and Compaq's plans, hear a presentation from Johnp< Wisniewski and, of course, enjoy free drinks and appetizers.   Agenda: 5:30 PM - 6: 15 PM Arrival and greetingso   6: 15 PM - 6: 45 PM ! Presentation from Leland Bolleterw and John Wisniewski of Compaqn -Discuss plans for OpenVMS -Security and OpenVMSo   6: 45 PM - 7:30 PM Mingle with community membersp Wrap up    Please RSVP by March 8th  < To RSVP please send an email to mike.umansky@pointsecure.comK <mailto:%20mike.umansky@pointsecure.com> and write in the subject line willw attend or will not attend.  / We look forward to seeing you all on March 13thp   ------------------------------  % Date: Thu, 28 Feb 2002 11:50:02 -0500,2 From: "Sue Skonetski" <susan.skonetski@compaq.com>7 Subject: Oracle Rdb Technical Update in NY - Invitation21 Message-ID: <hbtf8.342$fL6.6201@news.cpqcorp.net>c   Dear Newsgroup,:  D If you are interested in attending this please feel free to sign up.  
 Warm Regards,E   sueF       Oracle Rdb Technical Updatee  H The intended audience is customers, resellers, and employees. The topicsI will be technical, although the first presentation is a marketing/roadmaphI talk.We will meet at the Compaq offices at 2 Penn Plaza ( 7th Ave at 31st ( St), 8th Floor, New York/ Pacific Rooms.  L Space is limited, please register as soon as possible. We will offer a light lunch.   March 15, 2002   Agenda  5 11:00-11:10 Welcome and introductions, Nancy Davidsona  : 11:10 - 11:45 Rdb Business and Product Update, Kevin Duffy  , 11:45 - 12:15 Rdb Port to Itanium, Ian Smith  . 12:15 - 12:45 Oracle 9iAS and Rdb, Kevin Duffy  / 12:45 - 1:15 Java and XML with Rdb, Kevin Duffyg  2 1:15 - 1:45 Oracle Rdb Log Miner, Norman Lastovica  : 1:45 - 2:15 Oracle Rdb on OpenVMS Galaxy, Norman Lastovica  8 2:15 - 2:45 Oracle Rdb and Bit Mapped Indices, Ian Smith  2 2:45 - 3:30 SQl Highlights from Rdb 7.1, Ian Smith  * 3:30- 3:40 wrap up and close, Lynne Hummel      C If you would like to join us on March 15th, please send an email to3I Andrea.Sullivan@compaq.com. It is necessary to register to gain access toh
 the building.1   Bio'sM   Nancy Davidson  " Oracle Alliance Manager, NE Region   Compaq Computer Corporation6   Kevin Duffyo   Director, Rdb Development    Oracle Corporation  F Kevin Duffy is the Director of Development for Oracle Rdb Engineering,I located in Nashua, NH. He has been a manager with Rdb since 1989. All Rdbn development reports to Kevin.o  	 Ian Smith    Consulting Engineerh   Oracle Corporation      A Ian Smith is a Consulting Member of the Technical Staff of Oracle I Corporation and a Technical Leader for the Oracle Rdb product engineering K group located in Nashua NH, U.S.A. Ian has been a member of the engineeringaK team since Jan 1989 and has been working with Rdb since the beta testing ofsC the first version in 1984. Between 1982 and 1989 Ian was a databasebJ consultant working with customers to build and tune Rdb and DBMS (CODASYL)C systems. Ian is currently the SQL language and relational interface B architect for Oracle Rdb and has designed many of the new features) incorporated into Rdb in the last decade.h   Norman J. Lastovicao   Principal Engineer   Oracle Corporation  I Norman J. Lastovica is a Principal Engineer within Oracle Rdb EngineeringiL located in Nashua, New Hampshire. Mr. Lastovica is currently a member of theI KODA team and shares responsibility for the physical data storage, index,SL journaling, recovery, row cache, hot standby, and LogMiner components of the Rdb engine.    Lynne Hummel   Compaq  ! High Performance Sales Specialistb   NY Metro   973-586-0807   ------------------------------    Date: 28 Feb 2002 05:56:22 -0800( From: bob@instantwhip.com (Bob Ceculski)U Subject: Re: Prior Version Support (was: Re: Compaq adopts microsoft approach to VMS)o= Message-ID: <d7791aa1.0202280556.7fe6baf8@posting.google.com>   k hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) wrote in message news:<LAhf8.288$fL6.5957@news.cpqcorp.net>...o > Retitled.' > t > In article <fe52053.0202261622.4183b2e7@posting.google.com>, antony.wardle@optusnet.com.au (Antony Wardle) writes:G > :Anyone else concerned that unless you are running the latest versionuE > :of VMS then you are now on Prior version/best endeavor support for 	 > :fixes?g > I >   You have various choices here: stay current and use standard support, F >   lag behind on specific OpenVMS releases and acquire and use Prior B >   Version Support (PVS) contracts, stay on the PVS releases and I >   download the ECOs and use per-call as needed, or fall far behind and  I >   drop entirely off PVS and access to updated ECOs entirely.  Each has  J >   different costs, different risks, and different trade-offs, of course. > 7 > :What does this say about Compaqs committment to VMS/c> > :Lets force everyone to upgrade, doesn't that sound like MS? > G >   I will assume that this is a serious inquiry, but will mention thatME >   the phrasing approaches that of a troll -- and trolling might notr >   have the desired effect. > D >   This has been the policy since long before I joined the OpenVMS F >   Engineering group, and that is at least a year or two past by now.E >   A policy clearly not communicated particularly well in this case.l > F >   For related information on this topic, please search for "PVS" in G >   the OpenVMS FAQ.  (The relevent section is here included below the SE >   sig file.)  I've tried to keep the PVS stuff easily accessable ine >   the FAQ for a while now. > H >   The effort of supporting and testing an increasing number of OpenVMSI >   releases is huge, and is increasing.  OpenVMS maintains a unique set  G >   of compilers and the associated development environments and tools  F >   for each OpenVMS release with Prior Version Support (PVS) status, I >   and we currently have roughly a dozen source code pools in operation iI >   entirely in parallel at the present time -- changes from older pools  H >   need to propogate into the newer code pools, lest a code regression H >   occur.  And there have been more than a few changes over the years, F >   specifically meaning changes to older modules can involve merging D >   the change into each subsequent PVS release or currently-active B >   development release.  All of this reduces the cycles that are F >   available to work on new engineering designs and new code, and on F >   the new features and capabilities and platforms that the customer  >   folks (you) want to buy. > F >   Additionally, as we find and fix problems in older releases, folksC >   on current releases see these fixes incorporated as the code iseG >   propogated forward and merged.  (And I can discuss the source code eJ >   control topic at great length, too, if you desire detailed drivel. :-) >  > B > :We've only recently found out about it, and is seems that priorB > :version will be for any version of VMS more than 18 months old.H > :How many people upgrade the minute a new version is out? The hobbiestG > :probably, but what about the business customers who are running 24*7 B > :on critical systems. The upgrade planning can take a few months > :not to mention the testing. > E >   This choice is one of the reasons why we provide "landing zones" vD >   and Prior Version Support (PVS) contracts.  If you need or want A >   to stay on an older OpenVMS release and you want to maintain 9A >   support, you can acquire support for the specified subset of 8C >   OpenVMS releases.   These PVS releases as the ones that get theCF >   ECOs, and the ones where problem reports can be fielded.  We also H >   provide a map of releases that will continue to have support; these F >   are the so-called "landing zone" releases mentioned.  See the PVS " >   URLs listed below for details. > B >   A rule of thumb for PVS status -- the releases that are likelyA >   (but not entirely certain, of course) to be designated as PVSrC >   "landing zones" -- is the last release of a major release (eg: eE >   V5.5-2 and V5.5-2H4, V6.2), and the last release of the previous t@ >   "dot" release (eg: V7.2-2).   Again, see the PVS URLs below. > D >   Releases in PVS status have a substantial notification interval;G >   a substantial lead-time before the release(s) will exit PVS status.hH >   Typically two years lead-time, per the PVS website.  Details on the G >   notification intervals, PVS status, etc, are at the PVS URLs below.a >  > P >  ---------------------------- #include <rtfaq.h> -----------------------------P >       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    P >  --------------------------- pure personal opinion ---------------------------N >    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com >   J what about a decwindows cert patch for 7.1-1H2 ... they want us to upgradeL to 7.1-2 to get it, but that involves upgrading 35 boxes around the country! we have bronze support ...   ------------------------------  % Date: Thu, 28 Feb 2002 15:38:41 +0100a$ From: "Dr. Dweeb" <Dweeb@NoSpam.com>U Subject: Re: Prior Version Support (was: Re: Compaq adopts microsoft approach to VMS)p0 Message-ID: <fhrf8.111$nv4.2385@news.get2net.dk>  # Thanks Hoff !  Filled in some gaps.c   Dweeb.? "Hoff Hoffman" <hoffman@xdelta.zko.dec.nospam> wrote in messagee+ news:LAhf8.288$fL6.5957@news.cpqcorp.net...v >l
 >   Retitled.  >e> > In article <fe52053.0202261622.4183b2e7@posting.google.com>,5 antony.wardle@optusnet.com.au (Antony Wardle) writes: G > :Anyone else concerned that unless you are running the latest version E > :of VMS then you are now on Prior version/best endeavor support fors	 > :fixes?t >rI >   You have various choices here: stay current and use standard support, E >   lag behind on specific OpenVMS releases and acquire and use PrioruA >   Version Support (PVS) contracts, stay on the PVS releases and H >   download the ECOs and use per-call as needed, or fall far behind andH >   drop entirely off PVS and access to updated ECOs entirely.  Each hasJ >   different costs, different risks, and different trade-offs, of course. >u7 > :What does this say about Compaqs committment to VMS/s> > :Lets force everyone to upgrade, doesn't that sound like MS? > G >   I will assume that this is a serious inquiry, but will mention thatfE >   the phrasing approaches that of a troll -- and trolling might noto >   have the desired effect. > C >   This has been the policy since long before I joined the OpenVMS F >   Engineering group, and that is at least a year or two past by now.E >   A policy clearly not communicated particularly well in this case.  > E >   For related information on this topic, please search for "PVS" in>F >   the OpenVMS FAQ.  (The relevent section is here included below theE >   sig file.)  I've tried to keep the PVS stuff easily accessable in  >   the FAQ for a while now. >cH >   The effort of supporting and testing an increasing number of OpenVMSH >   releases is huge, and is increasing.  OpenVMS maintains a unique setF >   of compilers and the associated development environments and toolsE >   for each OpenVMS release with Prior Version Support (PVS) status,eH >   and we currently have roughly a dozen source code pools in operationH >   entirely in parallel at the present time -- changes from older poolsG >   need to propogate into the newer code pools, lest a code regressionrG >   occur.  And there have been more than a few changes over the years,aE >   specifically meaning changes to older modules can involve mergingeC >   the change into each subsequent PVS release or currently-active A >   development release.  All of this reduces the cycles that are E >   available to work on new engineering designs and new code, and on E >   the new features and capabilities and platforms that the customer  >   folks (you) want to buy. > F >   Additionally, as we find and fix problems in older releases, folksC >   on current releases see these fixes incorporated as the code is F >   propogated forward and merged.  (And I can discuss the source codeJ >   control topic at great length, too, if you desire detailed drivel. :-) >  > B > :We've only recently found out about it, and is seems that priorB > :version will be for any version of VMS more than 18 months old.H > :How many people upgrade the minute a new version is out? The hobbiestG > :probably, but what about the business customers who are running 24*7tB > :on critical systems. The upgrade planning can take a few months > :not to mention the testing. > D >   This choice is one of the reasons why we provide "landing zones"C >   and Prior Version Support (PVS) contracts.  If you need or wanta@ >   to stay on an older OpenVMS release and you want to maintain@ >   support, you can acquire support for the specified subset ofC >   OpenVMS releases.   These PVS releases as the ones that get the E >   ECOs, and the ones where problem reports can be fielded.  We also G >   provide a map of releases that will continue to have support; theseuE >   are the so-called "landing zone" releases mentioned.  See the PVSA" >   URLs listed below for details. >tB >   A rule of thumb for PVS status -- the releases that are likelyA >   (but not entirely certain, of course) to be designated as PVSRB >   "landing zones" -- is the last release of a major release (eg:D >   V5.5-2 and V5.5-2H4, V6.2), and the last release of the previous@ >   "dot" release (eg: V7.2-2).   Again, see the PVS URLs below. >ND >   Releases in PVS status have a substantial notification interval;G >   a substantial lead-time before the release(s) will exit PVS status.VG >   Typically two years lead-time, per the PVS website.  Details on the G >   notification intervals, PVS status, etc, are at the PVS URLs below.  >v >i( >  ---------------------------- #include' <rtfaq.h> ----------------------------- L >       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com, >  --------------------------- pure personal# opinion ---------------------------n1 >    Hoff (Stephen) Hoffman   OpenVMS Engineering  hoffman#xdelta.zko.dec.com >u >r >- > -- >d9 >    OpenVMS and Layered Products -- Support Information:  >M@ >      For information on Prior Version Support (PVS) and MatureB >      Product Support (including information on support end dates= >      for OpenVMS and various layered products), please see:  >t? >        http://www.compaq.com/services/software/ss_mature.htmloD >        http://www.compaq.com/services/software/ss_pvs_se_amap.htmlD >        http://www.compaq.com/services/software/ss_mps_pvs_eur.html >PE >      For information on supported versions of layered products, anda6 >      minimum required layered product versions, see: > C >        http://www.openvms.compaq.com/openvms/os/swroll/index.htmlG >D   ------------------------------    Date: 28 Feb 2002 07:14:29 -0800- From: pdafniotis@yahoo.com (Petros Dafniotis)bK Subject: Q: Why is XP-1000 RAM so expensive from Compaq vis-a-vis Kingston? = Message-ID: <e54adf36.0202280714.46145e1c@posting.google.com>i  B In another post I asked for XP-1000 ram. I need 4 modules of 256MB each.t   This is listed as:7 Compaq reference numbers are:	SN-MSP01-KD or 388262-B21e
 This is from:kH http://www.compaq.com/alphaserver/options/xp1000/xp1000_sn-msp01-kd.html  E Compaq Switzerland quoted this as 10,000 CHF; Compaq USA as 6,300 USDx so almost the same.   D Now I go to Kingston which I thought was rather reputable and I see:R http://www.ec.kingston.com/ecom/kepler/PartsInfo_Bod.asp?ktcpartno=KTC-XP1000/1024 which retails for US$ 729!!!  D This is a HUGE difference! Can I please get an explanation? I do notD mind at all paying twice or three times as much iff the RAM is going@ to be more reliable but the price factor of ~9 is just enormous.  
 Kind regards,  Petros ---  Petros Dafniotis, PhDs pdafniotis@yahoo.com   ------------------------------    Date: 28 Feb 2002 10:58:40 -0800) From: cpw531@hotmail.com (Clark Williams)cS Subject: Question on a Deskpro EP`s motherbaord and possibily of changing it out :) = Message-ID: <edb829c0.0202281058.69b32e4c@posting.google.com>n  : I have a Compaq Deskpro EP BX/P500 , System Serial Number:D 6941CJN4L090. I have basically explored very nook and cranny of thisF system, and I have a question about the motherboard. I have opened the? case and noticed the motherboard looks like a standard ATX formeF factor, so the question is, for flexibility in the future, does anyoneC know if I can take out this older compaq motherboard (Compaq 041Ch)eD and replace with a more modern standard atx motherboard??? I have toD admit I like the Deskpro chassis and I enjoy tinkering, but, being aF novice and somewhat inexperienced with ATX ( I have always worked withF AT ),I was wondering if the actual motherboard mounting of the chassisF are standard or proprietary. THe fact is that I don`t have another ATXB motherbaord to visually compare the mountings, and second I really? don`t want to take apart this system just for the prime fact ofeD exploration  ( its my home business PC ). I thank anyone who has any
 info on this.-<                                                             
 cordially,  E                                                                 Clark  Williams    D Just as a sidenote, my other system is an older custom system with aC Tyan Trinity 100 AT (S1590) motherboard and other components, and ibB can`t keep its case on because I am always experimenting with it (6 meaning also that it likes to get unstable alot more).   ------------------------------  + Date: Thu, 28 Feb 2002 09:47:43 +0100 (MET)d9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>t$ Subject: restricting access to ports; Message-ID: <01KESX25C5OI8ZL57Q@sysdev.deutsche-boerse.com>1  F On my hobbyist system, I want to avoid unwanted incoming connections, C since I have to pay for them.  My ISP can set up a filter to allow KI incoming connections to only certain ports.  However, this has the nasty 4F side effect that I can't establish an outgoing connection, since some 4 port is "randomly" allocated for the return traffic.  C Is there any way to tell VMS to allow incoming connections only to  G certain ports?  This would allow the ISP to block incoming requests to fI other ports, saving me the corresponding connection costs.  As long as I SE have enough free ports to handle return traffic, things should be OK.   H Another approach might be to restrict access to certain ports which are I common targets of unwanted incoming connections.  For example, there are nI Windows viruses which attack port 80, so I can just have that blocked (I OH use a different port for HTTP anyway).  Does anyone have any statistics @ concerning which ports are targeted by various things trying to B establish unwanted connections---viruses, port scans and the like?  A Perhaps some sort of intelligent firewall can solve my problem.   I However, this would obviously be at the ISP, and I don't want to rely on eA the ISP too much.  Thus, I would rather manage things as much as e possible on the VMS side.e  C I'm pretty sure that if I could restrict incoming access to just a 8D handful of ports (the server ports I want open and enough to handle C return traffic from outgoing connections) that my problem would be u solved.h   ------------------------------  % Date: Thu, 28 Feb 2002 04:06:33 -0500d- From: JF Mezei <jfmezei.spamnot@videotron.ca>i( Subject: Re: restricting access to ports+ Message-ID: <3C7DF313.97FFA9D@videotron.ca>r   Phillip Helbig wrote:tD > Is there any way to tell VMS to allow incoming connections only to > certain ports?  I Some software do have parameters that allow rejection of calls (or accepto calls only from named hosts).     V For TCPIP services, look at TCPIP> HELP SET SERVICE, especilly the /ACCEPT and /REJECT    I What I have done is install a router between the ISP and me and block thebM calls at the router, and run a SYSLOGD (thanks to John Vottero) server on the2K VAX to collect information on folks who try to connect to me. (router sendsi those logs to SYSLOG).  D Some guy just tried to scan all my ports (lots of bells on OPCOM !).   ------------------------------  + Date: Thu, 28 Feb 2002 10:16:21 +0100 (MET)f9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>.( Subject: Re: restricting access to ports; Message-ID: <01KESYCFH6LA8ZL57Q@sysdev.deutsche-boerse.com>e  D > For TCPIP services, look at TCPIP> HELP SET SERVICE, especilly the > /ACCEPT and /REJECT   ' I see HOSTS and NETWORKS, but no PORTS.n  F Note that I am not mainly trying to block initial incoming access to aD given port, but rather keep VMS itself from "randomly" allocating a G given port for return traffic associated with an outgoing connection I e initiated myself.   C I'm no network-protocoll expert, but presumably when I initiate an  C outgoing connection, I (or my machine), and not the receiving end, VH determine the port at my end used for listening to return traffic.  (If E the other side did this, the port might already be used by something t else.   > > What I have done is install a router between the ISP and me   G I have an ISDN router between me and the ISP (in fact, that is how the oF connection is made), but I have to pay for things which get as far as 9 the router, which is why I want to block them at the ISP.f   > and blockhE > the calls at the router, and run a SYSLOGD (thanks to John Vottero)iI > server on the VAX to collect information on folks who try to connect toy+ > me. (router sends those logs to SYSLOG). n  D Even if I know who made the unwanted connections, I don't see how I 8 could demand payment, which is why I want to block them.  F > Some guy just tried to scan all my ports (lots of bells on OPCOM !).  - Yes, it's this type of stuff I want to avoid.    ------------------------------  % Date: Thu, 28 Feb 2002 05:50:08 -0500e- From: JF Mezei <jfmezei.spamnot@videotron.ca>e( Subject: Re: restricting access to ports* Message-ID: <3C7E0B5E.4FF462@videotron.ca>   Phillip Helbig wrote:s > F > > For TCPIP services, look at TCPIP> HELP SET SERVICE, especilly the > > /ACCEPT and /REJECT  > ) > I see HOSTS and NETWORKS, but no PORTS.i  J But you do this for the defined services. For instance, for SMTP (port 25) you'd do, the SET SERVICE SMTP/ACCEPT=NETWORK=10.0.0.0  @ (you can specify the network mask, but am unsure of the syntax).  N With such a command, the SMTP would only accept incoming calls from your local) LAN network and nothing from the outside.e  M Note that if you don't have a service defined for a port, VMS will refuse theuG call. The dangerous portion though is for services that are dynamicallysG declared (for instance, the OSU web server delcares port 80 by itself).m  H > Note that I am not mainly trying to block initial incoming access to aE > given port, but rather keep VMS itself from "randomly" allocating aaH > given port for return traffic associated with an outgoing connection I > initiated myself.n  & Why are you trying to restrict those ?    H > I have an ISDN router between me and the ISP (in fact, that is how theG > connection is made), but I have to pay for things which get as far asl; > the router, which is why I want to block them at the ISP.m  N Ok, with ISDN I can understand why. A call request to your port 80 would forceJ the ISDN link to be established from the ISP to you only to have your hostL reject it, and the ISDN router keep the line active for X time, during which you are billed.c  I Do they have DSL or CABLE internet where you live ? Those are usually note; measured for connect time so that would solve your problem.s  H > > Some guy just tried to scan all my ports (lots of bells on OPCOM !). > / > Yes, it's this type of stuff I want to avoid.   J Hey, but that is the fun part :-) :-) :-)  It does give one a good idea of) what the script weenies are trying to do.i   ------------------------------  + Date: Thu, 28 Feb 2002 12:46:32 +0100 (MET)t9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>t( Subject: Re: restricting access to ports; Message-ID: <01KET32QYDMY8ZL57Q@sysdev.deutsche-boerse.com>   L > But you do this for the defined services. For instance, for SMTP (port 25)7 > you'd do the SET SERVICE SMTP/ACCEPT=NETWORK=10.0.0.0o > B > (you can specify the network mask, but am unsure of the syntax). > P > With such a command, the SMTP would only accept incoming calls from your local+ > LAN network and nothing from the outside.u  ? Right, but I WANT things from the outside, just not EVERYTHING.e  B On the other hand, everything comes through the ISDN router which 4 connects the local LAN to the ISP over an ISDN line.  O > Note that if you don't have a service defined for a port, VMS will refuse the-I > call. The dangerous portion though is for services that are dynamically I > declared (for instance, the OSU web server delcares port 80 by itself).   G That's not a problem in itself; I want to avoid unwanted requests from -E reaching my VMS machine at all, so that I don't have to pay for them.a  J > > Note that I am not mainly trying to block initial incoming access to aG > > given port, but rather keep VMS itself from "randomly" allocating asJ > > given port for return traffic associated with an outgoing connection I > > initiated myself.  > ( > Why are you trying to restrict those ?  K If I can restrict them to, say, ports 49125--49155 (or however many I need cE simultaneously), then I can tell the ISP to block incoming access to dE everything except these (which are listening for return traffic from oI connections I initiated), ports less than 1024 which I actually use, and sG special cases like 8000, 8080 for HTTP or whatever.  That way, if some 2G joker tries to access a random port as part of a port-scan strategy or dI whatever, or a virus going for a certain port, it will be blocked at the s# ISP and I don't have to pay for it.w  B I have no problem with the accesses hitting the machine; VMS just G refuses.  The problem is that I have to pay for incoming access (if it aB initiates a new connection; connections time out after 2 minutes).  H > I have an ISDN router between me and the ISP (in fact, that is how theG > connection is made), but I have to pay for things which get as far ase; > the router, which is why I want to block them at the ISP.s  J > Ok, with ISDN I can understand why. A call request to your port 80 wouldH > force the ISDN link to be established from the ISP to you only to haveE > your host reject it, and the ISDN router keep the line active for Xm% > time, during which you are billed. i   Exactly.  K > Do they have DSL or CABLE internet where you live ? Those are usually not = > measured for connect time so that would solve your problem.n  C Yes, but my last information from the ISP is that he can offer the sI service I need (dial-in-on demand access, static IP addresses) only over   ISDN.l  E Actually, the ISDN connection is not as bad as it seems and if I get pI things working right, it would be cheaper than DSL etc (I don't need the hG DSL bandwidth) and would also allow me to do things like dial in to my e own router.b  H I have a telephone-billing arrangement which doesn't charge for outgoingA calls on Sundays and holidays, so then I initiate a connection atsE midnight and keep it alive.  I allow robots in only on Sundays, so myaC web pages get accessed by robots---which I want---but only when thel@ connection is free.  Legitimate and/or non-robot access is not aF problem.  Someone once said that "intelligent" robots will black-list B pages and not index them if it is not always accessible, but this I doesn't seem to be the case---I get the same bots as always, but only on  H Sundays.  As long as I do big downloads etc myself only on Sundays, the G connection time and thus the costs are not that much (volume is not an e issue in any case).    ------------------------------  % Date: Thu, 28 Feb 2002 05:58:05 -0500u- From: JF Mezei <jfmezei.spamnot@videotron.ca>  Subject: RMS: rop: RAB$M_LIM, Message-ID: <3C7E0D3A.612C5E11@videotron.ca>  L I looked at the documentation for the rab$l_rop code of RAB$M_LIM and didn't& quite understand exactly how it works.  K If I do a $FIND, and then do sequential $GET with the LIM option, does thisrL mean that RMS will automatically signal to me an end of file (or other code)I once it gets to a record whose key no longer satisfies the original $FINDl specification ?    ------------------------------  # Date: Thu, 28 Feb 2002 17:53:35 GMT + From: Ryan Moore <rmoore@rmoore.dyndns.org>e  Subject: Re: RMS: rop: RAB$M_LIM= Message-ID: <Pine.LNX.4.31.0202280940280.31188-100000@jaipur>   I Not quite.  I've used the LIM option.  My understanding is that rab$b_rac I should be set to RAB$C_SEQ and the LIM bit should be set in the rab$l_roptF field.  Then when performing the GET operation, RMS will check the keyG specified in the RAB (l_kbf, b_ksz, and b_krf fields).  When it finds auG record that has a different key, it will return the status RMS$_OK_LIM.aH But it will *still* get (and lock, if appropriate) that next record withA the next key.  So be sure to do a sys$release() call to free this I unwanted record (I was bit by this little detail).  I guess if you get to / the end-of-file, you'll get an RMS$_EOF status.   E The key specified in the FIND operation will be irrelevant.  Only theoH key in the GET operations matter.  At least, that's how I interpret what) the documentation says about that option.r  I Oh, and if you do a FIND, and then GETs with the LIM bit set, it seems to G me you will skip the first record with that key.  Because the FIND will5J position RMS at the first record, the sequential get should get the *next*C record.  So you'd miss the first one.  I think.  When I use the LIMsE operation, I do a keyed GET to get the first record, and then the SEQt (w/LIM) GETs to get the others.s  $ On Thu, 28 Feb 2002, JF Mezei wrote:N > I looked at the documentation for the rab$l_rop code of RAB$M_LIM and didn't( > quite understand exactly how it works. >bM > If I do a $FIND, and then do sequential $GET with the LIM option, does this-N > mean that RMS will automatically signal to me an end of file (or other code)K > once it gets to a record whose key no longer satisfies the original $FIND- > specification ?-   ------------------------------    Date: 28 Feb 2002 07:20:00 -0600- From: koehler@encompasserve.org (Bob Koehler)S$ Subject: Re: Solution for the merger3 Message-ID: <vOC42G67Q5D3@eisner.encompasserve.org>   q In article <20020227234447.38667.qmail@web20209.mail.yahoo.com>, Fabio Cardoso <fabiopenvms@yahoo.com.br> writes:v > Well >  > Talking about DREAMS !!!!!  E    GM buys up all the streetcars and shuts them down so they can sell 7    more cars.  When cities worry, GM sells them busses..  A    Oh, wait, that was the background of "Who Framed Roger Rabit"..C    Oh, wait again, it really happened that way (OBTW, I was in New d@    Orleans when the Canal Street trolleys were "modernized" into    busses.)     / > - HP buys Compaq and stay commited to the DOEx5 > agreement. Saves OpenVMS, but Tru64 stills menaced.8, > HP will port TruCluster software to HP-UX.  F    HP buys Compaq to get VMS and Tru64 clustering, which they've been F    jealous of for a decade.  Further HP stops MPE so it won't compete :    with VMS, which has larger margins and market presence.  . > - IBM buys DELL or make an agreement saying:  F    IBM buys Dell and shuts them down because all those stinking little8    no profit PC's are cutting into their real bussiness.  G Reality check:  I'm not holding my breath waiting to buy my first cheapFD IA-64 laptop with VMS on it.  2004 is too far away; my Alpha will do fine until then.   ------------------------------    Date: 28 Feb 2002 17:37:33 +0100O From: pmoreau@dev.ath.cena.fr (Patrick MOREAU, CENA Athis, Tel: 01.69.57.64.40)h  Subject: Re: Sound on alpha/VMS?  Message-ID: <rasNhsB7kq0g@sable>  / In article <a5j9qi$87l$1@newslocal.mitre.org>,  : lewis@lumina.removethis.mitre.org (Keith A. Lewis) writes:I > I see some audio ports in the back of my Alphastation 500au.  Are theseh > usuable under VMS at all?-  -: > --Keith Lewis              klewis@mitre.remove-thisx.org@ > The above may not (yet) represent the opinions of my employer.  F You need MMOV (Multimedia Services for OpenVMS, the pack was generallyH available with VMS on a workstation). MMOV supports the olders MicrosoftI Compatible sound cards found on alphastations and also new Compaq Ensoniqe6 AudioPCI available for recent hardware (DS10 etc ...).  N MMOV gives you a basic sound player - recorder (WAV format or 8bit sun format)O but some interesting ports of Unix sound applications are available at the DECW D archive (mp3 player, midi player, midi sequenncer and score editor):  " http://decwarch.free.fr/audio.html   Enjoy !!   Patricka --O ===============================================================================eN pmoreau@ath.cena.fr  (CENA)     ______      ___   _           (Patrick MOREAU)4 moreau_p@decus.fr (DECUS)       / /   /     / /|  /|J CENA/Athis-Mons FRANCE         / /___/     / / | / |   __   __   __   __  N BP 205                        / /         / /  |/  |  |  | |__| |__  |__| |  |N 94542 ORLY AEROGARE CEDEX    / /   ::    / /       |  |__| | \  |__  |  | |__|N http://www.ath.cena.fr/~pmoreau/            http://www.multimania.com/pmoreau/O ===============================================================================    ------------------------------  % Date: Thu, 28 Feb 2002 10:04:22 +0100l3 From: "Aleksander Koodziejczyk" <olekk@kki.net.pl>l, Subject: Steps for installation OVMS 6.2 1H3- Message-ID: <a5krrj$49m$1@news2.ipartners.pl>b  D I'll be install OVMS 6.2, I never to do it before, it is difficult ?$ Can you tell me any steps, to do it.   Regards, Aleksander Koodziejczyk   ------------------------------  % Date: Thu, 28 Feb 2002 10:12:01 +0100e9 From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com>c0 Subject: Re: Steps for installation OVMS 6.2 1H3& Message-ID: <3C7DF461.314D29A@aaa.com>  @ Get the install manual from http://www.openvms.compaq.com:8000/.   or  = You should have got a install manual together with you media.n  2 And if it's difficult. No, not the second time :-)  < The actual steps depends on a number of things, such as what9 hardware you are using. It's almost impossible to give anr! step-for-step instruction here...y   Jan-Erik Sderholm.o  ! "Aleksander Koodziejczyk" wrote:t > F > I'll be install OVMS 6.2, I never to do it before, it is difficult ?& > Can you tell me any steps, to do it. > 
 > Regards, > Aleksander Koodziejczyk   ------------------------------  # Date: Thu, 28 Feb 2002 12:42:41 GMTa" From: "Hans Vlems" <hvlems@iae.nl>0 Subject: Re: Steps for installation OVMS 6.2 1H3/ Message-ID: <5Bpf8.46$Or2.1565@typhoon.bart.nl>   L Installing VMS is not difficult. Since V6.2 installs on VAX as well as Alpha I'llL try and cover both. But I have no Alpha at home so mistakes are very likely.5 Furthermore, my guess is that you have the kit on CD.aK First, establish the SCSI id of the CD player and the SCSI id of the targete disk.f On most VAXes try:   >>> SHOW DEV   On Alpha's try:,  
 >>> SHOW DISKu  F Before the installation, make sure you have the following information:# - a password for the SYSTEM account . - passwords for the FIELD and SYSTEST accountsG - a name for the machine (up to 6 ASCII characters, not case sensitive)s- - a SCSSYSTEMID (in fact the DECnet address).nH If this is a hobby system , use SCSSYSTEMID 1025 (that results in DECnet address 1.1)  C Let's assume that the CD drive is DKA400 and the target drive DKB0.d
 On the Alpha:i   >>> b -fl 0,0 dka400  I The system boots a small version of VMS and you'll get a menu. One of theFG options is "install VMS" or something similar. If you have licenses forV TCPIPlB and/or DECnet you can also install these network products as well.   >>> b dka400  6 After a while the system gives you the $ prompt. Type:  + $ backup dka400:[0,0]vaxvms062.b/save dkb0:k  . As soon as it has finished, reboot the system. Check your boot settings:v  1 >>> SHO BFLG                    ! should return 0 1 >>> SHO HALT                    ! should return 3(& >>> SHO BOOT                    ! DKB0  E If the latter was set to DKB0 then you do not need to reboot manuallyd  during the installation process.  
 >>> B dkb0  K At this point the partially installed VMS system will boot and  contine the- installationJ process. Depending on type of cpu that may take between 20 and 80 minutes.  E IMPORTANT: the installation process initializes the target disk so be- careful.J You need a 400 MB disk to install everything, but a 240 MB disk is a tight fit.L If you have the DW-MOTIF license, install DECwindows (or Motif as it is also called)u/ because that makes life handy with a PC around.t  	 Have fun,a   Hans   For the VAX the procedure is< Aleksander Koodziejczyk <olekk@kki.net.pl> wrote in message' news:a5krrj$49m$1@news2.ipartners.pl...oF > I'll be install OVMS 6.2, I never to do it before, it is difficult ?& > Can you tell me any steps, to do it. > 
 > Regards, > Aleksander Koodziejczyk >  >    ------------------------------  # Date: Thu, 28 Feb 2002 14:49:58 GMTr8 From: hammond@not@peek.ppb.cpqcorp.net (Charlie Hammond)0 Subject: Re: Steps for installation OVMS 6.2 1H31 Message-ID: <qsrf8.324$fL6.6080@news.cpqcorp.net>a  . In article <a5krrj$49m$1@news2.ipartners.pl>, 5 "Aleksander Koodziejczyk" <olekk@kki.net.pl> writes:b  E >I'll be install OVMS 6.2, I never to do it before, it is difficult ?y% >Can you tell me any steps, to do it.   F First, as with any new version of OpenVMS, it is *IMPORTANT* that you G revieiw the cover letter, release notes, and installation guide for thec version you are installing.     3     Please note V6.2 and V6.2-1H3 are different!!!!   B I am assuming that this is an OpenVMS Alpha V6.2-1H2 installation.  D If this is a NEW INSTALLATION (not an upgrade of an existing system)G then the instruction is simple -- Boot from the OpenVMS Alpha operatingaG system CD-ROM and follow the menus and instructions that are presented.0C (It will be a LOT easier if you read the installation guide first!):  J If this is an UPGRADE of an existing system, it is a bit more complicated.J This is because upgrading (or "updating", if you wish) to V6.2-1H3 is doneI using a VMSINSTAL kit.  You must first re-boot your exisiting V6.2 systemRF in a "minimum" configuation.  Then you invoke VMSINSTAL as outlined inH the installation guide.  NOTE: you can only upgrade to V6.2-1H3 from oneF of the follwing versions: V6.2, V6.2-1H1 or V6.2-1H2.  If you have anyF other (older) version, you will first have to upgrade to V6.2 and then to V6.2-1H3 as a second step.o  L Now a question -- WHY V6.2-1H3???  This is an old version.  Unless you have N some particular need for this old version, I would recommend that you install 3 V7.2-2 (or maybe V7.2-1 or V7.2-1H1 in some cases).    -- vK     Charlie Hammond -- Compaq Computer Corporation -- Pompano Beach  FL USAcH        (hammond@not@peek.ppb.cpqcorp.net -- remove "@not" when replying)J       All opinions expressed are my own and not necessarily my employer's.   ------------------------------  % Date: Thu, 28 Feb 2002 19:12:03 +0100e4 From: "Aleksander Koodziejczyk" <ako@simple.com.pl>0 Subject: Re: Steps for installation OVMS 6.2 1H3- Message-ID: <a5lrug$qr1$1@news2.ipartners.pl>   H > Now a question -- WHY V6.2-1H3???  This is an old version.  Unless you haveG > some particular need for this old version, I would recommend that your installS5 > V7.2-2 (or maybe V7.2-1 or V7.2-1H1 in some cases).a  K Thank you, but we haven't licence for VMS7.x. I'll be install V6.2-1H3. OuruF server is ALPHA1000 with 128MB RAM (I affraid it is to small for newerF version VMS). We use this server for developing programs (C, GEMBASE).  I We heve now VMS6.2, but something damaged with file system on system disksK and we must install/reinstall VMS (VMS ont this server was instaled in 1995  :) ).b  H Now, our server regular reboot when we use DCL command like "edit file".   Regards, Aleksander Kolodziejczyk   ------------------------------  % Date: Thu, 28 Feb 2002 10:31:19 +0000M* From: Ian Dean <ian.d.dean@baesystems.com> Subject: Use of PCA on Alpha. Message-ID: <3C7E06F7.E9AC19BC@baesystems.com>   Hi,YC 	I'm sorry if this is an old query, but it's the first time on thise group.  % I have a small Fortran file (TX.FOR):   
 	DO J = 1, 10T 		WRITE(*,*)'OK' 	END DOK 	END  H I am trying to get IO stats using PCA. The method works OK on a VAX, but( states the IO count is zero on an Alpha.   e.g.   ON VAX (VMS 5.4-2 PCA 3.0-8) ======   	FOR/DEB TX $ 	LINK/DEB=SYS$LIBRARY:PCA$OBJ.OBJ TX 	RUN TX  	....f
 	PCAC> SET PC,
 	PCAC> SET IOT	 	PCAC> GOn 	....h 	....,$ 	%PCA-I-ENDCOL, data collection ends 	$   	PCA TX.PCAs 	....c 	PLOT/IO 	.... ! 	....	Produces acceptable IO plotf     ON ALPHA (oVMS V7.1 PCA 4.5-4) ========   	FOR/DEB TX  	LINK/DEB TX 	RUN TXr 	....t
 	PCAC> SET PCo
 	PCAC> SET IO 	 	PCAC> GOa 	....  	....d$ 	%PCA-I-ENDCOL, data collection ends 	$   	PCA TX.PCAs 	....  	PLOT/IO< 	%PCA-W-NODATCNT, count is zero for I/O System Service Calls   Is this a known bug ?n Am I doing something wrong ?C I even set the loop size to 1000 !!! to ensure the program ran longn enough - still no difference. D Obviously the example given is a (extremely) cut-down version of the real McCoy.t TIAr 	Ian   ------------------------------    Date: 28 Feb 2002 08:19:15 -08005 From: pat.saunders@sis.securicor.co.uk (pat saunders)I. Subject: Re: vax/vms documentation - psi / x29< Message-ID: <bc0e3bd8.0202280819.af7766a@posting.google.com>  e "Bob Knowles" <bob.knowles@compaq.com> wrote in message news:<pV7f8.259$fL6.5523@news.cpqcorp.net>...o4 > "Alan Greig" <a.greig@virgin.net> wrote in message4 > news:0epp7ucgiq9hds4emoqigm275ubesp3bi9@4ax.com...5 > > On Wed, 27 Feb 2002 09:29:18 -0000, "Bob Knowles" # > > <bob.knowles@compaq.com> wrote:p > >c > >rN > > >As Pat specified VAX, and the VAX PSI product was retired before Phase V,K > > >sure as eggs is eggs the Phase V info won't be much (if any) help. Andc > >  > > Is that strictly true? > L > Nope. I checked the dates on the Alpha docs, and there was an overlap of 2 > or 3 yearsM > between the end of Reading's development of VAX PSI and the availability ofwM > the Alpha product, so Pat may well be running the old product over Phase V.  >  > br       Hi,c3   Yes I am using phase v , DECNET/OSI and VMS 5.5-2oE   We have a number of VAX's that work with X.25 but we did not make alB backup, so when a particular VAX failed we had to reconfigure from stratch.=   I still get PAD-F-NOLINES but have found out more details : +   I have a LAPB link to a x.25 mini switch.u-   Every ten secs I get the following errors :o  $   Message from user SYSTEM on KEWG99B Event: Link Setup Failed from: Node LOCAL:.KEWG99 LAPB Link DTE-0,-         at: 2002-02-28-17:16:00.693+00:00Iinft7         eventUid   D645CF84-2C6E-11D6-8008-AA0004009E2Av7         entityUid  3718B880-2C6E-11D6-8007-AA0004009E2A)7         streamUid  61C47CE0-2C6E-11D6-8008-AA0004009E2An     NCL>8 %%%%%%%%%%%  OPCOM  28-FEB-2002 17:16:00.75  %%%%%%%%%%%" Message from user SYSTEM on KEWG99B Event: Link Initialising from: Node LOCAL:.KEWG99 LAPB Link DTE-0,-         at: 2002-02-28-17:16:00.693+00:00Iinfo.         Link Initialising Reason=Maximum Retry7         eventUid   D645CF85-2C6E-11D6-8008-AA0004009E2At7         entityUid  3718B880-2C6E-11D6-8007-AA0004009E2Ar7         streamUid  61C47CE0-2C6E-11D6-8008-AA0004009E2Ab  E when I look at the DTE state, I find it is "Unsynchronised", I am notm( sure what this means or how to fix it ??   Any help appreciated.h   ta patg   ------------------------------  % Date: Thu, 28 Feb 2002 08:44:03 +0100a1 From: "Tomasz Dryjanski" <tdryjanski@hotmail.com>- Subject: Re: Very slow ftp, Message-ID: <a5kn44$m8$1@news2.ipartners.pl>  F > > I observe a very slow ftp transfer on my system (3.76 kB/s, ie. 30 kb/s). >nG > Is the AlphaServer the FTP client or the FTP server? What's the other-
 > machine?  F The AlphaServer is the FTP server, I used NT 4.0 and Win95 as clients.  E > Have you analyzed system performance on either system (overburdened1 > disks, swamped CPUs, etc.)?d    > Both the server and a client were almost idle during the test.    E > Have you tried a protocol analyzer to watch what's actually passingq > over the wire?   Nope. Perhaps I should try it.   T. D.i   ------------------------------  % Date: Thu, 28 Feb 2002 09:12:45 +0100t2 From: "Ren Schelbaum" <rene.schelbaum@datakom.at> Subject: Re: Very slow ftpG Message-ID: <3c7de676$0$28220$5039e797@newsreader01.highway.telekom.at>   B "Tomasz Dryjanski" <tdryjanski@hotmail.com> schrieb im Newsbeitrag( news:a5in0u$20qb$1@news2.ipartners.pl...K > I observe a very slow ftp transfer on my system (3.76 kB/s, ie. 30 kb/s).nH > I use an AlphaServer 1000A with DE500 network interface (10 Mb/s set).L > I have VMS 7.2-1H1 installed on it with all mandatory patches, and those IF > thought could affect ftp performance. I have also tried an insulated network.F > (one small hub, one laptop, two new patchcords) to eliminate networkE > problems. I have tried another DE500. And there is still no result.) >a2 > Have you got any idea what I could do to fix it? >y > TIAs >  > -- >r > T. D.  >a >d   Hi!   8 Had a similar problem some time ago. It turned out to beJ auto-negotiation-failure between the server and the ethernet-switch it wasJ connected to. One of them assumed full-duplex while the other assumed halfK duplex. That resulted in excessive collisions and a hughe performance-drop.aI I solved it by manually setting the switch and the server to full duplex.    Ren   ------------------------------  % Date: Thu, 28 Feb 2002 13:28:10 +0100l1 From: "Tomasz Dryjanski" <tdryjanski@hotmail.com>l Subject: Re: Very slow ftp- Message-ID: <a5l7os$dpr$1@news2.ipartners.pl>   K > I observe a very slow ftp transfer on my system (3.76 kB/s, ie. 30 kb/s).tH > I use an AlphaServer 1000A with DE500 network interface (10 Mb/s set).L > I have VMS 7.2-1H1 installed on it with all mandatory patches, and those IF > thought could affect ftp performance. I have also tried an insulated networkaF > (one small hub, one laptop, two new patchcords) to eliminate networkE > problems. I have tried another DE500. And there is still no result.n  D I have just observed, that ftp via WAN gives 800 kb/s (!). And it is reasonable performance.-5 So perhaps there is some TCPIP configuration problem?:I I have reviewed the configuration on the server, but perhaps you have gott
 some idea.   T. D.e   ------------------------------    Date: 28 Feb 2002 07:53:39 -0800? From: seos@nuigalway.ie (=?ISO-8859-1?Q?Seos_mac_C=E1rthaigh?=)a2 Subject: Re: VMS single Login with NDS e-directory= Message-ID: <faf6acd5.0202280753.1463f865@posting.google.com>s  9 Hi folks thanks for taking the time to reply to me query,w   I appreciate it,   Seosamh    ------------------------------  % Date: Thu, 28 Feb 2002 08:52:44 +0100w9 From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com>d, Subject: Re: What return codes mean success?' Message-ID: <3C7DE1CC.8DADE77C@aaa.com>   > Hm, one of the most easy-to-read descriptions off status codes? I'v seen comes from chapter 8 "Error Handling" in "Writing RealoE Programs in DCL, Second Edition" (Digital Press, ISBN 1-55558-191-9).w  5 Co-written by someone called Steve Hoffman, b.t.w :-)H   Jan-Erik Sderholm.      Hoff Hoffman wrote:i > J >   Please acquire and read the programming concepts manual in the OpenVMSH >   documentation set, and specifically please review the information in6 >   that manual on the OpenVMS condition value format. >    ------------------------------  % Date: Thu, 28 Feb 2002 03:20:39 -0500i- From: JF Mezei <jfmezei.spamnot@videotron.ca>i, Subject: Re: What return codes mean success?, Message-ID: <3C7DE855.36359267@videotron.ca>   Jan-Erik Sderholm wrote: G > Programs in DCL, Second Edition" (Digital Press, ISBN 1-55558-191-9).6 > 7 > Co-written by someone called Steve Hoffman, b.t.w :-)B  " Is that Hoff's evil twin brother ?   :-)l   ------------------------------   Date: 28 Feb 2002 09:49 CSTe' From: carl@gerg.tamu.edu (Carl Perkins) , Subject: Re: What return codes mean success?- Message-ID: <28FEB200209494718@gerg.tamu.edu>e  % Ernest <wmozart5@yahoo.com> writes...x7 }gartmann@immunbio.mpg.de (Christoph Gartmann) wrote in H }> Even return codes are bad, odd ones mean success. Thus, you just need }> to test the last bit. } 5 }Well, that was easier than I expected.  8^)  Thanks!t }  }Cheers, }Ernests  > Although, as has been pointed out, it isn't quite that simple.  = Going with just even-odd will count informational messages ass< success, which is probably OK most of the time but might not: be in some cases. As an example, sometimes you may want to> detect the informational messages and add them to your output.A This will also count warnings as errors, which is againt probablyn< OK most of the time but sometimes you may want to treat them differently.  B The actual status code part of each of your messages was the same,@ a 1. The difference was in the facility code. The numbers encodeB which type of thing it is reporting, what severity it is, and what
 it came from._  	 Consider:   	 $ z=65537  $ sh symb z 1   Z = 65537   Hex = 00010001  Octal = 00000200001 
 $ exit z+1+ %RMS-E-NORMAL, normal successful completion|
 $ z=196609 $ sh symb z|2   Z = 196609   Hex = 00030001  Octal = 00000600001
 $ exit z+1+ %DCL-E-NORMAL, normal successful completion= $ z=12124161 $ sh symb z=4   Z = 12124161   Hex = 00B90001  Octal = 00056200001
 $ exit z+1* %CC-E-NORMAL, normal successful completion  G (The +1 was to make them errors instead of successes so that the "exit"o+ command would actually report the message.)s   --- Carl   ------------------------------  % Date: Thu, 28 Feb 2002 19:09:07 +0100?9 From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com>d, Subject: Re: What return codes mean success?' Message-ID: <3C7E7243.BCF96EAB@aaa.com>0   The status code is 32 bits.-  E bit 0-2    : Severity  (0:warn, 1:success, 2:error, 3:info, 4:severe)S bit 1-14   : Status codeE bit 15     : (0:status code shared by multiple facilities, 1:facilitys	 specific):G bit 16-26  : Facility number (either a Compaq facility or a third-partyd	 facility) F bit 27     : (0:a Compaq fac or a third-pary fac registered by Compaq)E bit 28     : (1:inhibit thh DCL EXIT command or the $EXIT sys serv tom displays               the message) bit 29-31  : reservedy  ? I think I'v seen someware that you can (or could) register your G own application at DEC/Compaq, so you could have your own status codes.   B I'v written a short DCL hack to display the different parts of the status message.    $! P1 = status value $! $ mess_severity = p1 .and. %X7' $ mess_numb     = (p1 .and. %X7FF8) / 8>+ $ mess_facspec  = (p1 .and. %X8000) / 327682- $ mess_facnumb  = (p1 .and. %X7FF000) / 65536r1 $ mess_custfac  = (p1 .and. %X800000) / 134217728.2 $ mess_inhibdis = (p1 .and. %X1000000) / 268435456 $!A $ write sys$output "Message severity            : ",mess_severityt= $ write sys$output "Message number              : ",mess_numbb@ $ write sys$output "Message fac specific        : ",mess_facspec@ $ write sys$output "Message facility number     : ",mess_facnumb@ $ write sys$output "Message customer facility   : ",mess_custfacA $ write sys$output "Message inhibit dislay      : ",mess_inhibdiss  3 Example (The DCL hack is called parse_status.com) :t    
 $ dir fffffff ! %DIRECT-W-NOFILES, no files founde   $ @parse_status '$status't Message severity            : 0a  Message number              : 82 Message fac specific        : 1s Message facility number     : 1D Message customer facility   : 0r Message inhibit dislay      : 0    $ pu asfasdfF %PURGE-W-SEARCHFAIL, error searching for SYS$COMMON:[SYSMGR]ASFASDF.;* -RMS-E-FNF, file not found   $ @parse_status '$status'  Message severity            : 0n! Message number              : 583e Message fac specific        : 0>  Message facility number     : 19 Message customer facility   : 0: Message inhibit dislay      : 02   $ convert xxxxx zzzzz A %CONV-F-OPENIN, error opening SYS$COMMON:[SYSMGR]XXXXX.; as input  -RMS-E-FNF, file not found   $ @parse_status '$status'  Message severity            : 4u! Message number              : 531  Message fac specific        : 0   Message facility number     : 50 Message customer facility   : 0e Message inhibit dislay      : 0k  E So DIR i facility no 1, PURGE is no 19 and CONVERT is facility no 50.t   Regardsn Jan-Erik Sderholm   ------------------------------  % Date: Thu, 28 Feb 2002 11:27:33 +0100T( From: Paul Sture <paul.sture@bluewin.ch>/ Subject: Re: Why does ERRFMT give me RSZ error?M) Message-ID: <3C7E0615.4010003@bluewin.ch>d   JF Mezei wrote:.   > Lawrence Bleau wrote:l > : >>%%%%%%%%%%%  OPCOM  27-FEB-2002 21:44:11.39  %%%%%%%%%%%# >>Message from user SYSTEM on UMDSP ) >>ERRFMT - ERROR ACCESSING ERROR LOG FILEA! >>%RMS-F-RSZ, invalid record sized >> > F > In VAX VMS 7.2, a dump of the file reveals some interesting tidbits: > < > SYS$SPECIFIC:[SYSEXE]LISTPREPROC.DAT	-> what is that for ? > ' > There is something new I didn't know:  > O > Apparently ERRFMT has the ability to mail something to someone. There are two- > logicals defined:2 > ERRFMT$_SEND_MAIL "TRUE" > ERRFMT$_SEND_TO   "SYSTEM" > 0 > Under what conditions does ERRFMT mail stuff ? > Y > Also, check the logical for SYS$ERRORLOG to make sure it points to the right directory.4P > ( SYS$SYSROOT:[SYSERR]) If it points to another directory where there might beF > another errlog.sys file in a wrong file format, it might explain it. >  > R > And an easter egg too: "ERRFMT - I bring you bad news" somewhere in the program. >   F Yep, that's the subject of the MAIL it sends to SYSTEM as it's dying. G IIRC there's no text body, to minimize the message not getting through 6 due to lack of disk space.  H The first time I saw it had me falling out of my chair with laughter :-)   ------------------------------  % Date: Thu, 28 Feb 2002 16:50:47 +0000h( From: Nic Clews <sendspamhere@127.0.0.1>/ Subject: Re: Why does ERRFMT give me RSZ error?o) Message-ID: <3C7E5FE7.3F9BD327@127.0.0.1>e   JF Mezei wrote:b > ? > SYS$SPECIFIC:[SYSEXE]LISTPREPROC.DAT    -> what is that for ?  >   * Is it anything to do with LISTPREPOP.DAT ?  H (Preallocates chunks of nonpaged pool for subsequent boots, ERRFMT looks
 after it.)   -- s( Regards, Nic Clews CSC Computer Sciences nclews at csc dot coms   ------------------------------  % Date: Thu, 28 Feb 2002 17:17:11 +0000   From: Steve.Spires@yellgroup.com/ Subject: Re: Why does ERRFMT give me RSZ error?e: Message-ID: <OF6918578B.B9E3E990-ON00256B6E.005ED174@btyp>  D I hope so, because I've spent a while trying to find LISTPREPROC.DAT( amongst my systems and haven't found it.  D Started to think I had a problem, because it was close enough to the6 LISTPREPOP.DAT filename to make me think I needed one!   ;^Dd   Cheers   Steve S7        < Nic Clews <sendspamhere@127.0.0.1> on 02/28/2002 04:50:47 PM    To:        Info-VAX@Mvb.Saic.Com cc:-J From:      Nic Clews <sendspamhere@127.0.0.1>, 28 February 2002, 4:50 p.m.  & Re: Why does ERRFMT give me RSZ error?     JF Mezei wrote:  > ? > SYS$SPECIFIC:[SYSEXE]LISTPREPROC.DAT    -> what is that for ?  >a  * Is it anything to do with LISTPREPOP.DAT ?  H (Preallocates chunks of nonpaged pool for subsequent boots, ERRFMT looks
 after it.)   --( Regards, Nic Clews CSC Computer Sciences nclews at csc dot comi          F ______________________________________________________________________     [Information] -- PostMaster:D This transmission is intended solely for the addressee(s) and may beG confidential. If you are not the named addressee, or if the message hasbG been addressed to you in error, you must not read, disclose, reproduce,t$ distribute or use this transmission.  H Delivery of this message to any person other than the named addressee isG not intended in any way to waive confidentiality.  If you have receivednK this transmission in error please contact the sender or delete the message.,  
 Thank you.  D Yell Limited, Queens Walk, Oxford Road, Reading, Berkshire, RG1 7PT.; Registered in England and Wales, registered number 4205228.e  I Yellow Pages Sales Limited, Queens Walk, Oxford Road, Reading, Berkshire,>D RG1 7PT. Registered in England and Wales, registered number 1403041.   ------------------------------  # Date: Thu, 28 Feb 2002 17:21:15 GMTt1 From: "Mark D. Jilson" <jilly@clarityconnect.com> / Subject: Re: Why does ERRFMT give me RSZ error?E2 Message-ID: <3C7E674F.FA4F7F86@clarityconnect.com>  E This can happen for 1 of 2 reasons but both relate to a 'corrupt' EMBlH record.  When ERRFMT starts it will attempt to process the EMBs that areB stored in the system dump file (for newer versions of VMS on AlphaE SYS$ERRLOG.DMP) and any EMBs that are in memory and write them to thetA system errlog file.  If one of these EMBs are 'corrupt' then thiscE situation may occur.  The way to clear it is to either force a system D crash to rewrite the EMBs in the dump file or to create a new systemD dump file and reboot.  This situation is documented in DSNlink and aF future version of VMS will be able to workaround the issue without the' impact that current workarounds impose.    Lawrence Bleau wrote:g > J > I just noticed a problem: my error logging process (ERRFMT) is no longerL > running.  I restarted it with a  @SYS$SYSTEM:STARTUP ERRFMT  and monitoredJ > OPCOM messages.  It gave me 20 error messages via OPCOM then died again. > The error messages were: > : > %%%%%%%%%%%  OPCOM  27-FEB-2002 21:44:11.39  %%%%%%%%%%%# > Message from user SYSTEM on UMDSPs) > ERRFMT - ERROR ACCESSING ERROR LOG FILE0! > %RMS-F-RSZ, invalid record sizeo > N > I deleted all ERRLOG.SYS files in SYS$SPECIFIC:[ERRLOG] and retried it, sameN > result.  A DIR/FULL of the topmost ERRLOG.SYS files shows nothing anomalous;N > I show it below.  The disk (system disk) has loads of free space, is mountedK > okay, is writable, etc., so there's no reason ERRFMT shouldn't be able tocQ > create a file.  In fact, it *did*, as my tests show.  It just couldn't write to  > the file it created! > M > This is, like, wierd.  And the system was working okay last month, too.  InyP > fact, I'm in the habit of renaming my ERRLOG.SYS files to archive the last few? > months.  A DIR/FULL of the Jan 2002 file is also shown below.o > F > I've never seen this before, and haven't a clue as to how to fix it. > I > My system is a VAXstation 4000-90 running OpenVMS VAX V5.5-2.  I had nopJ > problems in this regard before, so the OS level shouldn't be the reason.Q > I haven't upgraded or patched anything yet this year, haven't installed any neweC > products, etc.  Something changed, but I don't know what.  Ideas?o > N > Btw, I tried a HELP/MESSAGE RSZ on another system, but it didn't give me any > hints. > ! > Directory SYS$SPECIFIC:[SYSERR]e > 8 > ERRLOG.SYS;63                 File ID:  (5240,10473,0)2 > Size:            2/3          Owner:    [SYSTEM]# > Created:  27-FEB-2002 21:44:11.70d' > Revised:  27-FEB-2002 21:44:11.91 (1)  > Expires:   <None specified>E! > Backup:    <No backup recorded>t  > File organization:  SequentialX > File attributes:    Allocation: 3, Extend: 0, Global buffer count: 0, No version limit8 > Record format:      Variable length, maximum 756 bytes > Record attributes:  None > RMS attributes:     None > Journaling enabled: None? > File protection:    System:RWED, Owner:RWED, Group:RE, World:d > Access Cntrl List:  None >  > Total of 1 file, 2/3 blocks. > ! > Directory SYS$SPECIFIC:[SYSERR]  > 8 > ERRLOG200201.SYS;1            File ID:  (1320,43751,0)2 > Size:           68/69         Owner:    [SYSTEM]# > Created:   1-JAN-2002 00:08:19.42 * > Revised:   1-FEB-2002 00:00:02.65 (4609) > Expires:   <None specified>-# > Backup:    4-FEB-2002 15:35:23.772  > File organization:  SequentialY > File attributes:    Allocation: 69, Extend: 0, Global buffer count: 0, No version limitw8 > Record format:      Variable length, maximum 304 bytes6 > Record attributes:  Carriage return carriage control > RMS attributes:     None > Journaling enabled: None? > File protection:    System:RWED, Owner:RWED, Group:RE, World:T > Access Cntrl List:  None >   > Total of 1 file, 68/69 blocks. >  > Lawrence Bleau > University of Maryland$ > Physics Dept., Space Physics Group > 301-405-6223 > bleau@umtof.umd.edut   -- .D Jilly	- Working from Home in the Chemung River Valley - Lockwood, NY0 	- jilly@clarityconnect.com			- Brett Bodine fan. 	- Mark.Jilson@Compaq.com			- since 1975 or so, 	- http://www.jilly.baka.com               -   ------------------------------  % Date: Thu, 28 Feb 2002 09:50:16 -0600w1 From: "Dave Gudewicz" <david.gudewicz@abbott.com>aL Subject: Re: Why is XP-1000 RAM so expensive from Compaq vis-a-vis Kingston?8 Message-ID: <a5ljou$nqb$1@fizban.fizban.pprd.abbott.com>  D I have often said that certain vendors think memory is made out of aJ precious metal, rather then well, sand.  The profit they must make on this is off the Richter scale.1  J I guess this thinking has not changed.  When I buy a system, I'll order itK with minimum memory, then buy third-party stuff for way less.  So far, thisb' hasn't been a problem, at least for me.i   Dave...k  : "Petros Dafniotis" <pdafniotis@yahoo.com> wrote in message7 news:e54adf36.0202280714.46145e1c@posting.google.com...oD > In another post I asked for XP-1000 ram. I need 4 modules of 256MB > each.r >e > This is listed as:9 > Compaq reference numbers are: SN-MSP01-KD or 388262-B21  > This is from:sJ > http://www.compaq.com/alphaserver/options/xp1000/xp1000_sn-msp01-kd.html >uG > Compaq Switzerland quoted this as 10,000 CHF; Compaq USA as 6,300 USDr > so almost the same.i >yF > Now I go to Kingston which I thought was rather reputable and I see: >-L http://www.ec.kingston.com/ecom/kepler/PartsInfo_Bod.asp?ktcpartno=KTC-XP100 0/1024 > which retails for US$ 729!!! > F > This is a HUGE difference! Can I please get an explanation? I do notF > mind at all paying twice or three times as much iff the RAM is goingB > to be more reliable but the price factor of ~9 is just enormous. >1 > Kind regards,  > Petros > ---i > Petros Dafniotis, PhDa > pdafniotis@yahoo.com   ------------------------------  # Date: Thu, 28 Feb 2002 18:09:57 GMTt* From: "Bill Todd" <billtodd@metrocast.net>L Subject: Re: Why is XP-1000 RAM so expensive from Compaq vis-a-vis Kingston?C Message-ID: <Vnuf8.214443$Aw2.16744887@bin7.nnrp.aus1.giganews.com>v  < "Dave Gudewicz" <david.gudewicz@abbott.com> wrote in message2 news:a5ljou$nqb$1@fizban.fizban.pprd.abbott.com...F > I have often said that certain vendors think memory is made out of aL > precious metal, rather then well, sand.  The profit they must make on this > is off the Richter scale.e >nL > I guess this thinking has not changed.  When I buy a system, I'll order itH > with minimum memory, then buy third-party stuff for way less.  So far, this) > hasn't been a problem, at least for me.e  I Hmmm.  Exactly the same advice I gave to Rob when he was pushing the idea-E that Dell's Itanic server was a great bargain compared with an Alpha.r0 Funny, he seems to have dropped that approach...   - bill   ------------------------------  % Date: Thu, 28 Feb 2002 09:55:49 -0600  From: fwheeler@csc.com% Subject: YA Amusing Message from DEC.s9 Message-ID: <OF34C3ABFD.9D691E64-ON86256B6E.00568E62@com>g  ? I was amused by the following that I stumbled upon on an HSC70.5B (Despite what you *might* be thinking, I believe TWATCH stands for Tape WATCH).   Cheers... /Scott Wheeler? Computer Sciences Corporation @ Lockheed Martin Aeronautics Co.s
 Fort Worth TXO   HSC>
 run twatch TWATCH-Q Interval (secs)?  ^Y6 Thanks for watching the Tape code, so few people do... HSC>   ------------------------------    Date: 28 Feb 2002 05:52:59 -0800( From: bob@instantwhip.com (Bob Ceculski)! Subject: Re: [Q] internet and VMSa= Message-ID: <d7791aa1.0202280552.3f815c38@posting.google.com>3  u martin@radiogaga.harz.de (Martin Vorlaender) wrote in message news:<3c79aec9.524144494f47414741@radiogaga.harz.de>...5+ > Bob Ceculski (bob@instantwhip.com) wrote:l8 > > "David J. Dachtera" <djesys.nospam@fsi.net> wrote...H > > > TCPware for OpenVMS from Process Software, http://www.process.com/K > > > TCPware was written from the ground up on VMS. It's kind of a pain toc< > > > manage, but is very fast for some important functions. > > N > > what are you talking about a pain to manage?  it is far easier than ucx orO > > multinet, and once you understand the tcpware_configure.com file, you don'tu > > even need $ NETCU anymore, > G > TCPWARE_CONFIGURE.COM doesn't carry an NETCU settings. It's all about,
 > CNFNET.COM.l > M > 1. I thought you violently opposed having to edit ASCII configuration filesnN >    at some time? Once you understand HTTPD.CONF, configuring Apache is easy. > N > 2. The WEBCNF (Web GUI) component has very limited functionality in TCPware. >  > cu,e
 >   Martin  G wrong!  tcpware_configure.com carries netcu settings ... one example ise         $ NETCU SET GATEWAYlF         this can be done by edt on the above con ... there are severalM         others I can do also ... when I say eliminate netcu almost completely L         I also mean cnfnet ... and editing tcpware_configure is a lot easier7         than editing 8 million conf files in apache ...c   ------------------------------   End of INFO-VAX 2002.116 ************************