1 INFO-VAX	Wed, 24 Jul 2002	Volume 2002 : Issue 405       Contents:$ Re: Another article mentions OpenVMS! BACKUP to remote disk via DECNET? % Re: BACKUP to remote disk via DECNET?  CWSB -Remote & -Raise Switches" Re: CWSB -Remote & -Raise Switches% Re: Cygwin/xfree86 and VMS/DecWindows ; Re: different behavior of a C++ binary in different machine * Re: DiskXcel vs OpenVMS 7.3 read cacheing?* Re: DiskXcel vs OpenVMS 7.3 read cacheing?4 RE: Excellent Commerzbank OpenVMS public testimonial( Re: How to access Oracle with DEC PascalA Re: How to create a shareable image from an object library (long) < HowTo delete a remote file via FTP without user interaction?@ Re: HowTo delete a remote file via FTP without user interaction?@ Re: HowTo delete a remote file via FTP without user interaction?@ Re: HowTo delete a remote file via FTP without user interaction?@ RE: HowTo delete a remote file via FTP without user interaction?@ Re: HowTo delete a remote file via FTP without user interaction?> Re: HSV snapshots vs. Spiralog ? (Was: CLI access to HSV's...)> Re: HSV snapshots vs. Spiralog ? (Was: CLI access to HSV's...) Re: Intel to smash AMD Re: Intel to smash AMD Re: Intel to smash AMD+ Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... + Re: Itanic2 - the cHumPaq spin continues... * Re: Need 80' Fortran/VAX training material* Re: Need 80' Fortran/VAX training material+ Re: Only 20% drop in VMS systems (was: wow) + Re: Only 20% drop in VMS systems (was: wow) + Re: Only 20% drop in VMS systems (was: wow)  OpenVMS Jobs in Europe Re: OpenVMS Jobs in Europe5 Re: PW600au problem (probably h/w) - long and tedious   Re: Questions about license paks  Re: Questions about license paks, Re: Raid Array 7000 firmware update question Re: RDB Questions  Re: Single Column text output? Re: Single Column text output?D Re: Size of the comp.os.vms readership ? , was: RE: OpenVMS webring:D Re: Size of the comp.os.vms readership ? , was: RE: OpenVMS webring:D Re: Size of the comp.os.vms readership ? , was: RE: OpenVMS webring:D Re: Size of the comp.os.vms readership ? , was: RE: OpenVMS webring: Software Tools Clearinghouse? ! Re: Software Tools Clearinghouse?  status tapedrive flashing  Re: status tapedrive flashing  Re: status tapedrive flashing  Re: Terminal input from DCL  Re: Terminal input from DCL I Re: usage of LIB$M_FIS_MIXEDCASE in the lib$find_image_symbol RTL routine  Re: VMS commitment Re: VMS commitment  F ----------------------------------------------------------------------  % Date: Wed, 24 Jul 2002 12:04:42 +0100 U From: Andrew Harrison SUNUK Consultancy <andrew_nospam.harrison_remove_this@sun#.com> - Subject: Re: Another article mentions OpenVMS 0 Message-ID: <ahm1k5$b59$1@new-usenet.uk.sun.com>   Kenneth Farmer wrote:   ! > Why HP could be king of servers ; > http://www.openvms.org/stories.php?story=02/07/23/5745879  >  > -- >  > Kenneth Farmer > http://www.Tru64.org > http://www.OpenVMS.org > http://www.LinuxHPC.org  >  >  >     2 Argggggggggg. So HP is going to be the king of the/ Server market because they have concentrated on * building a chipset for IA-64 that supports, up to 4 CPU's (but does it better than other/ chipsets that support more) and bigger capacity / systems will be delivered by clustering lots of  4 CPU Hintel boxes together.  - Oh                   my                   god   1 Sorry but this a repetition of the MS scaling out 1 dream which has as we all know never happened for  good technical reasons.   . And the example cited was a massively parallel. HPC design win. Great proves the point doesn't it.        Regards  Andrew Harrison    ------------------------------  % Date: Wed, 24 Jul 2002 09:51:00 -0600 + From: "Ransom Fitch" <rlfitch@peakpeak.com> * Subject: BACKUP to remote disk via DECNET?0 Message-ID: <001101c23329$e8c55f40$0a00a8c0@w2k>  F I need to copy files (and directory structure!) to a remote disk (LAN,G non-clustered) via DECNET.  Is there a way to use the BACKUP utility to  do this?  Using VMS 6.2.     Thanks,  Ransom Fitch   ------------------------------  % Date: Wed, 24 Jul 2002 12:39:36 -0400 1 From: "Mark D. Jilson" <jilly@clarityconnect.com> . Subject: Re: BACKUP to remote disk via DECNET?2 Message-ID: <3D3ED848.769A84D3@clarityconnect.com>  G Yes, create a saveset on the remote system and then expand that saveset  there.  
 Local node $ BACKUP/NOASSIST/VERIFY - disk:[dir...]*.*.* -3 node"username password"::disk:[dir]saveset.ext/SAVE   ( After backup is done, on the remote node $ BACKUP/NOASSIST/VERIFY - disk:[dir]saveset.ext/SAVE - disk:[*...]/BY_OWNER=ORIGINAL    Ransom Fitch wrote:  > H > I need to copy files (and directory structure!) to a remote disk (LAN,I > non-clustered) via DECNET.  Is there a way to use the BACKUP utility to  > do this?  Using VMS 6.2. > 	 > Thanks,  > Ransom Fitch   --  C Jilly	- Working from Home in the Chemung River Valley - Waverly, NY 0 	- jilly@clarityconnect.com			- Brett Bodine fan+ 	- Mark.Jilson@hp.com				- since 1975 or so  	- http://www.jilly.baka.com   ------------------------------    Date: 24 Jul 2002 02:41:19 -0700! From: oscar@mroscar.co.uk (Oscar) ' Subject: CWSB -Remote & -Raise Switches < Message-ID: <7c07dbc5.0207240141.d4defc2@posting.google.com>  	 Hi there,   D I have got CSWB up and running on an Alphaserver ds10 (OVMS 7.3) ansD all seems to be well - except that when using the -remote switch (toE open a page in an existing window)it does not bring the window to the  front.  A Under Netscape the default for the -Remote switch is to raise the E window to the front and you could manually set the behaviour with the  -raise and -noraise commands.   F The -Remote is fine it is just the raising which appears to be missing  / Is the feature missing? Unimplemented? Broken?    " Anyone know what I am doing wrong?   Cheers   Oscar    P.S.. I have been emailing 'OpenVMS.CSWB@hp.com' andF 'OpenVMS.CSWB@compaq.com' trying to get help from them (as the companyA I work for has a service contract) but they appear not to want to  reply to me :(   P.P.S.E Anyone else noticed that it is also stupidly so in eXcursion? (except  for me and Mr Blake)   ------------------------------  # Date: Wed, 24 Jul 2002 10:21:12 GMT ' From: Colin Blake <colin@theblakes.com> + Subject: Re: CWSB -Remote & -Raise Switches , Message-ID: <3D3E7F97.1000302@theblakes.com>  H I tried this and got the same results. I've entered a bug to track this:2 http://bugzilla.mozilla.org/show_bug.cgi?id=159092  D For support of the released version of CSWB you should contact your J support center. The OpenVMS CSWB mail account was just for the field test.   Colin.   ------------------------------   Date: 24 Jul 2002 07:58 CDT ' From: carl@gerg.tamu.edu (Carl Perkins) . Subject: Re: Cygwin/xfree86 and VMS/DecWindows- Message-ID: <24JUL200207584233@gerg.tamu.edu>   ' tadamsmar@aol.com (Tom Adams) writes... F }Anyone have any ideas why I am getting a "can't open display message"; }when I try to run a client on the VMS system?  (See above)   	 Security?   E Have you made certain that the system doing the displaying (X server) A is set to allow the system running the program (X client) to have  access to the display?   --- Carl   ------------------------------  % Date: Wed, 24 Jul 2002 07:21:03 -0400 2 From: rdeininger@mindspring.com (Robert Deininger)D Subject: Re: different behavior of a C++ binary in different machineJ Message-ID: <rdeininger-2407020721030001@1cust30.tnt1.nashua.nh.da.uu.net>  5 In article <3D3E1428.45324C1E@videotron.ca>, JF Mezei % <jfmezei.spamnot@videotron.ca> wrote:   O >Wasn't there some fairly major change with EV6 which required some programs to O >be recompiled with a newer compiler otherwise there would be problems with the H >executable on the newer platform ? (something about alignment I think).  H The problem was that a number of compilers had been generating code thatE was not compliant with the Alpha SRM (System Reference Manual), which G defines the alpha architecture.  Compilers were occasionally generating I load-locked/ store-conditional sequences that violated one of the rules.  J None of the pre-EV6 alpha chips cared; the code worded anyway.  On EV6 andC later the rule mattered, and these incorrect code sequences stopped  working.  H The SRM_CHECK tool still ships with VMS (in SYS$SYSTEM IIRC), and can beE used to look for bad code sequences in executables.  Be aware that it E gives false positives sometimes -- if it says a program is broken, it G might really be ok.  If it says it is ok, it is ok.  See any recent VMS # Release Notes manual for more info.    ------------------------------  % Date: Wed, 24 Jul 2002 10:31:02 +0100 ( From: Nic Clews <sendspamhere@127.0.0.1>3 Subject: Re: DiskXcel vs OpenVMS 7.3 read cacheing? ) Message-ID: <3D3E73D6.3B4C75CD@127.0.0.1>    Keith Brown wrote: > D > I would like to solicit opinions  on which read caching product isK > preferred by this group for OpenVMS Alpha systems. I have an I/O bottle I J > am trying to cure (read performance). I'm currently running V7.2-1 on anI > ES40 with 4GB ram and dual HSZ70's with 64MB writeback cache.  Should I L > upgrade to V7.3-1 and hope that helps or purchase a 3rd party product like9 > DiskXcel from Networking Dynamics, I/O Express, others?   G IO bottlenecks can be addressed with more than just caching. If you are G on RAID5 then you're probably optimizing the read performance. Have you B considered turning the writeback cache into a read cache only, and increasing its size?  H I presume you files are nicely defragmented, if indexed, are the indicesA tuned? Using global buffers? are the ACP_* caches at optimum size H (AUTOGEN and FEEDBACK), are the MOUNT qualifiers appropriate, are windowE turns occurring due to too small a cluster size, or fragmentation, or H too small windows? Are the files distributed to avoid volume contention?6 Later versions of VMS allow primary CPU IO offloading.  F Don't forget your application. I had one case of a database, where theH working set was too small, it was trying to cache data into memory, only- to be paged out [ultimately to the pagefile]!   0 The performance manual is a good place to start.  F Most solutions will probably work, but are you treating the problem or fixing the symptoms? --  ? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------  % Date: Wed, 24 Jul 2002 12:02:58 +0200 9 From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com> 3 Subject: Re: DiskXcel vs OpenVMS 7.3 read cacheing? ' Message-ID: <3D3E7B52.B1D11642@aaa.com>   : If your files are only/most read-only, DECram ("ram-disk")7 could olso be an option. If I'm not wrong, it should be 8 possible to shadow a DECram disk with a standard disk so> all writes will ba "saved", but reads comes from the ram-disk.  = But if you have 80-100 600Mb files, that is 48-60 Gb of data. > If these files are read sequentional, the caches will probably> not help anyway. The caches will just be flushed over and over> by "new" data. It might even go slower because of the overhead of managing large caches.   ? You don't say how your disks are configured. Make sure that you > get the best possible read I/O speed from your disks. Striping? comes to mind. And any cache product that implements read-ahead  caching would probably be good.   9 But, it very mauch depands of how your files are acessed. 4 One at a time ? Many in parallell ?  Seq or random ?> Do you also write some "output-file" at the same time ? If so,= make sure that is on different spindels as the "input-files". @ And try to read one file at a time from each spindel to minimize the disk-head movments.    Jan-Erik Sderholm.   L > In article <KGn%8.621829$352.132778@sccrnsc02>, kbrown2720@NOSPAMattbi.com > wrote: > 
 > >Hi all, > > E > >I would like to solicit opinions  on which read caching product is L > >preferred by this group for OpenVMS Alpha systems. I have an I/O bottle IK > >am trying to cure (read performance). I'm currently running V7.2-1 on an J > >ES40 with 4GB ram and dual HSZ70's with 64MB writeback cache.  Should IM > >upgrade to V7.3-1 and hope that helps or purchase a 3rd party product like : > >DiskXcel from Networking Dynamics, I/O Express, others? > > L > >Is it true that 7.3 still does file caching like 6.1 or 6.2 or does it doK > >block based caching? Any experience on how much improvement I can expect N > >from 7.3. How large can the cached files be? The data files I need to cacheL > >are 600MB in size and about will range from 80 to 100 in number. Some are  > >obviously hotter than others.   ------------------------------  % Date: Wed, 24 Jul 2002 12:26:11 -0400 * From: WILLIAM WEBB <WWEBB1@email.usps.gov>= Subject: RE: Excellent Commerzbank OpenVMS public testimonial - Message-ID: <0033000073796240000002L002*@MHS>   
 =0AWarren-   Just for the record-   I wasn't criticizing *you*.   + I was just making a subtly sarcastic remark , that unless it's an app grabbing data from a, database and creating DHTML with it, updates/ to a website (just like everywhere else) aren't  *instantaneous*.  * Perhaps *too* subtle. (should I use tags?)   Regards,   WWWebb   -----Original Message-----/ From: Info-VAX-Request@Mvb.Saic.Com at INTERNET % Sent: Tuesday, July 23, 2002 10:26 PM B To: Webb, William W Raleigh, NC; Info-VAX@Mvb.Saic.Com at INTERNET= Subject: RE: Excellent Commerzbank OpenVMS public testimonial     H I'm still waiting to get the official version from the vendor.. Then I = get & to make it into html and pretty it up.   -warren    --B ------------------------------------------------------------------9 Warren Sander                           OpenVMS Marketing B Hewlett-Packard Company         Work:  warren.sander@remove.hp.comH 200 Forest Street MR01-3/J1     Personal: sander@remove.ma.ultranet.com=  3 Marlboro, MA 01752                   (508) 467-4875 5    My opinions are my own and I only speak for myself ,          Read http://www.openvms.compaq.com/B ------------------------------------------------------------------7 "WILLIAM WEBB" <WWEBB1@email.usps.gov> wrote in message ' news:0033000073084675000002L052*@MHS...    Ed-    Warren's not *that* fast.   - The article to which you refer is dated 1999.   & They seem to have upgraded since then-  ( This one is dated July 2002 and entitled   "hp AlphaServer technology helps Commerzbank tolerate disaster on 9/11"   3 OpenVMS and GS160s are mentioned on the front page.   0 "Because of the intense heat in our data center,7 all systems crashed except for our AlphaServer GS160... 1 OpenVMS wide-area clustering and volume-shadowing 2 technology kept our primary system running off the) drives at our remote site 30 miles away."   / It's everything you'd want to see in an OpenVMS  promotional piece.  ' Kudos to whoever put this one together.    (Hurry up, Warren!)    WWWebb   -----Original Message-----/ From: Info-VAX-Request@Mvb.Saic.Com at INTERNET % Sent: Thursday, July 18, 2002 4:10 PM B To: Webb, William W Raleigh, NC; Info-VAX@Mvb.Saic.Com at INTERNET= Subject: RE: Excellent Commerzbank OpenVMS public testimonial     E Interestingly enough the Compaq Success Stories web site reports that 4 Commerzbank is using Tru64 for its database servers.   Full story is at: J <http://www.success-stories.compaq.com/cgi-bin/cssextusr/s=3Ddisplay/i=3D= 438>  # Here's an excerpt from the article.  What makes it work: 	 Products: ? Two AlphaServerTM 8400 systems running Tru64TM UNIX with Compaq ' StorageWorksTM, and an Oracle8 database 2 An AlphaServerTM 4100 as a standby recovery system   EdE **Please apply a generous amount of all the usual disclaimers here.**      > -----Original Message-----6 > From: rob@netcarrier.net [mailto:rob@netcarrier.net]' > Sent: Thursday, July 18, 2002 2:51 PMt > To: Info-VAX@Mvb.Saic.Como? > Subject: Re: Excellent Commerzbank OpenVMS public testimonialo >d >  > 7 > "Bob Ceculski" <bob@instantwhip.com> wrote in messagee9 > news:d7791aa1.0207181124.4bb916b4@posting.google.com... D > > "Sue Skonetski" <susan.skonetski@hp.nospam.com> wrote in message( > news:<ah6i8h$51l$1@web1.cup.hp.com>... > > > Dear Newsgroup,e > > >e >?E > > hear that Andrew, all with 4 alphas with a few chips each and thesF > > best clustering around ... with any other platform the above wouldE > > be impossible ... that's why I always say, an app is only as goodnD > > as the os it runs on ... if your os is down, so is your app, andC > > so are you, and so is your profits, and so is your stock price!o >  > = >      mmmmm Really seems to have helped Digital and Compaq'aE > stock  ......... >  >=3D=a   ------------------------------  + Date: Wed, 24 Jul 2002 08:05:52 -0600 (MDT)e" From: John Nebel <nebel@csdco.com>1 Subject: Re: How to access Oracle with DEC Pascal F Message-ID: <Pine.OSF.4.21.0207240757580.3788-100000@athena.csdco.com>  9 On Tue, 23 Jul 2002, Christian [iso-8859-1] M=FCck wrote:t   > Hello all, >=20F > I would like to know how to access an Oracle-DB (not Rdb) from a DECJ > Pascal application. I did not find any information about precompilers or4 > so on the net. Is there a way to use Rdb Sqlmod's?7 > We currently use DEC Pascal V5.5-51 and Oracle 8.1.7.  >=20 > Any suggestions ?s >=20 > Thanks > ChrisH >=20 >=20   Chris,  I Any VMS language (assuming it generates standard calls) should be able topF use the Oracle Call Interface.  It was a bit of a pain getting startedH because all the examples are in C which I know very little.  After that, Fortran worked fine.  3 Having a hard copy of the OCI manual really helped.V  
 John Nebel   ------------------------------  % Date: Wed, 24 Jul 2002 07:22:09 -0700S, From: "James Gessling" <jgessling@yahoo.com>J Subject: Re: How to create a shareable image from an object library (long)4 Message-ID: <ahmd6j$tsk9e$1@ID-46415.news.dfncis.de>  4 "David Mathog" <mathog@caltech.edu> wrote in message% news:3D3DBCB5.33CEE1E8@caltech.edu...b > James Gessling wrote:d >rB > > The trick to making the shareable image is to create an option9 > > file with all the entry points.  This is pretty easy.o >h > Well, usually.  See: >,H >   ftp://saf.bio.caltech.edu/pub/software/openvms/make_opt_from_olb.com >E= > I've run into cases (for instance, Xforms) where there werewB > global variables in the library which the SYMBOLS program didn't > catch.  L You're right.  This method can't find variables that need to be global, justJ entry points.  In fact it probably exposes too many entry points, it wouldJ be cleaner to go through the options file and remove those that don't need to be seen from the "outside".  K I find the trick of putting one or more object files into an object libraryoL to be able to list the entry point names to be generally useful, rather thanF trying to poke through source code where they may be hidden in various ways, like in header files.m   Jimo  ! p.s. and yes it could be scripted:   ------------------------------    Date: 24 Jul 2002 01:07:38 -0700* From: GMarttin@Hotmail.com (Geert Marttin)E Subject: HowTo delete a remote file via FTP without user interaction?p= Message-ID: <75be4573.0207240007.6257f41f@posting.google.com>p   Hi all,4  F I'm maintaining a (DEC Pascal) program that will download updates on aC regular basis. The program consists of a loop within which it will:e< - check if new files have been created on a remote directory' - copy these files to a local directoryt* - delete the files on the remote directoryE This worked fine with DecNet but in the future we'll be using FTP foro this.P  C The first two were easy to modify utilizing DIR/FTP and COPY/FTP inoF command files which are 'called' from the program (using $CREPRC). ButF I seem unable to find a (DCL) command that will delete files via FTP?!B I am able to set up a FTP connection manually and delete the filesD (with the FTP utility) but, well, that's not something my users want3 to be bothered with/I want to bother my users with.:  0 Anybody have any idea how to automate this task?   TIA...        Geert Marttin       PS:D6   DIGITAL TCP/IP Services for OpenVMS VAX Version V4.20   on a MicroVAX 3100-90 running OpenVMS V7.1    ,   DEC Pascal V5.6-64 for OpenVMS VAX Systems   ------------------------------  % Date: Wed, 24 Jul 2002 10:33:33 +0200.2 From: "Ren Schelbaum" <rene.schelbaum@datakom.at>I Subject: Re: HowTo delete a remote file via FTP without user interaction?oG Message-ID: <3d3e665c$0$33064$91cee783@newsreader02.highway.telekom.at>C  = "Geert Marttin" <GMarttin@Hotmail.com> schrieb im Newsbeitragc7 news:75be4573.0207240007.6257f41f@posting.google.com... 	 > Hi all,h >.H > I'm maintaining a (DEC Pascal) program that will download updates on aE > regular basis. The program consists of a loop within which it will:n> > - check if new files have been created on a remote directory) > - copy these files to a local directory , > - delete the files on the remote directoryG > This worked fine with DecNet but in the future we'll be using FTP forv > this.. >sE > The first two were easy to modify utilizing DIR/FTP and COPY/FTP in H > command files which are 'called' from the program (using $CREPRC). ButH > I seem unable to find a (DCL) command that will delete files via FTP?!D > I am able to set up a FTP connection manually and delete the filesF > (with the FTP utility) but, well, that's not something my users want5 > to be bothered with/I want to bother my users with.j >:2 > Anybody have any idea how to automate this task? >. > TIA... >d >      Geert Marttin >  >i >e > PS: 8 >   DIGITAL TCP/IP Services for OpenVMS VAX Version V4.2. >   on a MicroVAX 3100-90 running OpenVMS V7.1. >   DEC Pascal V5.6-64 for OpenVMS VAX Systems   Hi!s  ? You have to create a FTP-Commandfile and invoke FTP as follows:e  2 FTP hostname /USERNAME=username /PASSWORD=password /INPUT=your_ftp_commandfile   J You can create this commandfile in your applicalion and delete it, soon as the task is completed.   regardst   Ren   ------------------------------  % Date: Wed, 24 Jul 2002 11:22:06 +0200o: From: Karl Rohwedder <extern.karl.rohwedder@volkswagen.de>I Subject: Re: HowTo delete a remote file via FTP without user interaction?@/ Message-ID: <ahlrj4$32g2@doiweb4.volkswagen.de>c   Geert Marttin wrote:	 > Hi all,n > H > I'm maintaining a (DEC Pascal) program that will download updates on aE > regular basis. The program consists of a loop within which it will:M> > - check if new files have been created on a remote directory) > - copy these files to a local directoryn, > - delete the files on the remote directoryG > This worked fine with DecNet but in the future we'll be using FTP for  > this.g > E > The first two were easy to modify utilizing DIR/FTP and COPY/FTP iniH > command files which are 'called' from the program (using $CREPRC). ButH > I seem unable to find a (DCL) command that will delete files via FTP?!D > I am able to set up a FTP connection manually and delete the filesF > (with the FTP utility) but, well, that's not something my users want5 > to be bothered with/I want to bother my users with.  > 2 > Anybody have any idea how to automate this task? >  > TIA... >  >      Geert Marttin >  >  >  > PS:w8 >   DIGITAL TCP/IP Services for OpenVMS VAX Version V4.22 >   on a MicroVAX 3100-90 running OpenVMS V7.1    . >   DEC Pascal V5.6-64 for OpenVMS VAX Systems  & - maybe you can use DECnet/Osi over IP  - use RSH for the delete command -- o  - mit freundlichen Gruessen | with best regards    Karl Rohwedder               -C iT-Ingenieurteam     | Ellernbruch 11       | D-38112 Braunschweig rA Telefon: 0531/515521 | Telefax: 0531/515531 | Mobil: 0172/5434843 H  E-Mail: rohwedder(at)decus.decus.de        | iT-IngTeam(at)t-online.de .          karl.rohwedder(at)it-ingenieurteam.de DATEX-P: 4505018005::ROHWEDDER   ------------------------------  % Date: Wed, 24 Jul 2002 12:17:24 +0200o9 From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com>sI Subject: Re: HowTo delete a remote file via FTP without user interaction?s' Message-ID: <3D3E7EB4.56849388@aaa.com>f   A few thoughts...t  8 Could the "remote" application who creates the files not4 copy them to a "local" directory for you ? Then your9 own application would just be handling "local" files. Andi6 you don't need to have any "polling" over the network.   Or,t@ Breakout the FTP-hnadling from your Pascal app and build it with@ DCL. It's probably easier to manage that way. Or put in in a COM( file that is called from the Pascal app.   Or,,! Do the delete with a RECEX call :   4  $ RSH host /user=xxx /password=yyy DELETE some-file  = you can also add /LOG=zzz to catch any output from the DELETE  command.   Jan-Erik Sderholm.r   Geert Marttin wrote: > 	 > Hi all,i > H > I'm maintaining a (DEC Pascal) program that will download updates on aE > regular basis. The program consists of a loop within which it will:n> > - check if new files have been created on a remote directory) > - copy these files to a local directoryh, > - delete the files on the remote directoryG > This worked fine with DecNet but in the future we'll be using FTP fort > this.r >:   ------------------------------  % Date: Wed, 24 Jul 2002 09:36:02 -0400 & From: "DAniel Allen" <dallen@nist.gov>I Subject: RE: HowTo delete a remote file via FTP without user interaction?): Message-ID: <BCEGLBGJDODLELBJIADKEEBLCAAA.dallen@nist.gov>  < 	FTP is a pretty simple protocol.  I wouldn't thing actually> 	coding a custom function to perform the delete function would> 	be a very difficult task in Pascal or any other language e.g.> 	connect and USER, PASS, DELE, ... I have a vague recollectionA       of some discussion of an FTP API in conjunction with one orrD       more of the VMS TCP/IP products but that may have been wishful       dreaming...r   	Dan   -----Original Message-----/ From: Jan-Erik S=F6derholm [mailto:aaa@aaa.com]d& Sent: Wednesday, July 24, 2002 6:17 AM To: Info-VAX@Mvb.Saic.Coms< Subject: Re: HowTo delete a remote file via FTP without user interaction?     A few thoughts...   8 Could the "remote" application who creates the files not4 copy them to a "local" directory for you ? Then your9 own application would just be handling "local" files. Ando6 you don't need to have any "polling" over the network.   Or,t@ Breakout the FTP-hnadling from your Pascal app and build it with@ DCL. It's probably easier to manage that way. Or put in in a COM( file that is called from the Pascal app.   Or,t! Do the delete with a RECEX call :b  8  $ RSH host /user=3Dxxx /password=3Dyyy DELETE some-file  ? you can also add /LOG=3Dzzz to catch any output from the DELETEt command.   Jan-Erik S=F6derholm.    Geert Marttin wrote: >-	 > Hi all,a > H > I'm maintaining a (DEC Pascal) program that will download updates on aE > regular basis. The program consists of a loop within which it will: > > - check if new files have been created on a remote directory) > - copy these files to a local directoryo, > - delete the files on the remote directoryG > This worked fine with DecNet but in the future we'll be using FTP forr > this.  >"   ------------------------------  % Date: Wed, 24 Jul 2002 11:18:15 -0400f- From: JF Mezei <jfmezei.spamnot@videotron.ca>sI Subject: Re: HowTo delete a remote file via FTP without user interaction?e, Message-ID: <3D3EC534.4EF2FADC@videotron.ca>   "Ren Schelbaum" wrote:wA > You have to create a FTP-Commandfile and invoke FTP as follows:  > 4 > FTP hostname /USERNAME=username /PASSWORD=password > /INPUT=your_ftp_commandfilee    M Or create the subprocess with a mailbox device as input. That would allow therM main process to send whatever commands to the subprocess through the mailbox.oN Initially, you'd be at the equivalent of the $ sign, so you would issue a "FTPN host etc" command, and then you would be at the FTP> prompt and you could thenI issue your COPY DELET etc commands at will. When done, you would issue anaB "EXIT" command and then a "LOGOUT" command to kill the subprocess.  L The beauty of this is that not only is the subprocess preserved, but you canI also preserve your FTP session to the remote host and just have your mainlJ process issue FTP commands  with the subprocess remaining connected to theM remote host. Much more efficient than creating multiple subprocesses for eachpK FTP command (and especially since each COPY/FTP etc will establish a ttallya  new session to the remote host).  J You could also create the suprocess with its output to another mailbox andG then have your main program read from the mailbox to confirm success or?N failure, but that is a bit more involved since you'd have to loop a read untilM you get to some string indicating that FTP has completed the command and sentc all its output.H  L Before creating the process, you therefore need to create the mailboxes. andE supply those as arguments for subprocess input and optionally output.o   ------------------------------  # Date: Wed, 24 Jul 2002 09:07:40 GMT2* From: "Bill Todd" <billtodd@metrocast.net>G Subject: Re: HSV snapshots vs. Spiralog ? (Was: CLI access to HSV's...)bC Message-ID: <w7u%8.295995$Im2.15263517@bin2.nnrp.aus1.giganews.com>0  9 "Paul Repacholi" <prep@prep.synonet.com> wrote in message*' news:87r8hutliz.fsf@prep.synonet.com...cC > Jim.Johnson@software-exploration.nospam.com (Jim Johnson) writes:c >tG > > Unless you are saying that HSV disks become irretrievably corrupted'H > > in the event of a system crash (something that I cannot bring myselfA > > to believe), any and all caches held on any and all hosts aret > > irrelevant.I >R) > May... What is fsck used for? and when?   L It is used to attempt to bring incompetently-designed file systems back to aI state of internal consistency.  Competent file systems use careful-update J approaches or protect themselves with a transaction log, both avoiding theJ need for fsck and providing actual consistency up to some defined point inK time rather than just whatever structural consistency they can scrounge outp of the mess.  D As to 'when', it's used during recovery.  In the case Jim described,K 'recovery' includes using it (for incompetent file systems that require it)aH as the first step toward actually using a snapshot (though this requiresI that the snapshot be writable or be restored in raw form to a system thatoG can then be updated by running fsck).  Log-protected file systems wouldWH instead process their transaction log as the first step toward using theJ snapshot (again requiring that the snapshot be writable or be restored rawF to a system that could then be updated by processing the log).  ODS-2,L however, being a careful-update file system, could just process the snapshotG directly without having to write to it to create a consistent (at leaste, sufficiently consistent to be usable) state.   >nH > > The consistent view that I was referring to, and the only consistentC > > view that matters if the previous paragraph is true, is the oneeB > > presented by the storage itself -- that certainly involves theB > > caches on the disk device, and probably involves caches on theF > > controller.  It does NOT involve caches at any higher layer of the > > system., >rD > The consistant view may only exist on the user side of the kernel.  J The state Jim refers to is called a 'crash-consistent' state.  If a systemJ can recover from a crash, then it can use the same mechanisms to convert aL crash-consistent snapshot into something that can be processed (as described above).s    That J > is the state you must preserve. There must be a mechinism that runs fromE > userland to the metal to flush out ALL state, and totally serializec
 > everything.s  H No, there needn't be:  as long as the application is built to be able toD recover to a usable state after a crash, it can use exactly the sameJ mechanisms to recover a device-level snapshot to the same usable state (asK long as all application state is contained on the one device performing thehG snapshot; if it's spread more widely, some kind of synchronous snapshoto7 coordination across all applicable devices must exist).r  L Now, if you have a mechanism to flush all state to the storage device beforeH it takes its snapshot, you can certainly acquire a more up-to-the-minuteC picture.  But it's arguable that the difference is not particularly4H significant, since by definition the only additional state you'd captureD would be state the application was willing to lose if the system (orG application) died (i.e., not that important:  if it were, it would havea flushed it already).  ;  Note, that in a SAN, this means all the SAN is potentiallyo > involved in this.  >eD > If some of your data is cached in the host when you perform a snapD > and it does not get out first, you have corruped your data to some	 > extent.   L Only to the same extent that the application was willing to tolerate (and/or2 deal with on restart) if it or the system crashed.   - bill   ------------------------------  + Date: Wed, 24 Jul 2002 15:36:26 +0000 (UTC)5 From: david20@alpha2.mdx.ac.ukG Subject: Re: HSV snapshots vs. Spiralog ? (Was: CLI access to HSV's...)e+ Message-ID: <ahmhhq$8lq$1@aquila.mdx.ac.uk>0  p In article <w7u%8.295995$Im2.15263517@bin2.nnrp.aus1.giganews.com>, "Bill Todd" <billtodd@metrocast.net> writes: >i: >"Paul Repacholi" <prep@prep.synonet.com> wrote in message( >news:87r8hutliz.fsf@prep.synonet.com...D >> Jim.Johnson@software-exploration.nospam.com (Jim Johnson) writes: >>H >> > Unless you are saying that HSV disks become irretrievably corruptedI >> > in the event of a system crash (something that I cannot bring myself B >> > to believe), any and all caches held on any and all hosts are >> > irrelevant. >>* >> May... What is fsck used for? and when? >gM >It is used to attempt to bring incompetently-designed file systems back to aaJ >state of internal consistency.  Competent file systems use careful-updateK >approaches or protect themselves with a transaction log, both avoiding the2K >need for fsck and providing actual consistency up to some defined point inHL >time rather than just whatever structural consistency they can scrounge out
 >of the mess.4 >5E >As to 'when', it's used during recovery.  In the case Jim described,eL >'recovery' includes using it (for incompetent file systems that require it)I >as the first step toward actually using a snapshot (though this requires J >that the snapshot be writable or be restored in raw form to a system thatH >can then be updated by running fsck).  Log-protected file systems wouldI >instead process their transaction log as the first step toward using thebK >snapshot (again requiring that the snapshot be writable or be restored raw'G >to a system that could then be updated by processing the log).  ODS-2, M >however, being a careful-update file system, could just process the snapshotcH >directly without having to write to it to create a consistent (at least- >sufficiently consistent to be usable) state.k >  >>I >> > The consistent view that I was referring to, and the only consistenteD >> > view that matters if the previous paragraph is true, is the oneC >> > presented by the storage itself -- that certainly involves theiC >> > caches on the disk device, and probably involves caches on the-G >> > controller.  It does NOT involve caches at any higher layer of ther >> > system. >>E >> The consistant view may only exist on the user side of the kernel.r >tK >The state Jim refers to is called a 'crash-consistent' state.  If a system K >can recover from a crash, then it can use the same mechanisms to convert a M >crash-consistent snapshot into something that can be processed (as describedh >above). >l > ThatK >> is the state you must preserve. There must be a mechinism that runs from F >> userland to the metal to flush out ALL state, and totally serialize >> everything. >wI >No, there needn't be:  as long as the application is built to be able tonE >recover to a usable state after a crash, it can use exactly the samerK >mechanisms to recover a device-level snapshot to the same usable state (aseL >long as all application state is contained on the one device performing theH >snapshot; if it's spread more widely, some kind of synchronous snapshot8 >coordination across all applicable devices must exist). > M >Now, if you have a mechanism to flush all state to the storage device beforetI >it takes its snapshot, you can certainly acquire a more up-to-the-minutenD >picture.  But it's arguable that the difference is not particularlyI >significant, since by definition the only additional state you'd capture E >would be state the application was willing to lose if the system (or>H >application) died (i.e., not that important:  if it were, it would have >flushed it already).5 >$< > Note, that in a SAN, this means all the SAN is potentially >> involved in this. >>E >> If some of your data is cached in the host when you perform a snapoE >> and it does not get out first, you have corruped your data to somet
 >> extent. >aM >Only to the same extent that the application was willing to tolerate (and/or 3 >deal with on restart) if it or the system crashed.  >o >- billo  M Unfortunately fsck can sometimes not fix everything automatically on restart.eN You are then forced to reboot into single user mode and use fsck interactively3 to attempt the recovery. Not a pleasant experience.f  K A log-protected file system overcomes these problems but not everybody usesh6 such a file system. Use of UFS is still fairly common.  1 Another post mentioned ORACLE and online backups.   > From the "ORACLE Backup & Recovery handbook" from Oracle Press4 Database Design and Basic backup Rules (pages 61,62)   " + 1. Operate the database in ARCHIVELOG mode.o  D 2. Perform offline backups at least once a week if you don't need toJ operate your database 24 hours a day, 7 days a week. However, if your shop/ is operational 24x7, take daily online backups.  "n  M There are probably many Oracle sites which either supplement their online hotSD backups with offline cold backups or even only perform cold backups.    N Secondly the Oracle recommendations for how databases are setup has for years L specified that redo logs should be placed on a separate disk(s) to database I files.  Hence taking snapshots of the Oracle disks (with a running Oracle-K database) is NOT equivalent to the position after a system crash. Since therK timing of the snapshots may not capture the database files and redo logs onl% separate disks in a consistent state.eJ I am not aware of any controller based synchronisation feature which would< allow you to synchronise taking snapshots of multiple disks.    O Hence the need to shutdown the database and take the snapshots (if doing a coldiM backup) from the OS. Since by their nature such backups have to be taken whenoO users are not using the system they tend to occur late at night or early in they> morning. Hence the need for a host based scripting capability.     
 David Webb VMS and Unix team leader CCSS Middlesex University I   ------------------------------  % Date: Wed, 24 Jul 2002 10:29:48 +0200 = From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>h Subject: Re: Intel to smash AMD ) Message-ID: <3D3E657C.113AF45E@gtech.com>l   JF Mezei wrote:lP > And every action that Intel takes to make its 8086 more competitive with AMD's& > offering will hurt IA64 just as bad.   Why ?   > A redunction from $597 to $220 cuts $377 of the price. This isF a huge reduction for a low-end & mid-size PC's. But it does not really9 count that much for high-end PC's and especially servers.9   Arne   ------------------------------  % Date: Wed, 24 Jul 2002 08:35:27 -0700b' From: David Mathog <mathog@caltech.edu>S Subject: Re: Intel to smash AMD + Message-ID: <3D3EC93F.AC016E0E@caltech.edu>    Rob Young wrote: >         With theL >         upcoming 63% slice in Pentium 4 2.53 GHz prices (currently $597, aE >         63% price cut brings the 2.53 GHz part down to about $220).d  A That's the biggest single step price cut I've ever seen on a CPU.r> I can believe Intel won't be making any money on it.  I'm also? wondering if it isn't actually below the true production price.:A Dumping is the traditional method for wiping out competitors withc* fewer resources.  That's why it's illegal.  = Pretty obvious that Intel's strategy is to literally kill AMD3? in a brutal price war before AMD can ship Hammer and kill InteliD in fair competition.  Ie, Intel has more money in the bank than AMD,A so they'll sell their product at no profit (or even a loss) until > AMD dies.  Look for AMD to either accelerate Hammer or partnerH up with somebody big (Via?) to ride out the assualt.  Don't be surprisedA when the FTC jumps all over Intel.  Do be surprised if they do ite$ quick enough to make any difference.   Regards,   David Mathog mathog@caltech.edu   ------------------------------    Date: 24 Jul 2002 11:31:32 -0600+ From: young_r@encompasserve.org (Rob Young)  Subject: Re: Intel to smash AMDr3 Message-ID: <hKLt8cFK9ejE@eisner.encompasserve.org>a  U In article <3D3EC93F.AC016E0E@caltech.edu>, David Mathog <mathog@caltech.edu> writes:o > Rob Young wrote: >>         With theiM >>         upcoming 63% slice in Pentium 4 2.53 GHz prices (currently $597, atF >>         63% price cut brings the 2.53 GHz part down to about $220). > C > That's the biggest single step price cut I've ever seen on a CPU.T@ > I can believe Intel won't be making any money on it.  I'm alsoA > wondering if it isn't actually below the true production price.oC > Dumping is the traditional method for wiping out competitors withr, > fewer resources.  That's why it's illegal. >   G 	True in several aspects.  It isn't dumping though.  I highly doubt it cH 	is cheaper than manufacturing cost.  As Linley "Merced" Gwennap states:  5 http://news.com.com/2100-1001-246894.html?legacy=cnet.  P "The larger size means the Pentium 4 will cost around $80 to $90 to manufacture,N more than double the $40 manufacturing cost of the Pentium III, he estimated."  @ 	That analysis referred to .18 micron, we can assume .13 Pentium> 	4s (current crop) would cost somewhere in the neighborhood ofB 	$40-$60 to manufacture depending on yield.  As Linley points out,B 	manufacturing costs are some of the most closely guarded secrets	: 	in the industry.  That said, reasonable estimates abound.  D 	Intel has plenty of headroom left to squeeze AMD further.  Besides,D 	the top end part always costs more as folks buy latest and greatest? 	and they make the fat off the power users and compete with AMDh 	with volume parts.r  B 	Since 2.53 won't be top-end, it drops down to volume status.  TheG 	problem is AMD can't keep up performance and it is starting to show upc
 	big time.  ? > Pretty obvious that Intel's strategy is to literally kill AMDHA > in a brutal price war before AMD can ship Hammer and kill IntelaF > in fair competition.  Ie, Intel has more money in the bank than AMD,C > so they'll sell their product at no profit (or even a loss) untiln@ > AMD dies.  Look for AMD to either accelerate Hammer or partnerJ > up with somebody big (Via?) to ride out the assualt.  Don't be surprisedC > when the FTC jumps all over Intel.  Do be surprised if they do ite& > quick enough to make any difference.  = 	It is a brilliant strategy.  It really is about business andN! 	marketing and what not.  As for:   7 > Don't be surprised when the FTC jumps all over Intel.   D 	That is truly hilarious.  As if the FTC has a case.  They certainlyD 	don't.  Just because AMD's prices are too high, isn't Intel's fault 	;-).i  F 	Accelerate Hammer?  Haven't you heard yet?  It has slipped to Q1 2003 	intro, not Q4 2002.   Ooops!a   				Robm   ------------------------------  % Date: Wed, 24 Jul 2002 11:05:35 +0100 U From: Andrew Harrison SUNUK Consultancy <andrew_nospam.harrison_remove_this@sun#.com> 4 Subject: Re: Itanic2 - the cHumPaq spin continues...0 Message-ID: <ahlu59$9v9$1@new-usenet.uk.sun.com>   loopnz@hotmail.com wrote:a  w > "Bill Todd" <billtodd@metrocast.net> wrote in message news:<LPt_8.198334$iB1.10663413@bin4.nnrp.aus1.giganews.com>...T >  > D >>But we do glean the new nugget of information on page 29 that "TheI >>performance of the Itanium 2 processor's IA-32 engine is expected to beiM >>comparable with a 300 MHz Pentium Pro."  Wow - that'll sure cut Hammer IA32sM >>performance down to size!  Especially since if one runs a code mix heavy onhJ >>the IA32 side the IA32 code will consume most of the available processorE >>time even though only running at about 15% of native performance...  >> > H > Why again are we VMS users concerned about Hammer? Oh right, Elvis wasD > sighted in one of the VMS labs doing a secret VMS port to Hammer -F > after he lost his job in accounting trying to cook the books to showH > that VMS was making billions of dollars in profits and that Compaq x86+ > servers were dragging the company down...v >     " You should be terrified of Hammer.  E If IA-64 fails it will be for a number of reasons, price, performancetF availability, ISV support etc. ISV support and platform vendor support will be particularly critical.  C Getting ISV's who currently support IA-32 to do a full on migration- to IA-64 is a must.   ? Hammer represents a threat to IA-64 because it does not requires? the IA-32 ISV's to do a port if they don't need 64 bit support,s; while IA-64 does because its performance running IA-32 appsn emulated is terrible.   ? Becoming the defacto 64 bit processor for the vendors currently8> building Wintel systems is also a must, HP alone is not enoughE and any thoughts for example that HP-Compaqs Wintel revenues combined C will be enough are also missplaced. The latest market share numbers < which now include the combined HP-Compaq Intel business show> that nearly all the market share gains of amalgamating the two: vendors Intel divisions have been lost with HP losing 5-6%= market share mostly to Dell and posting a revenue drop 2x thes next worst performer.f  = Hammer represents a threat here because platform vendors will > be able to build a much bigger range of systems (in the Wintel8 market space) based on the same x86-64 ISA. AMD have low: cost low power consumption CPU's as well as hot fast CPU's5 under development for availability a long time beforee& Intel expect to do the same for IA-64.    : Like it or not the future of OpenVMS is IA-64, you are now; at the mercy of the Wintel Market. If IA-64 succeeds in thef7 Wintel market then OpenVMS has a long term platform ande4 a future (assuming HP do the right things). If IA-646 fails and Hammer may well be its undoing and Intel are2 forced to base their 64 bit ISA around x86-64 then4 you will not have a platform for OpenVMS and it will die.  3 So you should be very very interested in how Hammere5 does. Perhaps you have noticed that the OpenVMS choire4 who switched from Alpha forever to the big new thing4 of IA-64 overnight and seemingly without any remorse- have now developed a flaming Hammer sideline.e   Regardsr Andrew Harrison    ------------------------------  # Date: Wed, 24 Jul 2002 09:56:48 GMTo* From: "Bill Todd" <billtodd@metrocast.net>4 Subject: Re: Itanic2 - the cHumPaq spin continues...C Message-ID: <ARu%8.296638$Im2.15278341@bin2.nnrp.aus1.giganews.com>l  < "Brannon Batson" <Brannon_Batson@yahoo.com> wrote in message7 news:4495ef1f.0207231147.5bffe808@posting.google.com... 7 > "Bill Todd" <billtodd@metrocast.net> wrote in messagey? news:<%Q6%8.279257$Im2.14362024@bin2.nnrp.aus1.giganews.com>... 
 > > [snip]C > > Now, it's your turn (and boy, do I have some good questions)...I > >y
 > > - bill >e2 > Shoot away.  We'll see how many I can answer :-)   Since you insist:o  H 1.  As soon as the EV8 effort was killed, revisionists started assertingA that 2004 was probably an aggressive date that would have slippedhI significantly and that projected performance was likely over-blown.  YouroD friend Fred has even gone so far recently as to call EV8 'mythical'.  L My own impression was that the 2004 date was if anything a bit conservative,L given the amount EV8 had already slipped (in part because EV7 had slipped soH much - perhaps both largely due to neglect by Compaq, but that's not the@ point of *this* question), and that performance projections wereL well-grounded in reality.  But that's only an impression (just as I stronglyG suspect the revisionists are offering nothing but impressions, possiblyt
 biased ones).c  D So as an intimately-involved participant in this project that is nowK presumably of only historical interest, do you feel comfortable venturing alL more qualified opinion on EV8's schedule and capabilities than has been seen here?e  A 2.  EV8 with SMT was projected to offer twice the performance (intK multi-threaded server use) of EV8 without SMT (but presumably using all itsaI execution units in the service of a single thread), which given 4-way SMT K and EV8's goodly number of execution units seems believable (though if not, I I hope you cover it in answering the previous question).  But I never sawaK anything like an estimate of EV8's single-threaded performance (except fromiH Paul DeMone):  care to provide one?  (Again, it would seem to be ancientB history now and certainly not of any embarrassment to your current8 employer - and your former employer no longer exists...)  H 3.  My impression is that the primary motivation for EPIC was the beliefK that RISC OoO could not be made to work well, and that Alpha disproved this2E fairly decisively - thus removing any technical reason to pursue EPIC>I instead of RISC (especially since so many of McKinley's 'features' - bothnK current and proposed - appear equally applicable to a RISC processor shouldoI anyone be inclined to do so, and since McKinley requires so much more diekG area, power, fast on-chip cache, compiler intelligence, and developmentmK massaging - via profile feedback - to achieve the same level of performance K in the same process as, say, the 1.2+ GHz EV68C will in a month or so if HPoF doesn't delay its release to avoid making McKinley look bad).  WithoutL divulging proprietary secrets, could you suggest any (technical) reasons whyH EPIC might still be worth pursuing (other than, of course, the fact that@ Intel's and HP's courses have been set and the boats burned...)?  I 4.  Given that it's made by a company you don't and AFAIK have not workedtK for, would you feel comfortable giving your professional (but of course not'G Intel-supported) impression of Hammer in the context of the rest of thes
 64-bit world?i  D 5.  Are you guys ever going to write that tell-all book about Alpha?  K Well, I *tried* to keep to subjects that Intel wouldn't have any legitimate L objection to.  But you never know.  Feel free to offer some bonus answers toA substitute for any you'd feel uncomfortable expressing in public.c   Thanks,    - bill   ------------------------------  % Date: Wed, 24 Jul 2002 11:15:10 +0100tU From: Andrew Harrison SUNUK Consultancy <andrew_nospam.harrison_remove_this@sun#.com>)4 Subject: Re: Itanic2 - the cHumPaq spin continues...0 Message-ID: <ahlun8$a5c$1@new-usenet.uk.sun.com>   JF Mezei wrote:t  M > You can have as much propaganda and performance number fudging as you wish.s > K > But in the end, both Alpha and PA-RISC will eventually fade due to no newnN > performance improvements and IA64 will become HP's leading performance chip,L > possibly used by other companies. Pointless to argue what Alpha could haveP > been. We have to accept that IA64 will be HP's leading chip no matter what its > performance is.a > J > Sun managed to be extremely succesful with a chip that wasn't as fast asL > Alpha, there is no intrinsic reason that HP couldn't succeed with a slowerF > IA64 chip versus Power or whatever else exists in the server market. >     E Different Issue, Intel will not continue to invest in IA-64 if HP aretG their only customer. They have to get all the vendors who were buildingtC 4-8 CPU Wintel boxes to switch to IA-64. The cost of developing anduF FABing IA-64 will not be worth the returns with only HP as a customer.  G HP needed to get out of HP-PA design and build because the costs didn'taF justify the returns. If the IA-64 market simply ends up as replacementC for the HP-PA and Alpha processors then it won't take Intel long ton& reach the same conclusion that HP did.  H Sun has much higher US volumes than HP-PA or IBM Power or Alpha, in factC the unit volumes are about the same as all 3 combined. Sun also hashB a different model for developing and FABing SPARC which means that9 the costs are spread between 2 companies rather than one.g     Regardsr   Andrew Harrisonn   ------------------------------  % Date: Wed, 24 Jul 2002 12:26:40 +0100i( From: Nic Clews <sendspamhere@127.0.0.1>4 Subject: Re: Itanic2 - the cHumPaq spin continues...) Message-ID: <3D3E8EF0.B50DE081@127.0.0.1>   ( Andrew Harrison SUNUK Consultancy wrote: >  ...e< > Like it or not the future of OpenVMS is IA-64, you are now= > at the mercy of the Wintel Market. If IA-64 succeeds in the 9 > Wintel market then OpenVMS has a long term platform andn6 > a future (assuming HP do the right things). If IA-648 > fails and Hammer may well be its undoing and Intel are4 > forced to base their 64 bit ISA around x86-64 then6 > you will not have a platform for OpenVMS and it will > die.  C Ah, this is obviously some strange usage of the word 'die' I wasn't H previously aware of. If OpenVMS (and the remainder of HPs prime softwareH portfolio) is dependent on IA64, then simply fewer IA64s are produced if( not required for the Windows fraternity.  H So, you have the low ends on the (lets say) Hammer, and the high ends onE the 'others'. Some other chip dominating the market doesn't magicallyf mean others disappear.  E You've also missed that 'port' actually means 'portable' in this caseRH for OpenVMS. IA64 just happens to be the development platform currently., On that merit alone you mention a non issue.   -- d? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciencess nclews at csc dot comn   ------------------------------  % Date: Wed, 24 Jul 2002 13:25:01 +0100fU From: Andrew Harrison SUNUK Consultancy <andrew_nospam.harrison_remove_this@sun#.com>m4 Subject: Re: Itanic2 - the cHumPaq spin continues...0 Message-ID: <ahm6an$ckv$1@new-usenet.uk.sun.com>   Nic Clews wrote:  * > Andrew Harrison SUNUK Consultancy wrote: >  > ...a > < >>Like it or not the future of OpenVMS is IA-64, you are now= >>at the mercy of the Wintel Market. If IA-64 succeeds in theg9 >>Wintel market then OpenVMS has a long term platform anda6 >>a future (assuming HP do the right things). If IA-648 >>fails and Hammer may well be its undoing and Intel are4 >>forced to base their 64 bit ISA around x86-64 then6 >>you will not have a platform for OpenVMS and it will >>die. >> > E > Ah, this is obviously some strange usage of the word 'die' I wasn'tfJ > previously aware of. If OpenVMS (and the remainder of HPs prime softwareJ > portfolio) is dependent on IA64, then simply fewer IA64s are produced if* > not required for the Windows fraternity. >   = HP themselves could not justify the continued development and ? manufacture of CPU's just for consumption in their own systems.   > Without the other manufacturers and the Windows fraternity why> would Intel continue to do what HP themselves didn't think was	 economic.0  ; If HP port OpenVMS to IA-64 and then IA-64 dies because itsc< eaten by Hammer in the wintel space and is uncompetitve with> SPARC and Power outside that then OpenVMS will be left without= a platform to run on. An OS without a platform is dead to allf intents and purposes.u    J > So, you have the low ends on the (lets say) Hammer, and the high ends onG > the 'others'. Some other chip dominating the market doesn't magicallye > mean others disappear. > G > You've also missed that 'port' actually means 'portable' in this case-J > for OpenVMS. IA64 just happens to be the development platform currently.. > On that merit alone you mention a non issue. >     A So, you are HP senior managment, your nightmare scenario has justiA happened, Intel have faxed you to tell you that the IA-64 volumesc? don't justify continued development and that they are switchingt> to x86-64 but they will continue to supply you with Mckinley'sA for X months. You have just shelled out a bunch of cash to designnB a HW platform and port 3 major OS's to a platform that is dead andB have spent even more persuading the market that its an OK platform# and persuading ISV's to port to it.e  @ What do you do ? turn around and do it all over again with the 3B OS's. Remember doing the port isn't really the issue its designing? the HW platforms, compilers, getting the ISV's etc all of whiche you have to throw.  ; I very much doubt that OpenVMS would survive this scenario.    Regards  Andrew Harrison    ------------------------------  + Date: Wed, 24 Jul 2002 14:29:01 +0000 (UTC)d From: david20@alpha2.mdx.ac.uk4 Subject: Re: Itanic2 - the cHumPaq spin continues...+ Message-ID: <ahmdjd$7br$1@aquila.mdx.ac.uk>y  \ In article <3D3DC8B1.87AE2610@videotron.ca>, JF Mezei <jfmezei.spamnot@videotron.ca> writes:L >You can have as much propaganda and performance number fudging as you wish. > J >But in the end, both Alpha and PA-RISC will eventually fade due to no newM >performance improvements and IA64 will become HP's leading performance chip,cK >possibly used by other companies. Pointless to argue what Alpha could havecO >been. We have to accept that IA64 will be HP's leading chip no matter what itsp >performance is. >.I >Sun managed to be extremely succesful with a chip that wasn't as fast aseK >Alpha, there is no intrinsic reason that HP couldn't succeed with a slowereE >IA64 chip versus Power or whatever else exists in the server market.d >i  J But that is on the back of a large application portfolio built up over theL years when Sun price/performance was fairly good compared to the opposition.  O Itanic is in the unenviable position  of both being a poor performer and havingc0 no applications when compared to the opposition.    
 David Webb VMS and Unix team leader CCSS Middlesex University   ------------------------------  % Date: Wed, 24 Jul 2002 15:30:41 +0100o( From: Nic Clews <sendspamhere@127.0.0.1>4 Subject: Re: Itanic2 - the cHumPaq spin continues...) Message-ID: <3D3EBA11.D6F23FD3@127.0.0.1>   ) Andrew Harrison SUNUK Consultancy wrote: i? > HP themselves could not justify the continued development andlA > manufacture of CPU's just for consumption in their own systems.u > @ > Without the other manufacturers and the Windows fraternity why@ > would Intel continue to do what HP themselves didn't think was > economic.t > = > If HP port OpenVMS to IA-64 and then IA-64 dies because its > > eaten by Hammer in the wintel space and is uncompetitve with@ > SPARC and Power outside that then OpenVMS will be left without? > a platform to run on. An OS without a platform is dead to alla > intents and purposes.   B The first point is groundless and simply means that chip costs areC higher with lower volumes. Arguably this has already happened [withrD moving to IA64] but if HP customers want to buy systems with IA64 inD them, I'm fairly sure Intel will keep making IA64s if HP wants them.  A Intel will continue because someone is paying them, quite simply.m  G Digital quit the chip manufacturing business, passing it over to Intel,'F quite happily making (blindfolded of course ;-) ) Alpha chips, IBM hasG been churning out a few for Compaq's AlphaServers, so I don't quite seetF your point. The stuff currently coming out of Microsoft doesn't run on? any of these chips, yet the chips are still being manufactured.l  C > So, you are HP senior managment, your nightmare scenario has justnC > happened, Intel have faxed you to tell you that the IA-64 volumesuA > don't justify continued development and that they are switchingl@ > to x86-64 but they will continue to supply you with Mckinley'sC > for X months. You have just shelled out a bunch of cash to design D > a HW platform and port 3 major OS's to a platform that is dead andD > have spent even more persuading the market that its an OK platform% > and persuading ISV's to port to it.t > B > What do you do ? turn around and do it all over again with the 3D > OS's. Remember doing the port isn't really the issue its designingA > the HW platforms, compilers, getting the ISV's etc all of whicho > you have to throw.  D I suspect that OpenVMS will make movements to other platforms (as ifG this port didn't count) before the fate of IA64 one way or the other islH known. This has already been hinted at publicly with timescales *inside*G the existing expected AlphaServer support timeframe, never mind that ofm9 IA64. (In the presentation "IPF Base OS implementation").n  A OpenVMS is not being ported to IA64 specifically, it's having theeC hardware dependencies removed so it can run on IA64. Far from beingoD thrown away, the hard work will have been done. (I cannot comment on- HP-UX or NSK, I don't know them well enough).s  G The UNIX portability initiatives in OpenVMS are creating an environmentpD such that if XYZ's application compiles on a UNIX-98 environment, it= will also compile (and therefore be available on) the OpenVMSuF environment. While acceptably it does not mean it will happen, it does remove the barrier.   D Senior HP execs could say to Intel, "keep making the chips cos we'reH paying you", or in Blue Peter fashion, here's a version of the operatingA system for "the name is irrelevant" architecture we made earlier.    Compile and go.   t= > I very much doubt that OpenVMS would survive this scenario.i  G Correct marketing is what is important. You've also disregarded the diedG hards that use it anyway, and don't even care how fast or on what. Theym9 keep spending the money even in "this financial climate".f -- a? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciencese nclews at csc dot coma   ------------------------------  + Date: Wed, 24 Jul 2002 14:38:18 +0000 (UTC)s From: david20@alpha2.mdx.ac.uk4 Subject: Re: Itanic2 - the cHumPaq spin continues...+ Message-ID: <ahme4q$7br$2@aquila.mdx.ac.uk>   \ In article <3D3DCA69.6CCF9E3F@videotron.ca>, JF Mezei <jfmezei.spamnot@videotron.ca> writes:  >david20@alpha2.mdx.ac.uk wrote:/ >> VMSs future is (currently) tied to Itanic.  i >rN >Currently, VMS' future is tied to Alpha. It won't be tied to IA64 for anotherN >few years. The period between the declaration of Alpha's murder and the firstK >marketing of VMS on IA64 hardware (if such marketing ever materialises) isoK >perhaps one of the most critical period in this sick/weak operating systemtL >that is now forced to survive through a very difficult void period, merger,< >new architecture that is still an unknown quantity etc etc.   Sorry I should have said :-a  5 VMSs longterm future is (currently) tied to Itanic.  c    N The (currently) referred to the possibility/hope that HP might wake-up to the P need to provide an alternative long term future chip for VMS if things continue  badly for IA64.a    
 David Webb VMS and Unix team leader CCSS Middlesex University t   ------------------------------  % Date: Wed, 24 Jul 2002 11:40:51 -0400e- From: JF Mezei <jfmezei.spamnot@videotron.ca>e4 Subject: Re: Itanic2 - the cHumPaq spin continues..., Message-ID: <3D3ECA7F.F9ABBD08@videotron.ca>  ( Andrew Harrison SUNUK Consultancy wrote:G > If IA-64 fails it will be for a number of reasons, price, performance2   Define "failure" for IA-64.sW 	-will it fail to instantly capture the world and become industry-standard ? Certaintly9= 	-will it fail to blow other chips's performance ? Certaintly D 	-will it fail to instantly capture wintel server market ? Probably.2 	-will it fail to give HP a usable processor ? No.4 	-will it give HP a competitive edge ? Probably not.  N I tend to see IA64 as a liability for HP, not an asset. They are stuck with anF inferior chip to Alpha that will remain, for the foreseable future, as1 proprietary to HP as Alpha was to Digital/Compaq.b  K However move the clock forwards 5 years, and what happens if the 8086 (bothtM Intel's and AMD's Hammer) start to lag behind the IA64 ? At that point,  IA64rL might start to capture some of the "industry standard" market and look a lot better than it does today.   There are three big questions:N -Will Hammer be able to keep up with its 1970s 8086 "core" against 1990s Power and IA64 IN THE LONG TERM ?u  N -What are Intel's true intentions with regards to the 8086 product line ? WillM Intel continue to push that chip's speed as much as possible or will it allow I it to start to lag others since Intel will be able to offer IA64 for highi
 performance ?   M -Will Windows ever require the processing power of such processors or will itoL be relegated to desktops running Office and servers running Exchange and IIS to help spread viruses ?  I If Windows doesn't get those high end "enterprise" stuff, then it will bet: perfectly happy to stay with the 8086 even in 32 bit mode.    M When Alpha first came out, I also recall comments about it being very hot etcrN etc. So I would suspect that Intel will, over the next few years, improve that aspect of the chip.l  M I think that IA64 is still in "beta" mode. That image is strenghetened by the.L fact that it isn't ready to be incorporated into commercial enterprise class) systems capable of running VMS or Tandem.p   ------------------------------  % Date: Wed, 24 Jul 2002 16:21:02 +0100 U From: Andrew Harrison SUNUK Consultancy <andrew_nospam.harrison_remove_this@sun#.com> 4 Subject: Re: Itanic2 - the cHumPaq spin continues...0 Message-ID: <ahmgkp$g2k$1@new-usenet.uk.sun.com>   Nic Clews wrote:  + > Andrew Harrison SUNUK Consultancy wrote: y > ? >>HP themselves could not justify the continued development andtA >>manufacture of CPU's just for consumption in their own systems.t >>@ >>Without the other manufacturers and the Windows fraternity why@ >>would Intel continue to do what HP themselves didn't think was >>economic.r >>= >>If HP port OpenVMS to IA-64 and then IA-64 dies because its > >>eaten by Hammer in the wintel space and is uncompetitve with@ >>SPARC and Power outside that then OpenVMS will be left without? >>a platform to run on. An OS without a platform is dead to all  >>intents and purposes.l >> > D > The first point is groundless and simply means that chip costs areE > higher with lower volumes. Arguably this has already happened [withcF > moving to IA64] but if HP customers want to buy systems with IA64 inF > them, I'm fairly sure Intel will keep making IA64s if HP wants them. >     C Continuing to fabricate and continuing to develop are two differentl? things. Would Intel commit to the expenditure necessary to fundeA future generations of IA-64 processors if HP are their only majore? customer, I don't know but since HP themselves don't think thatd= doing a specific processor for HP alone is justifiable from a 7 cost benefit basis its unlikely that Intel will either.m    C > Intel will continue because someone is paying them, quite simply.g >     = Well of course they will but then will HP continue to pay, if < HP end up paying for all the IA-64 development costs and the= FAB costs spread over a much lower volume of IA-64 processors)= than they were expecting then eventually HP will have to give-; up. If it didn't make sense to bear the cost of HP-PA to HP 9 then simply giving the process to Intel but still bearingf7 the full cost from an HP standpoint is also unlikely tod be attractive.  8 The whole point of IA-64 was so that HP could get out of7 being a CPU developer and manufacturer and allow othersb8 e.g Intel to bear the costs of this. This will only work7 however if Intel can get a bigger market than HP itselfP- of of course if HP can massively grow itself.c    I > Digital quit the chip manufacturing business, passing it over to Intel,fH > quite happily making (blindfolded of course ;-) ) Alpha chips, IBM hasI > been churning out a few for Compaq's AlphaServers, so I don't quite seemH > your point. The stuff currently coming out of Microsoft doesn't run onA > any of these chips, yet the chips are still being manufactured.  >     : Well if you beleive Compaq then Alpha wasn't viable either9 hence the Alphacide and Compaq was only talking about thet; development costs not the FAB costs because they had as youP said gone FABless.  @ The IA-64 model says that eventually Intel will be designing and@ FABing IA-64's. If Intel are forced to major on x86-64 then they@ will have design efforts split across two competing incompatibleC 64 bit CPU's. If they only have HP as a market for IA-64 with their A competion for all other design wins in the Wintel space being AMDe? then the team getting the most resources will inevitably be theR' x86-64 team rather than the IA-64 team.b  = HP is currently the market leader with ~15% of the PC market.i? So in the Wintel market alone if IA-64 becomes HP's proprietary1> CPU and the rest of the market marches towards x86-64 then 85%> of the current market is up for grabs. In this situation Intel? would be mad to concentrate on IA-64 development and would haver! to focus their efforts on x86-64.e     regardse   Andrew Harrisona   ------------------------------  % Date: Wed, 24 Jul 2002 11:53:50 -0400i- From: JF Mezei <jfmezei.spamnot@videotron.ca>r4 Subject: Re: Itanic2 - the cHumPaq spin continues..., Message-ID: <3D3ECD89.AF82E9D5@videotron.ca>  ( Andrew Harrison SUNUK Consultancy wrote:I > HP needed to get out of HP-PA design and build because the costs didn'taH > justify the returns. If the IA-64 market simply ends up as replacementE > for the HP-PA and Alpha processors then it won't take Intel long to ( > reach the same conclusion that HP did.  I It all depends on the type of deal between HP and Intel. Compaq said thatoK Alpha wasn't profitable because it found a way to fudge the numbers to show H what it wanted to show. If Compaq had wanted to, they wouldn't even haveE needed to fudge numbers to kill off the unprofitable wintel business.t  N Alpha was part of a very profitable business that Compaq wasn't interested in.  M HP will be able to portray the IA64 costs as being instrumental in generatingdN most of HP's profits, even if IA64 is relegated to running only the enterpriseM class systems. Alpha was the bastard child that Compaq didn't want. HP is the J golden child that was conceived in part by HP and HP will do its upmost to protect and cherish it.c  I Ironically, I think that HP would have been far better off if it had told-N Compaq to outsource Alpha to Intel and then proceed to abandon IA64.  HP wouldN have only needed to migrate HP-UX to Alpha, VMS and Tru64 would have remained,2 and Tandem would have continued the port to Alpha.  N Just imagine what the Digital engineers could have done with Alpha if they hadL all of Intel's money and resources available instead of working in the tight confones of Compaq's budgets.>   ------------------------------  % Date: Wed, 24 Jul 2002 12:10:54 -0400l- From: JF Mezei <jfmezei.spamnot@videotron.ca> 4 Subject: Re: Itanic2 - the cHumPaq spin continues..., Message-ID: <3D3ED187.499DACC1@videotron.ca>  ( Andrew Harrison SUNUK Consultancy wrote:? > HP themselves could not justify the continued development andeA > manufacture of CPU's just for consumption in their own systems.t   Reality check here:t  M Was PA-RISC at a technological dead end or was its acrhitecture fully capablet? of being continued into a 64 bit world and remain competitive ?d  G If PA-RISC was at a technological dead end, then it isn't a question ofiF "continued development" but rather "develop a brand new architecture".  J Developping a brand new architecture costs a whole lot more than continued= incremental improvements of an existing architecture. Right ?l  J Once the IA64 is developped and goes into "incremental improvements" mode,J then its costs should be  lower, and it will be generating *some* revenue,J which is much better than what IA64 has been doing since the mid 1990s. AsJ long as IA64 generates a bit of profit for Intel, then I see no reason forL Intel not wanting to continue the job of producing HP's proprietary chip andF try to recuperate some of the investment that was wasted in that chip.  K If a company realises that one of its office buildings is underused, it canjJ sell the building and then reinvest the capital in a new plant to increase4 production. Better , more profitable use of capital.  G But how much capital is tied up with IA64 ? If they kill IA64, will anyoN capital be usable for a more profitable venture, or will it be just a questionI of reducing headcount ?  If there isn't much capital freed up, then therepG isn't much point in killing IA64 as long as it generates a tiny profit.t   ------------------------------  % Date: Wed, 24 Jul 2002 12:26:55 -04003- From: JF Mezei <jfmezei.spamnot@videotron.ca>e4 Subject: Re: Itanic2 - the cHumPaq spin continues..., Message-ID: <3D3ED548.F1C36354@videotron.ca>   Nic Clews wrote:F > I suspect that OpenVMS will make movements to other platforms (as ifI > this port didn't count) before the fate of IA64 one way or the other is0J > known. This has already been hinted at publicly with timescales *inside*I > the existing expected AlphaServer support timeframe, never mind that ofn; > IA64. (In the presentation "IPF Base OS implementation").s > C > OpenVMS is not being ported to IA64 specifically, it's having them5 > hardware dependencies removed so it can run on IA64b  M Such statements are reminiscent of Bob Ceculski's "EV8 will live inside IA64"tG statements. I think that will all wish to see VMS moved to a succesfull N architecture, not one that is seen as a big dud with about as much uncertainty= about long term prospects as Alpha had prior to June 25 2001.p  H The problem is that HP is very unlikely to fund a port of VMS to anotherM platform. Remember that the port is presumably funded by Intel. Moving to yeteL another platform would be against stated policy of concentrating on a single2 platform, as Carly and Curly stated so many times.  K Now, the VMS engineers are being smart in using this intel funded effort tofN restructure VMS to be more portable so that if VMS is ever sold, it can easilyL be ported to another platform. But I don't see this as an indication that HP= has intentions to allow VMS to be ported to another platform.    ------------------------------  % Date: Wed, 24 Jul 2002 17:24:18 +0100 U From: Andrew Harrison SUNUK Consultancy <andrew_nospam.harrison_remove_this@sun#.com>t4 Subject: Re: Itanic2 - the cHumPaq spin continues...0 Message-ID: <ahmkbd$h70$1@new-usenet.uk.sun.com>   JF Mezei wrote:P  * > Andrew Harrison SUNUK Consultancy wrote: > ? >>HP themselves could not justify the continued development andoA >>manufacture of CPU's just for consumption in their own systems.  >> >  > Reality check here:o > O > Was PA-RISC at a technological dead end or was its acrhitecture fully capableVA > of being continued into a 64 bit world and remain competitive ?t >     ? No PA is a 64 bit processor now and there is no suggestion thatt9 given the resources HP couldn't have kept it competitive.n  < It has suffered from being the processor that IA-64 is going6 to replace, because of this HP have held back on major7 developments in favour of IA-64. Even with this in mindh5 PA is relatively competitive with IA-64 so HP are nowu4 putting resource into downplaying how close PA is to IA-64.   Regardsl Andrew Harrisons   ------------------------------  % Date: Wed, 24 Jul 2002 07:49:08 -0400u2 From: Atlant Schmidt <atlantnospam@mindspring.com>3 Subject: Re: Need 80' Fortran/VAX training materiala. Message-ID: <3D3E9433.ED8E0903@mindspring.com>   Didier Morandi wrote:   , > the pedagogy (is this an English word? :-)   Oui!   Atlant   ------------------------------  % Date: Wed, 24 Jul 2002 15:39:55 +0200c- From: Didier Morandi <Didier.Morandi@Free.fr>d3 Subject: Re: Need 80' Fortran/VAX training materialr' Message-ID: <3D3EAE29.213DB078@Free.fr>a   Merci.  J BTW, I just finished to read the French version of the book "The Eight" byA Katherine Neville (Ballantines Books 1988, Le Cherche Midi 2002).   N Pleasant to find that many expressions were written in French in the original.  V I do recommend this book to those who do not know it, particularily if you play chess.  + Is Katherine a famous writer in the States?t   D.   Atlant Schmidt wrote:i >  > Didier Morandi wrote:  > . > > the pedagogy (is this an English word? :-) >  > Oui!   ------------------------------  % Date: Wed, 24 Jul 2002 13:54:42 +0200 9 From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com>h4 Subject: Re: Only 20% drop in VMS systems (was: wow)' Message-ID: <3D3E9582.BA98D256@aaa.com>    Sorry Andrew, but you wrote :e  +  "We are still waiting on both of these..."w  
 Who's "We" ??d   I'm I part of the "We" ?: No, I can't be, since I'm not waiting for anything of what you described !n  ? Can't all one-to-one stuff on this NG be moved into an ordinaryn= mail conversation between those 2 (or sometimes 3) involved ?    Best Regards Jan-Erik Sderholm.e    ( Andrew Harrison SUNUK Consultancy wrote: > , > BTW you were going to get back to me with: > 1.1 > Datapoints that support the Alpha is fast SPARCe >         is slow conjecture.i > 2.4 > Where you have had a sucessfull technical argument >         with me in the past. > * > We are still waiting on both of these...   ------------------------------  % Date: Wed, 24 Jul 2002 12:16:56 +0100pU From: Andrew Harrison SUNUK Consultancy <andrew_nospam.harrison_remove_this@sun#.com> 4 Subject: Re: Only 20% drop in VMS systems (was: wow)0 Message-ID: <ahm2b2$bcb$1@new-usenet.uk.sun.com>   jlsue wrote:  D > Once again, Andrew, you show your incredible inability to read and
 > understand.n > D > You can not prove me incorrect in what I say - mostly because I amH > speaking about real-world experience - so you try to twist things with# > innuendo and disparaging remarks.' >v    A > Please don't waste our time with more of this garbage until youi > actually have a clue.- >     < No you are the time waster, you jump into a discussion about7 SMP vs Cluster scalability with examples that you admit @ a primarely HA and are then unable to recollect what scalability you observed in real life.  = If you have forgotten your "real-world" experience taught you@: then it wasn't really a very usefull real-world experience was it.   : If you had a clue you would have jumped in with real world9 examples of cluster scalability, you didn't so I think inn: the circumstances you have proven to be the person without a clue.d  * BTW you were going to get back to me with: 1. n/ Datapoints that support the Alpha is fast SPARC  	is slow conjecture. 2. a2 Where you have had a sucessfull technical argument 	with me in the past.c  3 We are still waiting on both of these, wouldn't youa1 be better off trying to salvage Alpha and you own 8 tattered reputations rather than jumping into a argument> where you had to admit to having "forgotten" the only peice of2 information that might have made your contribution usefull or interesting.r   Regardso Andrew Harrisona     >  > G > On Thu, 18 Jul 2002 13:31:34 +0100, Andrew Harrison SUNUK ConsultancyV6 > <andrew_nospam.harrison_remove_this@sun#.com> wrote: >  >  >> >>jlsue wrote: >> >>H >>>On Wed, 10 Jul 2002 18:42:40 +0100, Andrew Harrison SUNUK Consultancy7 >>><andrew_nospam.harrison_remove_this@sun#.com> wrote:i >>>o >>>e >>>j >>>>jlsue wrote: >>>> >>>> >>>>J >>>>>On Tue, 09 Jul 2002 19:23:18 +0100, Andrew Harrison SUNUK Consultancy9 >>>>><andrew_nospam.harrison_remove_this@sun#.com> wrote:  >>>>>r >>>>>s >>>>>i >>>>>f >>>>>2	 >>>>>>1. "7 >>>>>>You need to be specially trained to manage an OPS 8 >>>>>>	environment for starters, this might in itself be5 >>>>>>	a hint to you that managing an OPS cluster hasl: >>>>>>	additional skill requirements to a single instance. >>>>>> >>>>>> >>>>>>E >>>>>I think you overstate the complexities involved.  Sure, there isvH >>>>>additional training, so what?  Every time you get a new version youD >>>>>need some bit of training for the changes/new features as well.. >>>>>And it's not as if there are no benefits. >>>>>A >>>>>s >>>>>i/ >>>>Come on you must understand the difference.  >>>>3 >>>>There is an Oracle 8i DBA course or a 9i course'> >>>>and then once you have done that if you want to administerE >>>>a OPS or RAC cluster there is a 8i OPS course or a 9i RAC course.a >>>>3 >>>>Its an additional layer that you have to learn.s >>>> >>>>G >>>Hmm... Don't see how anything I said above contracticts what  you'verG >>>said.  Training is part of the package for any of this, and there isrH >>>little-to-no difference whether this training is for new versions, orI >>>for an add-on that provides more features.  I've been through trainingcD >>>in both cases.... most intelligent folks wouldn't have a problem, >>>imho. >>>q >>>t >>>oE >>>>>It's a trade-off.  Many, many customers find the availabilty andrG >>>>>scalability features of VMSclustering far outweigh this complexitys% >>>>>that you keep spouting on about.  >>>>>t >>>>>" >>>>> E >>>>Get back to the point, Kerrys claim is that from a DBA standpointmF >>>>tuning and running a DBMS on a large SMP system is as difficult as4 >>>>tuning and running a DBMS on a cluster of nodes. >>>>@ >>>>This is BS and no DBA will support his claim as we all know. >>>> >>>> >>>>C >>>As I detail below, very little tuning has ever been needed in mytH >>>environments.  SGA sizes were the most common piece that needed to beC >>>modified - other than that, it was typically related to keys not H >>>existing where they needed to be.  And OPS did not make that any more >>>complex.e >>>p >>>r >>>_	 >>>>>>2. @9 >>>>>>Tuning is very different, in addition to the normale8 >>>>>>	tuning issues of data placement etc you also need8 >>>>>>	to consider partitioning data, something that you9 >>>>>>	may do on an SMP system but for different reasons.  >>>>>> >>>>>> >>>>>>I >>>>>What?  Well, none of this was true in the VMSclusters I ran that had I >>>>>Oracle clustered on them.  Tuning consisted of merely making the SGAnG >>>>>as large as we wanted to reduce real I/O to disk, and there was NOo+ >>>>>need whatsoever to partition the data.s >>>>>  >>>>>  >>>>>eE >>>>Really, so what was your scalability, did the DBMS deliver 2x thed< >>>>thoughput on two nodes when compared with one node ????? >>>>A >>>>I have been responsible for OPS clusters on Solaris which are>B >>>>based on OPS to provide very high reliability, not scalability? >>>>no tuning for it but then it wasn't the reason that OPS wasp >>>>being used.o >>>> >>>> >>>>I >>>Well, scalability is only one measurement.  To tell the truth, I can'tnH >>>recall the scalability on these apps, but I do know that it wasn't anF >>>issue, or the managers wouldn't have kept supporting it with budgetH >>>dollars.  But as I stated earlier, it's all a trade-off between cost,D >>>speed, and reliability.  You can pick any two, but not all three. >>>h >>>v >>> >>Since this discussion is about scalability it seems odd that= >>you would have bothered contributing since you don't appearl4 >>to remember whether your point is relevant at all. >> >> >>	 >>>>>>3. i8 >>>>>>There are financial penalties for using OPS, which9 >>>>>>	any administrator in his/hers right mind will needv >>>>>>	to consider.e >>>>>> >>>>>> >>>>>>K >>>>>Sure, as well as the financial penalties of increasing the size of theuH >>>>>single server when other apps on the same server grow.  I know thisI >>>>>one from experience.  Oracle apps remained constant, other apps were J >>>>>growing, requiring a server upgrade, which meant over $250K (USD) for+ >>>>>Oracle upgrades we really didn't need.  >>>>>  >>>>>n >>>>> < >>>>Nope sorry. Learn your charging model before responding. >>>>> >>>>Oracle charge on a per CPU basis, so my 32 CPU server will< >>>>cost 32x the price of a single CPU system. Then they add? >>>>OPS on top. So is you have 2 x 16 CPU nodes you get chargeds4 >>>>the standard 32 CPU charge plus the cost of OPS.
 >>>>[snip...]c >>>>; >>>>The deal looks even worse when you realise that currentp= >>>>large GS boxes deliver less throughput per CPU than Sun'sf= >>>>and in a cluster will not unless we are talking somethingeB >>>>like TPC-C deliver linear or even remotely linear scalability. >>>>9 >>>>The best case ratio which is if you don't have volume = >>>>deal and are buying per CPU puts a cluster being used foro= >>>>scalability at more than 2x the Oracle licensing costs tos; >>>>get the same throughput. And that ignores the increasedg  >>>>TCO of the clustered option. >>>> >>>> >>>>C >>>This isn't all based on facts.  TCO takes many, many things intopC >>>account.  And availability does provide value to most apps.  AndrH >>>whether a single GS box delivers less throughput or not (and I do notG >>>believe this is true in the environments I've work in), if the total I >>>cost of the hardware to provide equal-or-better throughput is the sametF >>>as the single-server model, then so what?  In the environments I'veC >>>worked in , clustering did NOT increase the management costs anysI >>>appreciable amount, and the value of availability certainly outweighedt
 >>>that cost.o >>>e >>>t >>C >>Which bit did you actually disagree with, you will recollect thats@ >>you, Kerry et al have all failed to provide data that counters@ >>my 2-3 320's to one F15000 points. Or the points about GS320's% >>delivering less throughput per CPU.  >>H >>If you have examples of like for like publically available performanceC >>data that prove your point then produce it otherwise what you areP  >>suggesting is pure conjecture. >>C >>I also have no doubt that managing a cluster using staff equipped C >>with no training but with a copy of OpenVMS Clusters/OPS ClusterssB >>for dummies in snuggly in their back pocket is cheap, however asC >>you know most people prefer their staff to be properly trained oreC >>to have extensive experience both of which you end up paying for.t >>D >>It is also apparent from you subsequent posts that your experienceF >>does not extend to having to configure and tune apps for scalabilityG >>in a cluster, just to be highly available. This discussion however ise? >>about the costs of getting apps to scale in SMP and Clusteredy@ >>environments so while your information might have been usefull) >>in some other context it isn't in this.i >> >> >>	 >>>>>>4.  ; >>>>>>Your storage requirements are entirely different, you = >>>>>>	now need a shared storage subsystem rather than single  >>>>>>	host attach.m >>>>>> >>>>>> >>>>>>G >>>>>Not a problem.  It's completely transparent in my VMSclusters.  In I >>>>>fact, when these customers see how other servers & clusters work wrtn >>>>>storage, they cringe. >>>>>t >>>>>t >>>>> C >>>>You seem to miss the point again, the best and cheapest storageoD >>>>for a datawarehouse is often JBOD drives, the read-ahead caching% >>>>in a array just gets in the way.   >>>> >>>>F >>>No, JBOD drives INCREASE the management, tuning, and administrationI >>>costs for systems, and especially for Oracle databases.  An enterprise F >>>storage configuration (a.la. Cpq/hp EVA subsystems) actually lowersG >>>this tuning and administration cost because the hardware manages theDI >>>stastical load balancing rather than the DBA.  This is a tough one forPI >>>DBAs to get past:  Many of the ones I've dealt with have spent so much I >>>time and effort managing table spaces on spindles that it is difficulttI >>>for them to give up.  But it's just not necessary anymore when all I/Oa+ >>>can be spread over hundreds of spindles.d >>>r >>>  >>? >>BS, A RAID array doesn't manage storage, it provides you with ? >>an interface that allows you to manage the storage. With JBOD ? >>drives you have a host based volume manager which you need toe@ >>configure for mirroring, stripes, stripe size etc, you have toC >>do exactly the same thing on a RAID array using its configurationn= >>manager, the LUN's don't just create themselves and presento
 >>themselves.a >>@ >>The tuning and administration buttons you have to push, stripe@ >>width, interleave etc are just the same using JBOD as hardwareB >>RAID, which in addition may require you to play with the cachingA >>as well which in any case for most DW apps has a tendancy to be % >>more of a hinderance than an asset.t >>? >>In addition if you want multi-pathing you need to manage thattB >>at the host, something you don't need for mirrored striped JBOD. >>? >>Not one of your best efforts at a technical response you must,
 >>try harder.e >> >> >> >>>  >>>>If you don't want a HAE >>>>datawarehouse system and a lot of people don't then clustering to E >>>>get scalability for DW will increase your storage costs, you needd. >>>>dual pathing etc and reduce your choices.  >>>> >>>>E >>>Clustering will not provide scalability for all apps.  However, intA >>>many, many application workloads, we can configure the cluster5E >>>activity to stastically balance among servers in a sort-of-similarII >>>fashion to the I/O spread used in disk arrays.  I've done it.  Does itoC >>>work for large, long-running batch jobs... well, yes and no.  Itm >>>depends on lots of things.  >>>s >>>  >>? >>But then the specific discussion is one about the scalabilitywD >>of clusters, SMP boxes and the relative ease of getting it in both@ >>environments. The fact that you are uncertain that clusters do@ >>deliver scalability in the terms Kerry was describing suggestsC >>that you should have an offline discussion with him and come backv# >>with a more coeherant party line.o >> >> >> >>>>All drives/arrays withinF >>>>reason can be connected to a single system subsets of this can be H >>>>connected to clusters. Compaqs own clustered TPC-H number for ES45'sD >>>>shows this with a relatively complex switch/array infrastructureD >>>>imposed by a need to have connectivity from all the nodes in the >>>>cluster to all the storage.u >>>> >>>>H >>>Well, VMSclusters had shared storage with CI technology over 15 yearsI >>>ago.  This was NOT a "relatively complex switch/array infrastructure".dI >>>It may have been pricey, but at the time it was the only game in town,oH >>>so it's kind of a cheap shot to call it "costly" when there's nothing >>>else to compare.e >>>t >>>M >>D >>But this has as you know been largely been replaced by FC which is? >>a realtively complex switched infrastructure so your point is  >>hardly relevant is it. >> >> >> >>K >>>>>Support/Schmupport.  You'd never imagine how many times vendors try touK >>>>>pull that crap.  If problems occur with software, it's a simple mattermJ >>>>>to reproduce the problem with only one server running the app... thatG >>>>>shuts the vendor up as far as finger-pointing goes.  Rarely, and IwI >>>>>mean RARELY (as in, never in 17 years) have I had a problem that waso >>>>>cluster-specific. >>>>>e >>>>>n >>>>> F >>>>Akk, and you are customer facing, you would propose, configure andF >>>>install a platform to host an application that wasn't qualified or  >>>>supported for that platform. >>>> >>>>I >>>However, not all customers who I deal with today, as a consultant, arePG >>>as concerned that each and every thing they do has complete blessingeI >>>from a vendor.  As a consultant, it's my job to inform them the limitsgH >>>to which they can expect support, and if they choose to go outside ofI >>>that environment, then I can help them understand the trade-offs.  And B >>>if they choose to continue unsupported, I can help them develop; >>>operating disciplines that address their business needs.  >>>w >>>i >>0 >>Blimey you really are a consultant arn't you ! >>G >>Do you tell them that the platform solution you have proposed to them>< >>is what is forcing them to consider a unsupported solution) >>as far as the apps vendor is concerned.  >> >> >>H >>>This does not mean that any unsupported option is good.  But one mustI >>>understand the risks and implications before one can determine what is0F >>>reasonable, and also fully realize what assuming that risk in-house >>>really means. >>>m< >>>For example, we write many, many scripts to do day-to-dayD >>>administrative functions.  These are "unsupported" by any vendor.E >>>However, all system administrators do this regularly.  If problems)I >>>crop us, we take the responsibility to support them to the extend thatoB >>>the issues are not related to bugs in vendor-supplied software. >>>l >>>  >>= >>This is rather different from say configuring say a cluster = >>of DBMS servers to support an application which is a config - >>that the ISV specifically does not support.) >> >> >>J >>>>>If the vendor's software works in a cluster at all (i.e., they didn't> >>>>>pull some real bone-headed programming mistakes) then anyH >>>>>cluster-related problems are not usually related to their software, >>>>>per se. >>>>>  >>>>>n >>>>>uE >>>>So even the risk of this happening would not stop you proposing a D >>>>cluster without having the vendors apps qualified for a cluster. >>>> >>>>G >>>You have no idea how I develop or make proposals.  Don't jump so far H >>>without knowing what you're talking about.  When I make a proposal orD >>>recommendation, it includes solutions that I know from first-hand >>>experience. >>>c >>>t >>5 >>No I don't but I am beginning to get an idea :):):)  >> >> >>Or rather ):):):): >> >>	 >>Regardse >> >>Andrew Harrisonj >> > 3 > Not speaking for anyone, certainly not DEC/Compaqr/ > (get rid of the xxxz in my address to e-mail)a >    ------------------------------  % Date: Wed, 24 Jul 2002 15:43:01 +0100 U From: Andrew Harrison SUNUK Consultancy <andrew_nospam.harrison_remove_this@sun#.com>m4 Subject: Re: Only 20% drop in VMS systems (was: wow)0 Message-ID: <ahmedg$f8l$1@new-usenet.uk.sun.com>   jlsue wrote:  G > On Thu, 18 Jul 2002 13:02:20 +0100, Andrew Harrison SUNUK Consultancyr6 > <andrew_nospam.harrison_remove_this@sun#.com> wrote: >  >  >> >>jlsue wrote: >> >>I >>>On Wed, 10 Jul 2002 18:05:04 GMT, "Bill Todd" <billtodd@metrocast.net> 	 >>>wrote:t >>>g >>>h >>>n$ >>>>I've in no way asserted that SMPL >>>>*substitutes* for clustering in terms of availability, just that *withinO >>>>whatever availability constraints exist* (e.g., that might require at leastnL >>>>two clustered systems plus whatever additional systems one might need toP >>>>keep performance acceptable should one fail) larger individual cluster nodesI >>>>are usually preferable to greater numbers of smaller cluster nodes.  a >>>> >>>>G >>>Well, I have several points of experience that contradict this view.o< >>>There have been quite a few environments where small, VMSD >>>workstation-based clusters of 15-20 nodes out-performed a single,G >>>large, SMP-based solution, and did it at a much, much lower cost andw >>>higher availability.a >>>pB >>>For example, a VAXstation Model 90, fully configured with 128MBF >>>memory, FDDI card, and unlimited openvms user licenses could be hadI >>>for less than $50K back in the early 1990s.  And the license units foraH >>>layered products was 10 (or 20) units.  This CPU was similar to a VAXH >>>4000-610 system with a license rating of 1200 units and costing about >>>8X as much. >>>tG >>>We'd cluster these together with AlphaServer 3000-800s that 'served'w< >>>the disks to the cluster.  All connected to a gigaswitch. >>>wG >>>Performance was phenomenal, availability was fantastic, and cost wasc >>>amazingly low.  >>>> >>>  >>>: >>< >>Right so what was the workload, you are great at this kind; >>of "reference" which does not have enough information for.@ >>anyone to make an informed judgment but less good at providing >>the full deck. >> > ? > You are so full of shit it keeps coming out of your keyboard.nE > I put the same amount of "detail" into my reference that was in the>
 > quoted one.> >     4 Do you always resort to abuse when you get stuck ???    E > Just to keep you a little more informed, the clustered environmentsn= > include over 250 different applications.  Some of them werek@ > e-mail/messaging, some were scientific & statistics, some were2 > database - both Rdb and Oracle (V6.x varieties). >     - You still seem unable to answer the question.v  ; Were you running a single Oracle DBMS across multiple nodesi0 what was the workload, what was the scalability.  9 Phenominal, fantastic and amazingly low may well be exacti= measures in yuor world but they arn't in most of my customersh2 unless you happen to be talking to a sales person.    B >>And why would license units be a direct measure of performance ? >> >> > C > Hmm... I  was trying to show scalability, cost, and availability.tH > License units refer to cost - and if you're talking about Oracle, this > is a big issue.v >     > Humm, so what you are really saying is that Digitals licensing; system didn't make any sense based on the capability of thee servers.  > Interesting but not in the context of this discussion when the2 subject is scalability of clusters vs SMP systems.   Regards.   Andrew Harriasoo   ------------------------------  + Date: Wed, 24 Jul 2002 10:12:54 -0700 (PDT) . From: Fabio Cardoso <fabiopenvms@yahoo.com.br> Subject: OpenVMS Jobs in Europel@ Message-ID: <20020724171254.74296.qmail@web20209.mail.yahoo.com>  
 Dear Sirs,  6 As my contract is almost over here in Brazil (Set, 30)3 and I dont want to renew (after 3 years is good to r6 breath new air), if someone in Europe needs an OpenVMS  5 SysAdmin, and depending on I accept to be downgraded >5 to a Senior Operator, I am available... Event to 6 - e 12 months contracts...   Preferences for:   Germany  Italy 
 Nederlands Portugal$ Scandinavian Countries:   may be !!! Sweden Spains United Kingdom   No Balkans please... f   Regards    FC a     =====y ========================== Fbio dos Santos Cardoso OpenVMS System Manager Rio de Janeiro - Brazil  fabiopenvms@yahoo.com.br ==========================  2 __________________________________________________ Do You Yahoo!?( Yahoo! Health - Feel better, live better http://health.yahoo.comn   ------------------------------  % Date: Wed, 24 Jul 2002 19:22:39 +0200t- From: Didier Morandi <Didier.Morandi@Free.fr> # Subject: Re: OpenVMS Jobs in EuropeS' Message-ID: <3D3EE25E.C29932BC@Free.fr>s  0 Don't send me your resum, I have it already :-)   D.   Fabio Cardoso wrote: >  > Dear Sirs, > 8 > As my contract is almost over here in Brazil (Set, 30)4 > and I dont want to renew (after 3 years is good to8 > breath new air), if someone in Europe needs an OpenVMS > 6 > SysAdmin, and depending on I accept to be downgraded6 > to a Senior Operator, I am available... Event to 6 - > 12 months contracts...   ------------------------------   Date: 24 Jul 2002 13:48:55 GMT) From: wkb@freebie.xs4all.nl (Wilko Bulte)s> Subject: Re: PW600au problem (probably h/w) - long and tedious? Message-ID: <3d3eb047$0$12292$e4fe514c@dreader4.news.xs4all.nl>   V In <3D39DC9A.4B6220F1@pressenter.com> Lyndon Bartels <lbartels@pressenter.com> writes:   >Mark Daniel wrote:  >> s" >> Thanks for the input Christoph. >> t' >> I have changed the disks previously.f, >> There is currently nothing on SCSI bus A. >>  I >> > dkb0.0.0.1010.0            DKB0                          RZ29B  0016sI >> > dkb100.1.0.1010.0          DKB100                        RZ28D  0010gI >> > dqb0.0.0.207.0             DQB0        TOSHIBA CD-ROM XM-6202B  1110l$ >> > dva0.0.0.0.1               DVA0I >> > mkb600.6.0.1010.0          MKB600                        TLZ09  0172 C >> > ewa0.0.0.3.0               EWA0              00-00-F8-76-1C-F6 I >> > pka0.7.0.1004.0            PKA0                  SCSI Bus ID 7  5.57eC >> > pkb0.7.0.1010.0            PKB0                  SCSI Bus ID 7 C >> > pqa0.0.0.107.0             PQA0                       PCI EIDEtC >> > pqb0.0.0.207.0             PQB0                       PCI EIDEe >> vK >> SCSI bus A shows errors on it (the KLAATU$PKA0 device - which is the VMS>H >> device for SCSI controller - I think) even without any devices on it.H >> When there was I would get an error log entry saying the system diska% >> (DKA0) had gone into mount verify.e >> eE >> Is there any way to disable the on-board SCSI controller I wonder?r >>      D >I didn't think the PWSs had onboard SCSI... onboad IDE yes, but not >SCSI.  , The GL variant of the Miata has onboard SCSI   Wilko> --' |   / o / /_  _   				wilko@FreeBSD.orgC2 |/|/ / / /(  (_)  Bulte				Arnhem, the Netherlands   ------------------------------  % Date: Wed, 24 Jul 2002 11:42:31 +0100l( From: Nic Clews <sendspamhere@127.0.0.1>) Subject: Re: Questions about license paks ) Message-ID: <3D3E8497.AA846D6F@127.0.0.1>t   Stuart Johnson wrote:c >  lH > 1) Can I sell the license paks? I've seen licenses for sale on eBay. IK > remember that, in the "old" days at least, certain licenses could be sold>I > with a machine and others (layered products) coundn't. I haven't a clues' > about selling the license paks alone.s  F To add to Larry's response. We at CSC administer systems for ourselvesF and on clients behalf, maintaining, upgrading, installing, supporting.> We are therefore duty bound to do it by the rules, and also toB understand it. Legal teams and experts negotiate deals and explain4 things to 'us', keeping ourselves and clients clean.  E The most important thing to realize about a software licence is it isaE effectively a right to use NOT a right to copy or distribute, and the E date and TYPE of the licence strictly dictate what the licence allowsiG you to do, in regards to what version, who for, on what system and whatdH your responsibilities are as a licensee. Just because you've been handedG some bit of paper, or some strings of characters typed into the licencex3 utility 'work', does not necessarily make it legal.n  E Strictly, whomever you got that box from broke the licence agreement. F The *only* licence you should have is the base licence (and I mean theH paper), which travels with the hardware. Certain shrink wrapped licencesC (the pathworks ones may be examples) can optionally be transferred,kE subject to their terms and conditions. ALl other licences should have ? been removed and deleted. They are the intellectual property offF Digital/Compaq/HP. There is a process called 'novation' where licencesE can be transferred from one party to another, via the licensor. Everyu# single licence issue can be traced.n  C VMS and subcomponents are good because of the investment in skilled>C engineers, licensing and support finance that investment, providingiG newer and better versions. You are getting what you pay for, but if youeA stop paying for it, we'll all stop getting it. OK there's more tooE economics than this, but anyone reading this and trading illegally ondH ebay should at least have a guilty conscience, and not be surprised whenG taken to court. Anyone else has hopefully had a quick heads up on beings legal.  C *Disclaimer* The Terms & Conditions that you got with your (legallyr/ obtained) licence will give you the full story.t -- >? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciencest nclews at csc dot coma   ------------------------------  % Date: Wed, 24 Jul 2002 13:06:27 +0100e* From: "Richard Brodie" <R.Brodie@rl.ac.uk>) Subject: Re: Questions about license paksd+ Message-ID: <ahm57a$gdq@newton.cc.rl.ac.uk>i  Y "Nic Clews" <sendspamhere@127.0.0.1> wrote in message news:3D3E8497.AA846D6F@127.0.0.1...t  G > The most important thing to realize about a software licence is it isrG > effectively a right to use NOT a right to copy or distribute, and thesG > date and TYPE of the licence strictly dictate what the licence allowsiI > you to do, in regards to what version, who for, on what system and whats* > your responsibilities are as a licensee.  K Subject to local law, of course; software licences are notorious for havinguK clauses that are unenforcable. However, I agree that reading the licence isgG a much better starting point than assuming whatever you can technicallyo do is permitted.   ------------------------------   Date: 24 Jul 2002 13:52:08 GMT) From: wkb@freebie.xs4all.nl (Wilko Bulte)z5 Subject: Re: Raid Array 7000 firmware update questionI? Message-ID: <3d3eb108$0$12302$e4fe514c@dreader4.news.xs4all.nl>   X In <yoXSza1NiXFE@malvm7.mala.bc.ca.> nothome@spammers.are.scum (Malcolm Dunnett) writes:    , > I've just acquired a used Raid Array 7000.  - > It currently has HSOF 7.0 firmware on it, I * >understand 7.3(?) is the current version.  4 > What are the implication of using the 7.0 firmware4 > in a VMS cluster ( VMS 7.2-2, disk array on shared0 >SCSI bus with two hosts )? Alternately, what is/ >the proper method of getting updated firmware?s. >Is it available for download or does one have >to purchase it?  3 It is not free. But I *think* HSZ70 has gone EOL so > how easy it would be to get a later HSOF version I don't lnow.  1 >The unit has dual controllers - do I need to buy , >two platform kits so that I can get two new0 >program cards ( I already have 2 program cards,' >but with the older firmware on them) ?>  / Fieldservice should be able to get loose cards.y   W/   --' |   / o / /_  _   				wilko@FreeBSD.orgo2 |/|/ / / /(  (_)  Bulte				Arnhem, the Netherlands   ------------------------------  % Date: Wed, 24 Jul 2002 08:25:34 +0200t: From: Karl Rohwedder <extern.karl.rohwedder@volkswagen.de> Subject: Re: RDB Questions/ Message-ID: <ahlh84$6ia4@doiweb4.volkswagen.de>d  * Alan Winston - SSRL Admin Cmptg Mgr wrote: > In article <20020723084440.5808.qmail@web21003.mail.yahoo.com>, =?iso-8859-1?q?Tadimeti=20Keshav?= <keshav_tadimeti@yahoo.co.uk> writes: >  >>Hello all,0 >>I guess COBOL programs containing embedded SQL7 >>statements requires Oracle CDD without which it won't  >>compile. Is this correct?l >>8 >>Re: JCC's RDB list, I had set it to DIGEST. Ever since% >>I haven't been recieving any mails.v >>5 >>TDMS: Where can I find documents related to the nowy8 >>discontinued TDMS? Does Anyone have any old documents? >  > K > Really, don't get involved with TDMS if you don't have to.  If you reallyx! > need forms, how about DECForms?m > I > Are you on VAX or Alpha?  TDMS didn't even get ported to Alpha, and thea5 > emulator (from PRAXA) isn't available to hobbyists.s > 	 > -- Alani > Q > ===============================================================================y2 >  Alan Winston --- WINSTON@SSRL.SLAC.STANFORD.EDUO >  Disclaimer: I speak only for myself, not SLAC or SSRL   Phone:  650/926-3056lO >  Physical mail to: SSRL -- SLAC BIN 69, PO BOX 4349, STANFORD, CA  94309-0210hQ > ===============================================================================x > . I'e just read, that TDMS will make it to IA64.   -- n  - mit freundlichen Gruessen | with best regards.   Karl Rohwedder               mC iT-Ingenieurteam     | Ellernbruch 11       | D-38112 Braunschweig wA Telefon: 0531/515521 | Telefax: 0531/515531 | Mobil: 0172/5434843 H  E-Mail: rohwedder(at)decus.decus.de        | iT-IngTeam(at)t-online.de .          karl.rohwedder(at)it-ingenieurteam.de DATEX-P: 4505018005::ROHWEDDER   ------------------------------  % Date: Wed, 24 Jul 2002 08:56:52 -0400>0 From: "Syltrem" <syltremspammenot@videotron.com>' Subject: Re: Single Column text output?l4 Message-ID: <ctx%8.6946$H67.36836@tor-nn1.netcom.ca>   Could you post a
 $ DIR/FULL and a> $ DUMP/REC=COUNT=5 of the spooled file?  I Are other files displayed normally, in the editor (could be your editor's 	 settings)d   --   SyltremeI http://pages.infinit.net/syltrem (OpenVMS related web site - en franais)l> To reply to myself directly, remove .spammenot from my address  F "Kiasu Surfer" <verykiasu@hotpop.com> a crit dans le message de news:- gn6sjug2aki5fplce93dqm5pr7tdm9liv8@4ax.com...uL > I had a weird experience in spooling a log file from my DCL script calling > Oracle scripts.s > B > When I use TYPE to list out, the spooled text displayed as norm.H > But when I EDIT it, the visual display show the spooled text as single	 character ! > in single column vertical down.p >e
 > Should be:-s- > ORACLE Server Manager... blah blah blah....  >  > In Edit mode:- > Oh > Ry > Ap > Cd > Lo > Eg >.. > Can someone help me to rectify this problem.	 > Thanks.f
 > Regards, >e > Kevin Lai0 >> > **************** > ** SPAM BLOCK ** > ****************A > REPLACE "verykiasu" with "iamverykiasu" at verykiasu@hotpop.comn   ------------------------------  % Date: Wed, 24 Jul 2002 11:05:55 -0400o- From: JF Mezei <jfmezei.spamnot@videotron.ca>m' Subject: Re: Single Column text output?m, Message-ID: <3D3EC251.83A16D86@videotron.ca>   Kiasu Surfer wrote:eB > When I use TYPE to list out, the spooled text displayed as norm.R > But when I EDIT it, the visual display show the spooled text as single character! > in single column vertical down.a    L You have to look at the file structure of the created file first (DIR/FULL).M What probably happens is that the file is created with attributes that do not K match the file's contents. You can use SET FILE/ATTRIB to fix those. ( HELPc has much information on this,   M You can then use DUMP to look at the raw file contents to see if there is anyeN special characters between characters which would cause the editor to think itN is an end of line. If there are no special characters between letters, then it$ would have to be the file structure.  K For instance, if the file were created with a fixed record size of 1 and noeK carriage control attributes, then edit would behave as you described, whilesM type would simply dump the raw contents to the screen and the file would thena appear to be OK.  8 Some of the SET FILE/ATTRIB you way to try are probably: 	RFM:STMLF MRS:0 LRL:0    K If when you edit the file you see little "CR" characters at the end of eachdE line, you can also fiddle with RAT:CR and RAT:NONE in the attributes.    ------------------------------  % Date: Wed, 24 Jul 2002 19:39:40 +0010o% From: paddy.o'brien@zzz.tg.nsw.gov.auuM Subject: Re: Size of the comp.os.vms readership ? , was: RE: OpenVMS webring:e5 Message-ID: <01KKHGKKFBBM000N2T@tgmail.tg.nsw.gov.au>l   Rob Buxton wrote:   D >comp.os.vms is certainly a very active newsgroup. Actual readershipF >would be difficult to quantify. I certainly agree with the quality of7 >postings, but you do see the same, not so many, names.  >>E >I know there's probably a lot like me who only access the news groupsC >say once a day and only contribute when there's an issue that theyn* >believe they can offer something new on.   8 Last paragraph, I agree entirely, I am of a similar ilk.  E Last bit of the first paragraph, not so sure about the "not so many".m  O Since I have been subscribed to Info-VAX, "big" names have come and gone.  The rN late CJL, and people like Jerry Leichter, Richard Gilbert and Pat Rankin come O to mind from my early years.  All very technical, which is often lacking these nA days.  And of course Hoff has been around since Jurassic Park :-)u  O We currently have a new crop of names with a large correspondence quota.  But,  F there are many who have cropped up periodically over the years (in no N particular order, Brian Tillman, Bob Koehler, Larry Kilgallen come to mind -- N mainly because I have seen posts from them today).  Most of the more prolific N correspondents in these later years are less interested in the technicalities N of VMS, but in positive or negative doom-saying and who's got the biggest and N best equipment :-)  I can readily understand the reasons that c.o.v. has gone 	 this way.r  L But there are also a lot of names that I would never recognise nor probably O see again.  They "lurk" and appear when they have problems, or a speciality to aL offer.  I believe there are a lot like this, and I know quite a few here in  Sydney.h  H It is in that sense that I see your reference to regular names as being + somewhat misleading of the population here.    Regards, Paddy   ------------------------------    Date: 24 Jul 2002 06:53:53 -0700% From: Alan Greig <a.greig@virgin.net>eM Subject: Re: Size of the comp.os.vms readership ? , was: RE: OpenVMS webring:>) Message-ID: <ahmbhh01s68@drn.newsguy.com>o  5 In article <01KKHGKKFBBM000N2T@tgmail.tg.nsw.gov.au>,u' paddy.o'brien@zzz.tg.nsw.gov.au says...vP >Since I have been subscribed to Info-VAX, "big" names have come and gone.  The O >late CJL, and people like Jerry Leichter, Richard Gilbert and Pat Rankin come    L Hey, you can't mention Carl without Ehud! Wonder what he's up to these days.   Search Result 2 7 From: Ehud Gavron 602-570-2546 (GAVRON@IT.SUNQUEST.COM)sN Subject: Carl and Ehud are told to be good or the world will end (news at 10)  Newsgroups: comp.os.vmseB View: (This is the only article in this thread) | Original Format  Date: 1992-12-02 07:10:05 PST   s  > From:	SMTP%"GAVRON@ALPHA.SUNQUEST.COM" 25-NOV-1992 18:12:54.28
 To:	GAVRON CC:	 Subj:	Re: Carl/Ehud       . Date:    Wed, 25 Nov 1992 18:12:46 -0700 (MST)E From:    GAVRON@ALPHA.SUNQUEST.COM (Ehud Gavron 602-570-2000 x. 2546) 6 Message-Id: <921125181246.32a10aa8@ALPHA.SUNQUEST.COM> Subject: Re: Carl/Ehud4 To:      ccdarg@dct.ac.uk, gavron@vesta.sunquest.com% X-Vmsmail-To: SMTP%"ccdarg@dct.ac.uk"n  : In article <1992Nov25.094629.1785@dct.ac.uk>, you write...E #know what you're doing to future employment prospects for yourselvesw, #but I very much doubt that it's positive...  9 	That's because you're still thinking that being employeda; 	is a favor someone does you.  In fact, it's a favor you dor> 	them that you're willing to solve all their problems for less 	then all their profit.s  ? 	There are many employers who consider this much more importantn< 	than pissing small net.people off.  I wouldn't work for any 	other.E  & #Stop frightening people from posting.   	I don't think so. #  #-- D #Alan Greig                            Janet: Alan@UK.AC.DUNDEE-TECH; #Dundee Institute of Technology	   Internet: Alan@DCT.AC.UKe7 #Tel: (0382) 308810                 Int: +44 382 308810w6 #              -- There is only one true conspiracy --  e 	Ehud   C -- Ehud Gavron        (EG76)      gavron@spades.aces.com   ------------------------------    Date: 24 Jul 2002 09:45:14 -0600+ From: young_r@encompasserve.org (Rob Young) M Subject: Re: Size of the comp.os.vms readership ? , was: RE: OpenVMS webring:t3 Message-ID: <GMRPPiK3iNjE@eisner.encompasserve.org>n  Q In article <ahmbhh01s68@drn.newsguy.com>, Alan Greig <a.greig@virgin.net> writes:i7 > In article <01KKHGKKFBBM000N2T@tgmail.tg.nsw.gov.au>,V) > paddy.o'brien@zzz.tg.nsw.gov.au says...2Q >>Since I have been subscribed to Info-VAX, "big" names have come and gone.  The  P >>late CJL, and people like Jerry Leichter, Richard Gilbert and Pat Rankin come  > N > Hey, you can't mention Carl without Ehud! Wonder what he's up to these days. >  > Search Result 2 9 > From: Ehud Gavron 602-570-2546 (GAVRON@IT.SUNQUEST.COM) P > Subject: Carl and Ehud are told to be good or the world will end (news at 10)  > Newsgroups: comp.os.vms-D > View: (This is the only article in this thread) | Original Format   > Date: 1992-12-02 07:10:05 PST  >  : > @ > From:	SMTP%"GAVRON@ALPHA.SUNQUEST.COM" 25-NOV-1992 18:12:54.28 > To:	GAVRON > CC:	 > Subj:	Re: Carl/Ehud. >  >  > 0 > Date:    Wed, 25 Nov 1992 18:12:46 -0700 (MST)G > From:    GAVRON@ALPHA.SUNQUEST.COM (Ehud Gavron 602-570-2000 x. 2546) 8 > Message-Id: <921125181246.32a10aa8@ALPHA.SUNQUEST.COM> > Subject: Re: Carl/Ehud6 > To:      ccdarg@dct.ac.uk, gavron@vesta.sunquest.com' > X-Vmsmail-To: SMTP%"ccdarg@dct.ac.uk"  > < > In article <1992Nov25.094629.1785@dct.ac.uk>, you write...G > #know what you're doing to future employment prospects for yourselves . > #but I very much doubt that it's positive... > ; > 	That's because you're still thinking that being employed = > 	is a favor someone does you.  In fact, it's a favor you do @ > 	them that you're willing to solve all their problems for less > 	then all their profit.C > A > 	There are many employers who consider this much more important > > 	than pissing small net.people off.  I wouldn't work for any	 > 	other.  > ( > #Stop frightening people from posting. >  > 	I don't think so. > #  > #-- F > #Alan Greig                            Janet: Alan@UK.AC.DUNDEE-TECH= > #Dundee Institute of Technology	   Internet: Alan@DCT.AC.UKn9 > #Tel: (0382) 308810                 Int: +44 382 308810g8 > #              -- There is only one true conspiracy -- >  e > 	Ehud> >  > > --  > Ehud Gavron        (EG76)      > gavron@spades.aces.com >   ? 	Ehud is legendary in several different directions.  One of thee@ 	brightest to grace this forum, painful in a Carl Lydick kind of= 	way.... that said... here is a little known fact about Ehud:/  7 http://www.wired.com/news/business/0,1367,19285,00.html-  N A Venezuelan online casino doled out more than US$1 million to buy Wall Street -- dot com. N Players Sportsbook and Casino, based on tiny Margarita Island off the coast ofN Venezuela, had the US$1.03-million winning bid in an auction for the rights to the wallstreet.com domain.    F "I'm shocked that this kind of value exists for a domain name, but I'mN overjoyed, too," said Ehud Gavron, owner of the flashy Web address since 1994.M Gavron, the 32-year-old president of a Tucson, Arizona-based Internet servicen8 provider, will split the loot with two other co-owners.   @ 	You can do various net searches for Ehud Gavron and follow what1 	he is up to if interested.  He was/is in Tucson.s   				Robs   ------------------------------  % Date: Wed, 24 Jul 2002 11:22:37 -0400'- From: JF Mezei <jfmezei.spamnot@videotron.ca>oM Subject: Re: Size of the comp.os.vms readership ? , was: RE: OpenVMS webring: , Message-ID: <3D3EC63A.C2C006BF@videotron.ca>  M Actual size may not be so important, but trend might be. Perhaps the owner ofiN the info-vax mailing list might be able to give a ballpark estimate of whetherI the list size has remained more or less the same or whether it would haveb& grown or shrunk (just a bit or a lot).  N Of course, there may be a significant number who migrated from the list to theE comp.os.vms newsgroup, so the loss of a subscriber of infovax may not  represent the loss of a reader.a   ------------------------------  % Date: Wed, 24 Jul 2002 15:36:03 +0200w- From: Didier Morandi <Didier.Morandi@Free.fr>w& Subject: Software Tools Clearinghouse?' Message-ID: <3D3EAD42.1707A102@Free.fr>s  K We are probably a large number of us having tools we developped one day fory special purposes.a  O Today, there are a certain number of sites gathering these tools, and also somecN 1600 bpi tapes and TKs here and there, plus the DECUS tapes, plus the Freeware CDs, plus..s  T What do you think about creating a WEB page indexing (easily) all these useful data?  I This STC used to exist within DEC internally. Everyone was able to uploadeP worldwide via DECnet a piece of code with some documentation. A retrieval systemN was very powerful to find back some coding examples on request (where are you, Lynn Yarbrough?).s  P Maybe this idea could be developped and, why not, the whole DEC STC lib could be made public.   Your advice?   D. -- e2   ------------------------------------------------2 MORANDI Consultants  http://Didier.Morandi.Free.fr0   19 chemin de la Butte, 31400 Toulouse, France.2 Tel.: +33 (0)6 7983 6418 - Fax: +33 (0)5 6154 19282 OpenVMS, APPLE, Computer Security, Migration plans2 --------------------------------------------------   ------------------------------  % Date: Wed, 24 Jul 2002 10:10:21 -0400r! From: Jim Agnew <jpagnew@vcu.edu>o* Subject: Re: Software Tools Clearinghouse?' Message-ID: <3D3EB54D.A70B02B5@vcu.edu>a  H I *like* this...  I really, really LIKE this...  I'll email you a cup of coffee..   jimt   Didier Morandi wrote:  > M > We are probably a large number of us having tools we developped one day fort > special purposes.a > Q > Today, there are a certain number of sites gathering these tools, and also someeP > 1600 bpi tapes and TKs here and there, plus the DECUS tapes, plus the Freeware
 > CDs, plus..e > V > What do you think about creating a WEB page indexing (easily) all these useful data? > K > This STC used to exist within DEC internally. Everyone was able to uploadeR > worldwide via DECnet a piece of code with some documentation. A retrieval systemP > was very powerful to find back some coding examples on request (where are you, > Lynn Yarbrough?).e > R > Maybe this idea could be developped and, why not, the whole DEC STC lib could be > made public. >  > Your advice? >  > D. > --4 >   ------------------------------------------------4 > MORANDI Consultants  http://Didier.Morandi.Free.fr2 >   19 chemin de la Butte, 31400 Toulouse, France.4 > Tel.: +33 (0)6 7983 6418 - Fax: +33 (0)5 6154 19284 > OpenVMS, APPLE, Computer Security, Migration plans4 > --------------------------------------------------   ------------------------------  % Date: Wed, 24 Jul 2002 22:19:46 +0900h& From: "David Lee" <phongle@kornet.net>" Subject: status tapedrive flashing+ Message-ID: <ahm87q$s0p$1@news1.kornet.net>   L I have an ES-40 computer with the tape drive unit attached to it.  Recently,	 I noticedlJ that the status light on this tape drive flashing slowly.  I tried to hard7 power itdown and bring it up, but it still won't clear.hJ I tried to clean it with the cleaning tape several timesbut still won't go away.  I am running VMS 7.2.1c  ) Does anyone know how to clear this light?k   Thanks   ------------------------------  % Date: Wed, 24 Jul 2002 10:29:48 -0400-- From: Jonathan Boswell <jsb@ost.cdrh.fda.gov>3& Subject: Re: status tapedrive flashing0 Message-ID: <3D3EB9DC.C814E6F7@ost.cdrh.fda.gov>   David Lee wrote:C > I have an ES-40 computer with the tape drive unit attached to it.t   Which tape drive?  A DDS3 DAT?  L > I tried to clean it with the cleaning tape several timesbut still won't go > away.t  M A good cleaning tape of the appropriate type should fix it.  Eventually thesenL reach end of life and must be discarded.  Also, I note that my ES40s shippedO with DDS4 cleaning tapes even though my Compaq SDT-9000 drives are DDS3.  ThesesD cleaning tapes do indeed clear the slowly flashing status indicator.   ------------------------------   Date: 24 Jul 2002 12:07 CDTo' From: carl@gerg.tamu.edu (Carl Perkins)l& Subject: Re: status tapedrive flashing- Message-ID: <24JUL200212072599@gerg.tamu.edu>o  1 Jonathan Boswell <jsb@ost.cdrh.fda.gov> writes...D }David Lee wrote:pD }> I have an ES-40 computer with the tape drive unit attached to it. }  }Which tape drive?  A DDS3 DAT?= } M }> I tried to clean it with the cleaning tape several timesbut still won't goa }> away. } N }A good cleaning tape of the appropriate type should fix it.  Eventually theseM }reach end of life and must be discarded.  Also, I note that my ES40s shippedBP }with DDS4 cleaning tapes even though my Compaq SDT-9000 drives are DDS3.  TheseE }cleaning tapes do indeed clear the slowly flashing status indicator.   E Huh? As far as I can tell, a DDS cleaning cartridge is a DDS celaning.F cartridge. They don't seem to make DDSx varieties for each generation.  D (Seagate is apparently the exception. They sell one that they say isC useable for their DDS4 and earlier drives, part number 91301M, that D is different in some unspecified way than the one they sold for DDS3D and earlier drives, part number 91301S which they warn against using? in their DDS4 drive, but they seem to be the only ones who do.)   A If you put in the cleaning cartridge and it is ejected just aboutpD immediatly then it has hit the end of the cleaning tape and you needB a new one (the tape is never rewound - it runs a little of it eachC time). A real cleaning run should keep the tape in the drive for atr, least several seconds - maybe as long as 30.   --- Carl   ------------------------------  % Date: Wed, 24 Jul 2002 09:33:53 +0200e2 From: "Frits A.M. Storms" <frits@storms.tmfweb.nl>$ Subject: Re: Terminal input from DCL? Message-ID: <3d3e58d7$0$94891$e4fe514c@dreader3.news.xs4all.nl>8  4 <paddy.o'brien@zzz.tg.nsw.gov.au> schreef in bericht/ news:01KKG1S49S0Y000M6D@tgmail.tg.nsw.gov.au...p > Didier Morandi wrote:P >oE > >This raises a different issue. DCL is not a high-level programmingh	 language.s > ItG > >cannot perform ASTs (except the read/timeout/err), it cannot performo > >QIO-nowait,K > >not computer GOTOs. If this automatic validation field feature is really-K > >requested, yes it has to be programmed with a 3rd level language (or FMSi :-)b > >  > >D.s > >  > >"David J. Dachtera" wrote:  > >../..H > >> My point was that without being able to trap individual keystrokes,L > >> there's no practical way to limit the count of characters entered so as; > >> to limit entry into a field to a predetermined length.e > >>E > >> You can do that in BASIC with INKEY$, but DCL has no equivalent.- > / > I'm sure I've mentioned this before, but ....w >eL > Although I've written much QIO interface code within my applications, manyL > moons ago I found a nice little macro package (compiles on Alpha too) thatJ > provided a simple interface to several of my command files, and saved me* > re-inventing the wheel for that purpose. >p/ > http://vms.process.com/fileserv-software.htmla >iI > Look for the ASK package, and courtesy of Hunter Goatley, for those whot don'th< > know this site, browse through some of the other packages. >D > Regards, PaddyL Have been using ASK for years, very useful (wrote a menuscript based on it),I but the limitations of it (it cannot perform single character input) haver" driven me to starting this thread. regards, Fritsl   ------------------------------  % Date: Wed, 24 Jul 2002 19:09:55 +0010w% From: paddy.o'brien@zzz.tg.nsw.gov.aua$ Subject: Re: Terminal input from DCL5 Message-ID: <01KKHFIOT50I000M9U@tgmail.tg.nsw.gov.au>    Frits,  M >Have been using ASK for years, very useful (wrote a menuscript based on it),hJ >but the limitations of it (it cannot perform single character input) have# >driven me to starting this thread.t    . I trust we are talking about the same program.  M Years ago before we got involved with CMS and friends, all our source was in -O manifold .TLBs.  I massively rewrote a .COM file that was being used to handle n$ them when I first started work here.  N I used ASK for the help key (F15) and Q to quit.  Various others that I would O have to revisit the .COM file that has not been used since early '90s (--1990s e that is :-)s  N Without revisiting, my memory is that ASK returned a single key which I would L parse in DCL.  Perhaps it just returned codes for function keys and not the $ standard Alphameric keyboard keys.    K Whatever, I found it a handy little no-re-inventing tool to use in several e command procedures.n   Regards, Paddy   ------------------------------   Date: 24 Jul 2002 07:07 CDT ' From: carl@gerg.tamu.edu (Carl Perkins)iR Subject: Re: usage of LIB$M_FIS_MIXEDCASE in the lib$find_image_symbol RTL routine- Message-ID: <24JUL200207075969@gerg.tamu.edu>T  2 ksubrama1@yahoo.com (Kannan Viswanathan) writes...A }In my code , I am using the lib$find_image_symbol call to invokeeE }functions in a shared image. The modules in the shared image as well A }as the code that invokes  are compiled with switches "as_is" and 2 }"shortened" and hence want to turn on the flag inC }lib$find_image_symbol which will invoke symbols without converting( }them to uppercase.d } ) }Here is my code segment to achieve this.  }..t/ }unsigned longword flags = LIB$M_FIS_MIXEDCASE;  }..t@ }status = LIB$FIND_IMAGE_SYMBOL(&log_dsc, &func_dsc, &retval, 0,	 }&flags);e } B }This is on alpha/openvms 7.3 and the entire code is compiled with }pointer=64. } @ }The above segment doesn't seem to work. When I step through the5 }debugger, the message I get at the point of invoking  }lib$find_image_symbol is  } # }%LIB-F-INVARG, invalid argument(s)o } G }When I change the flags argument to "0", the RTL routine  succeeds buti< }upcases the symbol by default which is not I want to happen } 7 }I will greatly appreciate your experience in using theh# }LIB$M_FIS_MIXEDCASE for this call.m }  }Thanks and Regards, }Kannan.  I I'm not certain of this, but it may be complaining that the 4th parameter H is not valid. The last two are optional, but to specify the last one youG may need to specify the next to last one as a valid value. "0" is not aiF valid descriptor address. You might try making that "&log_dsc" too, orE something appropraite. When you change the last one to "0" it works -sH not because "0" is a valid address to find the flags at, but because "0"H is what is used by these things to represent "the optional parameter was not specified".e   --- Carl   ------------------------------  % Date: Wed, 24 Jul 2002 15:01:43 +0100eU From: Andrew Harrison SUNUK Consultancy <andrew_nospam.harrison_remove_this@sun#.com>  Subject: Re: VMS commitmento0 Message-ID: <ahmc02$eie$1@new-usenet.uk.sun.com>   Main, Kerry wrote:  	 > Andrew,  > ! > Re: buying new ERP / CRM pkgs -a > I > <<< At the interfaces level 10 years ago customers would roll their own"B > integration/messaging now there is a much greater use of EAI/ETL > products instead.>>> > B > While there will continue to be a market for these products, theH > previous hype for these large pkgs curing world hunger has dropped off > with most cust's now.r    B I am always more interested in the numbers than what Analysts say.  < SAP posted flat revenues for the last half year, pretty good: with the economic downturn and not evidence that companies have stopped spending.  ? Spending on Oracle applications is also flat 2002 vs 2001 againa< given the current economic downturn not really evidence that- customers are deserting ERP/CRM in a big way.y    J > From Gartner (since many Cust's like to quote these guys): July 11, 2002G > "The enterprise application software markets, which saw unprecedented F > growth in the late 1990s, have come back to reality in 2001. This isH > because of the economy and the CFO/CIO focus on doing more with the ITF > purchases of the past than in putting in new systems. And the futureF > looks just as tight for 2002 for the worldwide customer relationshipF > management (CRM), supply chain management (SCM), enterprise resource, > planning (ERP) and collaborative markets." >     C Well as I said despite the downturn both Oracle and SAP have posted-@ results for their respective businesses which are basically flatB for 2002 vs 2001, given a ~30% decline in IT expenditure over this6 period their relative performance is quite reasonable.   Regards  Andrew Harrisonn    	 > Regardse >  > Kerry Main > Senior Consultante > Hewlett-Packard Canada# > Consulting & Integration Servicesc > Voice: 613-592-4660t > Fax   : 613-591-4477 > Email: Kerry.Main@hp.com >  >  > -----Original Message-----) > From: Andrew Harrison SUNUK Consultancy 7 > [mailto:andrew_nospam.harrison_remove_this@sun#.com] t > Sent: July 19, 2002 1:15 PMk > To: Info-VAX@Mvb.Saic.Comt > Subject: Re: VMS commitmentd >  >  >  >  > David Froble wrote:e >  > * >>Andrew Harrison SUNUK Consultancy wrote: >> >> >>= >>>Since the mid 90's most big corporates have been following H >>>a strategy of moving from developing their own apps to buying as much >>>o >  >>>as they can.h >>> D >>>The rise in fortunes of Seibel, SAP, Oracle applications etc are ; >>>examples of this effect turning up in revenues to ISV's.  >>>A >> >  > G > It would be wrong just to concentrate on the Applications Suites like6J > SAP and Oracle, customers are also getting out of developing middleware,A > interfaces etc as well. 10 years ago many investment banks used>J > middleware and market data systems that they rolled themselves, now manyJ > have replaced these with packages from vendors like Reuters, Vitria etc. > E > At the interfaces level 10 years ago customers would roll their own B > integration/messaging now there is a much greater use of EAI/ETL > products instead.  > I > One Seibel project I recently worked on also required Vitria, AB InitioeH > and a number of other products to manage the interfaces between SeibelA > and the other applications, bespoke or otherwise running in the  > customers environment. > J > Just having Seibel available on the platform isn't the end of the story,H > applications stacks have become very much more complicated composed ofF > products from multiple vendors all of which need to be available and4 > properly supported on the platform being proposed. > , > OpenVMS also falls at this hurdle as well. > 	 > RegardsT > Andrew Harrison  >  >  > D >>First, not all application packages are the same.  Some are ratherA >>rigid, saying that "you WILL run your business in this manner".  >>	 > Others s > E >>try to allow as much flexibility as possible.  Both scenarios have O5 >>problems, and yes, they do work for some companies.a >>H >>If you're running a business where you and everyone else in that type  >>ofC >>business do the same thing with no hope of increasing your markete >> > share  > > >>by doing things in a better manner, AND, there is a software >> > application  > F >>that fits your business rather well, then packaged applications will >> > be $ >  >>a winner for you.i >>H >>If your business requires you to 'do it just a bit better' in order toE >>be successful, then a rigid package just won't do, and some of the oH >>flexible packages are rather complex (and expensive) to install, use, 
 >>and modify.  >>E >>Then there are the examples of custom home grown applications that t >>giveC >>such a bad reputation.  All the terms and ship via codes are hardE >>	 > coded, f > C >>along with custom logic for each one.  Modification and changing sE >>business needs are expensive, inflexible, and serve to provide job 0  >>security for a bunch of hacks. >>A >>In the middle are the small ISVs that produce applications for  
 >>verticalH >>markets, in-house developed applications that are well designed, well G >>implemented, and flexible, the packages that get adopted by in-house  @ >>people who add company specific business requirements, and the >> > packages t > H >>that get modified and maintained by the ISV or other outside technical >> >  >>help.  >>G >>The last group of applications are particularly suitable for VMS witht& >>it's robust development environment. >>F >>So why do some companies attempt to move to the SAPs and such, when  >>it'sG >>not really a good fit for them?  Too many reasons, some of them with   >>some validity. >>D >>1) The clueless bean counter who finally advanced to a position ofF >>authority figures that he can cut costs to enhance profits, and ends >> > up   > - >>cutting out the muscle and keeping the fat.a >>G >>2)The 'big ego' IT staff that perpetuates the custom 'everything hardoF >>coded' systems that serve the IT staff much better than they do the  >>business needs., >>F >>3) The managers that listen to the 'Authur Andersons', who are much  >>moreH >>interested in selling $250/hr SAP consultants to the customer, than in >> >  >>the customer's well being. >>
 >><many more>t >> >>Dave >> >  >    ------------------------------  % Date: Wed, 24 Jul 2002 12:16:47 -0400q- From: JF Mezei <jfmezei.spamnot@videotron.ca>b Subject: Re: VMS commitmente, Message-ID: <3D3ED2E8.1832B819@videotron.ca>  ( Andrew Harrison SUNUK Consultancy wrote:A > SAP posted flat revenues for the last half year, pretty good...a> > Spending on Oracle applications is also flat 2002 vs 2001...  G Ahh, but is that spending for new sites, or spending for maintenance ofp existing sites ?  H What would be interesting metric is the number of ACTIVE sites running a particular package or OS.-  M The 400,000 VMS sites is important to prevent HP from killing VMS. But we allvL know that this number if meaningless because it isn't qualified, so it couldS mean 400,000 licences issued, or 400,000 systems still under maintenance contracts.   K A while back Lotus Notes was the rage. What is happening to it these days ?gG A while back, Novell was the rage, what is happening to it these days ?s   ------------------------------   End of INFO-VAX 2002.405 ************************