1 INFO-VAX	Sat, 03 Jun 2000	Volume 2000 : Issue 308       Contents:' RE: Actual vs Theoretical Network Speed ' Re: Actual vs Theoretical Network Speed  Re: Comments on ABS v3.0% Computing internet address for a user  DEcnet plus to DEcnet Phase IV" Re: DEcnet plus to DEcnet Phase IVP DIFF /IGNORE=(SPACING) questiion - was sent before with wrong	subjectline, sorryP Re: DIFF /IGNORE=(SPACING) questiion - was sent before with wrong subjectline, s0 DTSESSION crashes for some users, but not others Re: future of PMDF and VMS Re: future of PMDF and VMS Re: future of PMDF and VMS Re: future of PMDF and VMS Re: Migrating from VMS to AIX  Re: Migrating from VMS to AIX , Re: None dare call it Legacy (ISE goes live), Re: None dare call it Legacy (ISE goes live), Re: None dare call it Legacy (ISE goes live)A Re: OpenVMS commentaries (was Re: Gartner commentary on Wildfire) 1 OpenVMS Date Change software for application only * Re: OpenVMS vs Tru64 Pathworks performance* Re: OpenVMS vs Tru64 Pathworks performance Pipe throughput, sigh, slow..." Re: Pipe throughput, sigh, slow..." Re: Pipe throughput, sigh, slow..." Re: Pipe throughput, sigh, slow...0 Re: Please help a beginner with puzzling code...0 Re: Please help a beginner with puzzling code...0 Re: Please help a beginner with puzzling code...3 Re: POS/credit card verification with OpenVMS Alpha 3 Re: POS/credit card verification with OpenVMS Alpha  Re: SImple graphs fro VMS  Tadpole Tuning Re: Tadpole Tuning Tadpole Tuning- Re: Tru64 poised to open source DLM, AFS, LVM  Re: UCX$POP_SERVER problems  Re: VAX on Intel?  Re: VAX on Intel?  Re: VAX on Intel?  Re: VAX VMS 7.2 Bug? Re: VAX VMS 7.2 Bug? Re: VAX VMS 7.2 Bug? Re: VEST / DECmigrate  Re: VEST / DECmigrate 7 Re: VMS File Caching Futures (Was: Re: Andrew whatever) 7 Re: VMS File Caching Futures (Was: Re: Andrew whatever) 7 Re: VMS File Caching Futures (Was: Re: Andrew whatever) 7 Re: VMS File Caching Futures (Was: Re: Andrew whatever) 7 Re: VMS File Caching Futures (Was: Re: Andrew whatever) 7 Re: VMS File Caching Futures (Was: Re: Andrew whatever) 7 Re: VMS File Caching Futures (Was: Re: Andrew whatever) 7 Re: VMS File Caching Futures (Was: Re: Andrew whatever) 7 Re: VMS File Caching Futures (Was: Re: Andrew whatever) 4 What is the simplest way to network two VMS systems?8 Re: What is the simplest way to network two VMS systems?8 Re: What is the simplest way to network two VMS systems?8 Re: What is the simplest way to network two VMS systems?? Where to find VXT load image  was - Re: where to find firmware? $ Re: Which is the language of VAX/VMS Re: Wildfire Announcement  RE: Wildfire Announcement  RE: Wildfire Announcement   F ----------------------------------------------------------------------  % Date: Fri, 02 Jun 2000 19:32:42 +0100 8 From: John Macallister <J.Macallister1@physics.ox.ac.uk>0 Subject: RE: Actual vs Theoretical Network SpeedL Message-ID: <B8D7D072032FD411BE7A0090279FA2401ED5AC@ppnt41.physics.ox.ac.uk>  L No special optimisation, just our standard systems. Maxima observed on quiet6 networks or when connected on to ports on same switch.   John  B Name: John B. Macallister  E-mail: j.macallister1@physics.ox.ac.ukH Post: Nuclear and Astrophysics Laboratory, Keble Road, Oxford OX1 3RH,UKA Phone: +44-1865-273388 (direct)  273333 (reception)  273418 (Fax)      -----Original Message-----8 From: Gord Coulman [mailto:nospam_gcoulman@ccinet.ab.ca]# Sent: Friday, June 02, 2000 6:37 PM  To: John Macallister0 Subject: Re: Actual vs Theoretical Network Speed    C John Macallister <J.Macallister1@physics.ox.ac.uk> wrote in message F news:B8D7D072032FD411BE7A0090279FA2401ED5AB@ppnt41.physics.ox.ac.uk...J > We see up to around 60 Mbits/sec here on 100 Mbits/sec switched EthernetC > with UNIX and PC boxes but much less than that with VMS: around 6 
 Mbits/sec.D > These figures are observed maxima i.e. tending towards the figures expected > on dedicated systems.   J John, that 60 Mbits/sec is more like what I expected to get.  What kind ofA equipment are you using?  Have you done any kind of optimization?    Gord.    ------------------------------  % Date: Fri, 02 Jun 2000 16:36:06 -0700 + From: "Barry Treahy, Jr." <treahy@mmaz.com> 0 Subject: Re: Actual vs Theoretical Network Speed( Message-ID: <393844E6.3F5F85F3@mmaz.com>  J Actually I think I had a error in mine too, is not a true megabyte 2^21 or 1048576 bytes?  K Regardless, if you are on a 100Mb network, you're having problems.  Are you J running half or full duplex and are you sure that both ends are configuredO correctly?  I know it is apples and oranges, but I have an Alpha running DU 4.0 N and I could not let the NIC autonegoiate, it would always get it wrong and the performance was terrible.   $ What about CRC or packet loss?  Any?   Barry    Gord Coulman wrote:   D > Barry, yes the first column is MegaBytes.  Here's how I calculated
 > throughput:  > ) > 7 Megabytes X 8 bits/byte = 56 Megabits . > 56 Megabits / 4 seconds = 14 Megabits/second > J > Perhaps I oversimplified things.  How did you get 17.92 Megabits/second? > 6 > Barry Treahy, Jr. <treahy@mmaz.com> wrote in message# > news:3937E712.9805C8A@mmaz.com... N > > Gord, first assuming that when you state MB, you really mean a megabyte of > dataM > > and not one million bytes and second of all, dod you really mean megabits  > under N > > the speed column?  Assuming MB=megabutes and Mbit is megabits, I calculate > out K > > the 4 second transfer at 17.92 megabits per second.  The second file at  > 18.77 N > > Mb/sec and 16.13 Mb/sec.  I'm assuming that you are running on a 100 Mb TP= > > ethernet?  Are you seeing collisions, CRC or packet loss?  > J > I don't have any network diagnostic equipment and wasn't near the hub toN > watch the blinky lights.  The network operates normally, but I am seeking to > find out:  > , > a.  Am I getting the bandwidth I paid for?2 > b.  If not, is there anything I can do about it? >  > Gord.  >  > > 	 > > Barry  > >  > > Gord Coulman wrote:  > > K > > > Searching for a network bottleneck the other day, I transferred a few  > large @ > > > files and measured how long it took.  Here are my results: > > >  > > > Size      Time    Speed ! > > > (MB)    (sec)    (Mbit/sec)  > > > 7            4        14 > > > 22          12      14.7 > > > 241        153    12.6 > > > I > > > This was between a couple of W98 PCs on a 100Mbit switched LAN over  > CAT5. H > > > The switches are 3Com Switch 3300's and the NICs are Intel 10/100. > > > K > > > My question:  I know that actual network throughput is always smaller  > thatH > > > theoretical, but how much smaller?  What is a reasonable number to	 > expect?  > > >  > > > Gord Coulman > >  > > -- > > C > > Barry Treahy, Jr  *  Midwest Microwave  *  Vice President & CIO  > > E > > E-mail: Treahy@mmaz.com * Phone: 480/314-1320 * FAX: 480/661-7028  > >  > >  > >    --  ? Barry Treahy, Jr  *  Midwest Microwave  *  Vice President & CIO   A E-mail: Treahy@mmaz.com * Phone: 480/314-1320 * FAX: 480/661-7028    ------------------------------  # Date: Fri, 02 Jun 2000 18:49:53 GMT  From: itjck01@my-deja.com ! Subject: Re: Comments on ABS v3.0 ) Message-ID: <8h8vka$3d5$1@nnrp1.deja.com>   8 In article <8h32he$3db$1@fizban.fizban.pprd.abbott.com>,4   "Dave Gudewicz" <david.gudewicz@abbott.com> wrote:E > Thanks for the comments.  I look forward to your benchmark results.  > 	 > Dave...  >   E I saw about 14gigs/hour/tape drive on TL896 with ABS 3.0b and 2 KZPBA D tape controller pipes on an unloaded AlphaServer 4100 with 2 gigs ofD memory and a fairly well tuned ABS backup account.  With 6 drives weB could get about 80 gigs/hour to tape in compaction mode.  (We have: patch for ABS 3.0b to get compaction/compression working.)  A I also did a test using OpenVMS 7.2-1 Backup with compaction to a H single TZ89 drive of the TL896, and saw pretty much the same throughput.  D I get about 8gigs/hour with a TZ877/KZPAA configuration on this same system.   A I thought the TZ89 drive/controller (TL896) would perform better. E Perhaps my backup accounts (ABS and other) are not tuned as well as I @ thought.  When I get a chance, I will reexamine and do some more testing.  - Are the above throughputs reasonable, anyone?    :) jck   --$ Free personal opinion is what I post    & Sent via Deja.com http://www.deja.com/ Before you buy.    ------------------------------  % Date: Fri, 02 Jun 2000 15:39:43 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> . Subject: Computing internet address for a user, Message-ID: <39380D7D.EA5C7721@videotron.ca>  M I would like an application to "compute" the internet address of a user. (eg: 7 how someone on the internet could reach that VMS user).   G Obviously, I can get the vms username and nodename on which the user is M running the program.  The program will have a configuration file, so it could $ potentially contain the domain name.  L If I were simply to have:  <username>@<node>.<domain_name>  would this cover all cases ?   M  If I would to allow <username>@<domain_name> as well, would that cover those   cases not covered by the above ?  < Is there a way to programatically get a cluster alias name ?   ------------------------------  % Date: Sat, 03 Jun 2000 01:51:37 -0600 " From: Ed Nath <enath@nps.navy.mil>' Subject: DEcnet plus to DEcnet Phase IV @ Message-ID: <3.0.6.32.20000603015137.0127d360@mail.nps.navy.mil>  D How do I deinstall decnet plus and go back to decnet phase IV...I am/ running a vax 4000 with version 7.1 of open VMS @ **************************************************************** Ed Nath > Lab Manager, Secure Computing and Simulation Laboratory (SCSL) U. S. Naval Postgraduate School  Code: GSC/WG 1411 Cunningham Road, Room 239 Monterey, Ca 93933   (831)656-3014, DSN 878-4014    ------------------------------  % Date: Fri, 02 Jun 2000 16:59:54 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> + Subject: Re: DEcnet plus to DEcnet Phase IV , Message-ID: <39382042.CDB726A2@videotron.ca>   Ed Nath wrote: > F > How do I deinstall decnet plus and go back to decnet phase IV...I am1 > running a vax 4000 with version 7.1 of open VMS   P To upgrade from DECNET PLUS to DECNET-4, you need to deinstall DECNET PLUS first  6 I beleive STOP/NETWORK DECNET if I remember correctly,  J then use the PRODUCT SHOW PRODUCT to get a list of installed products, and then   PRODUCT REMOVE <product name>   . (can't remember the exact name of DECNET PLUS)  D You then use VMSinstall to install the DECNET-4 kit. I did that on aN worktation and it worked quite well. Of course, you will need to reconfigure aK lot of stuff after, but since DECNET-4 is so much easier to manage for most  networks, it is not a problem.   ------------------------------  % Date: Fri, 02 Jun 2000 17:48:42 -0400   From: norm.raphael@jamesbury.comY Subject: DIFF /IGNORE=(SPACING) questiion - was sent before with wrong	subjectline, sorry 4 Message-ID: <C22568F2.0076F63B.00@jklh21.valmet.com>  @ I have a text file that differs from its predecessor (mostly) in@ that the old one has lines that start with a single space, while> in the new one, that space has been edited out.  Nevertheless,= when I do $DIFF/IGNORE=(SPACING) newfile.txt;<cr> the leading ' space in the new file is _not_ ignored.    Is this intended behavior?  ; Is there a switch to get DIFF to see them as matched lines?    Here is the help file entry.   DIFFERENCES   	   /IGNORE            /IGNORE=(keyword[,...])   B      Inhibits the comparison of the specified characters, strings,A      or records; also controls whether the comparison records are D      output to the listing file as edited records or exactly as theyE      appeared in the input file. If you specify only one keyword, you E      can omit the parentheses. The keyword parameter refers to either C      a character or a keyword. The first set of keywords determines E      what, if anything, is ignored during file comparison; the second E      set of keywords determines whether or not ignored characters are E      included in the output. The following keywords are valid options       for the /IGNORE qualifier:   "      Keyword          Item Ignored   [snip]  C      SPACING          Extra blank spaces or tabs within data lines.    [snip]   -Norm    ------------------------------  # Date: Fri, 02 Jun 2000 22:29:38 GMT  From: mlyczewski@ddpwa.comY Subject: Re: DIFF /IGNORE=(SPACING) questiion - was sent before with wrong subjectline, s 4 Message-ID: <39383511.105722793@client.news.psi.net>  @ This is the symbol I have set up for doing diffs on COBOL files.F It takes care of spaces... I suspect that your problem is not leading, but trailing spaces.    @   DIFC == "DIFFERENCES/COMMENT_DELIMITER=* /IGNORE=(BLANK_LINES,$ COMMENTS, SPACING, TRAILING_SPACES)"  E On Fri, 02 Jun 2000 17:48:42 -0400, norm.raphael@jamesbury.com wrote:    >  >  > A >I have a text file that differs from its predecessor (mostly) in A >that the old one has lines that start with a single space, while ? >in the new one, that space has been edited out.  Nevertheless, > >when I do $DIFF/IGNORE=(SPACING) newfile.txt;<cr> the leading( >space in the new file is _not_ ignored. >  >Is this intended behavior?  > < >Is there a switch to get DIFF to see them as matched lines? >  >Here is the help file entry.  >  >DIFFERENCES > 
 >  /IGNORE >   >        /IGNORE=(keyword[,...]) > C >     Inhibits the comparison of the specified characters, strings, B >     or records; also controls whether the comparison records areE >     output to the listing file as edited records or exactly as they F >     appeared in the input file. If you specify only one keyword, youF >     can omit the parentheses. The keyword parameter refers to eitherD >     a character or a keyword. The first set of keywords determinesF >     what, if anything, is ignored during file comparison; the secondF >     set of keywords determines whether or not ignored characters areF >     included in the output. The following keywords are valid options  >     for the /IGNORE qualifier: > # >     Keyword          Item Ignored  >  >[snip]  > D >     SPACING          Extra blank spaces or tabs within data lines. >  >[snip]  >  >-Norm >  >  >    ------------------------------  % Date: Fri, 02 Jun 2000 16:55:56 -0500 0 From: James Bellinger <jnb@hep.physics.wisc.edu>9 Subject: DTSESSION crashes for some users, but not others 4 Message-ID: <39382D6C.27C9BE7D@hep.physics.wisc.edu>  . We have a few AXP machines running VMS 6.1 and DecWindows Motif 1.2-4. , Some users can log into the console and work) with no trouble under the CDE system, but - one of our users has the window manager crash  immediately.  , I checked his login.com file and it exits if) the process is DTSESSION, so that doesn't  seem to be a factor.  * At the moment a DECW$TE_0140 process seems1 to be taking an inordinate amount of CPU as well,  if that means anything.   . (Yes, I deleted the old DECW*.* files and blew1 away his DT directory.  Didn't help, same trouble  when he logged in again.)   , 	Can someone point me to what might be going wrong? 	Thanks   . .............................................. From DECW$SM.LOG I find:  0 Executing mcr cde$system_defaults:[bin]dtsession  ; %SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual  address=00000000007D& 2000, PC=FFFFFFFF80408F0C, PS=0000001B  2   Improperly handled condition, image exit forced.1     Signal arguments:   Number = 0000000000000005 1                         Name   = 000000000000000C 1                                  0000000000010000 1                                  00000000007D2000 1                                  FFFFFFFF80408F0C 1                                  000000000000001B        Register dump:9     R0  = FFFFFFFFFFFFFFF3  R1  = 00000000007D2000  R2  =  000000007B1EF1709     R3  = 00000000006738B8  R4  = 000000007B22C008  R5  =  00000000006738B09     R6  = 0000000000000048  R7  = 0000000000000000  R8  =  00000000007BA7609     R9  = 0000000000000010  R10 = 0000000000000010  R11 =  00000000000000409     R12 = 00000000007CD368  R13 = 000000007B99C000  R14 =  00000000000031009     R15 = 0000000000000000  R16 = 000000002E60C088  R17 =  00000000006738F89     R18 = 0000000000000033  R19 = 0000000000000000  R20 =  00000000007CD3689     R21 = 000000002E6238DB  R22 = 00000000007BAAC8  R23 =  00000000007D1FFC9     R24 = 00000000007BA758  R25 = 00000000007D2000  R26 =  000000002E60C07B9     R27 = 00000000007BA790  R28 = 0000000000000004  R29 =  000000007AFD20609     SP  = 000000007AFD2060  PC  = FFFFFFFF80408F0C  PS  =  200000000000001B   Exiting XSESSION.COM   --/ James N. Bellinger      U.Wisconsin at Madison  ! jnb@hep.physics.wisc.edu.antispam    ------------------------------  $ Date: Fri, 2 Jun 2000 13:33:04 -05001 From: "Dave Gudewicz" <david.gudewicz@abbott.com> # Subject: Re: future of PMDF and VMS 8 Message-ID: <8h8ueg$75v$1@fizban.fizban.pprd.abbott.com>  H I can't answer your questions specifically, but I can say that Compaq isK looking into this situation.  I also can't predict the future, but it would J make a whole lot of sense for Compaq to make a bid on the PMDF product for VMS and get it under its wing.   Hope this helps a little.    Dave...   0 <system@niuhep.physics.niu.edu> wrote in message% news:8h8r1t$l63$1@husk.cso.niu.edu...  > Hello,= >     I am seriously considering purchasing PMDF-MTA for VMS.  > G >     While Sun may drop support for PMDF on VMS at some point is there F > any reason to believe that the current version of PMDF will cease to& > function for future versions of VMS? > G >     My understanding is that PMDF is a pretty solid product and given H > that VMS maintains pretty decent backward compatability this shouldn't > be a problem.  > C >     Can anybody see a problem with this?  Is IMAP due for a major 3 > change that would make the PMDF product obsolete?  > A >     I am currently running IUPOP3 and MX 5.1.  I need the IMAP4 B > capabilities that PMDF has.  Is it possible to just use the IMAPB > facilities and to continue, at least temporarily with IUPOP3 and > MX?  > ( > Thank you for any advice you can give. >  > Robert > Morphis@physics.niu.edu 8 > Real Women change tires abuse@uu.net postmaster@uu.net9 > Real Men change diapers                 security@uu.net    ------------------------------  $ Date: Fri, 2 Jun 2000 22:51:30 +0400+ From: "Ruslan R. Laishev" <laishev@dls.net> # Subject: Re: future of PMDF and VMS ? Message-ID: <aiTZ4.17099$XX4.273439@news-east.usenetserver.com>    Hi !  A     Why you don't want to bay PSC or TCPWare/Multinet with IMAP4?   . system@niuhep.physics.niu.edu wrote in message" <8h8r1t$l63$1@husk.cso.niu.edu>... >Hello, < >    I am seriously considering purchasing PMDF-MTA for VMS. > F >    While Sun may drop support for PMDF on VMS at some point is thereE >any reason to believe that the current version of PMDF will cease to % >function for future versions of VMS?  > F >    My understanding is that PMDF is a pretty solid product and givenG >that VMS maintains pretty decent backward compatability this shouldn't  >be a problem. > B >    Can anybody see a problem with this?  Is IMAP due for a major2 >change that would make the PMDF product obsolete? > @ >    I am currently running IUPOP3 and MX 5.1.  I need the IMAP4A >capabilities that PMDF has.  Is it possible to just use the IMAP A >facilities and to continue, at least temporarily with IUPOP3 and  >MX? > ' >Thank you for any advice you can give.  >  >Robert  >Morphis@physics.niu.edu7 >Real Women change tires abuse@uu.net postmaster@uu.net 8 >Real Men change diapers                 security@uu.net   --I + C U, SysMan at DLS ...................................................+ G  http://www.radiusvms.com         |             Cel:  +7 (901) 971-3222 G  http://www.levitte.org/~rlaishev |             Fax:  +7 (812) 115-1035 I + Flying by Su-27........................................ Frying on VMS +    ------------------------------  % Date: Fri, 02 Jun 2000 15:30:04 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> # Subject: Re: future of PMDF and VMS , Message-ID: <39380B3A.1D5C5CD0@videotron.ca>  $ system@niuhep.physics.niu.edu wrote:A >     I am currently running IUPOP3 and MX 5.1.  I need the IMAP4 B > capabilities that PMDF has.  Is it possible to just use the IMAPB > facilities and to continue, at least temporarily with IUPOP3 and > MX?   K If you are unconfortable with PMDF's future, you may want to look at Office H Server (ALL-IN-1) with provides character  cell, POP IMAP as well as WEB access to your emails.  1 The SPD is at http://www.digital.com/info/SP6149/    ------------------------------   Date: 2 Jun 2000 19:38:38 GMT # From: system@niuhep.physics.niu.edu # Subject: Re: future of PMDF and VMS + Message-ID: <8h92fu$3mh$1@husk.cso.niu.edu>   - "Ruslan R. Laishev" <laishev@dls.net> writes:  >Hi !  > # >    Why you don't want to bay PSC    . Never heard of it.  Can you give me more info?    >or TCPWare/Multinet with IMAP4?                                 > I think cost.  PMDF for my purposes costs $1200 (+1000 if I go with the security option)   ? complexity: I have 5 Alphas and a Vax, I would just as soon run ' the same tcpip software on all of them.   ! Thanks for the suggestion though.    Robert Morphis@physics.niu.edu 8 Real Women change tires			abuse@uu.net postmaster@uu.net7 Real Men change diapers                 security@uu.net    ------------------------------  $ Date: Fri, 2 Jun 2000 13:03:00 -0500- From: "Jon Power" <jon.power@sector7XXXX.com> & Subject: Re: Migrating from VMS to AIX8 Message-ID: <WBSZ4.161302$VR.2630487@news5.giganews.com>  I None that I would recommend (although I may be somewhat biased). It makes L sense to use the same vender for RMS and DCL and there is cross play betweenG them. Our RMS uses the exact same FAB/RAB/XAB/NAM interface as VMS. The K Filename fields can be specified with VMS filespecs. logical:[dir]name.ext. H For the name translation to convert on the fly between VMS spec and UNIXI spec it must tie into the logical name handler which although can be used E standalone (through UNIX utilities to get/set logical name) is easier  through DCL.   What language are you porting ?   J If "C/C++" the FAB/RAB structures will work. If FORTRAN you may also weantJ to consider using our VX/FPT which converts VMS fortran to UNIX/NT fortranJ and also convers the embedded indexed file read/write routines to internalF functions giving you RMS indexed/relative funtionality within fortran.   /Jon    7 "David A Froble" <davef@tsoft-inc.com> wrote in message ' news:3934BB21.EFF362C4@tsoft-inc.com...  > maztang@my-deja.com wrote: > > J > > We are looking into converting our one remaining VMS system to AIX forH > > ease of support and consistency amongst our mid-range servers.  I'veI > > found Sector7's software for running RMS and DCL on unix systems, and K > > Boston Business Computing's VCL and EDT+.  Are there any other products I > > out there along these lines?  RMS and DCL are what we really need, to I > > make the transition as smooth as possible.  We'd like to have as many D > > choices as possible to test before we make a commitment with one6 > > vendor.  Any help you all can give would be great. > >  > > Thanks.  > >  > > Jay  > > * > > Sent via Deja.com http://www.deja.com/ > > Before you buy.  > E > Any help?  Ok.  If you really want ease of support and consistancy,  convert your > AIX systems to VMS.  > 0 > An answer you'd have to expect on comp.os.vms. >  > Dave >  > --6 > David Froble                       Tel: 724-529-04506 > Dave Froble Enterprises, Inc.      Fax: 724-529-0596= > 170 Grimplin Road               E-Mail: davef@tsoft-inc.com  > Vanderbilt, PA  15486    ------------------------------  # Date: Fri, 02 Jun 2000 19:53:56 GMT ! From: jim@accelr8.com (Jim Reiss) & Subject: Re: Migrating from VMS to AIX; Message-ID: <ohUZ4.22844$5k2.42398@dfw-read.news.verio.net>   , Jon Power <jon.power@sector7XXXX.com> wrote:J >None that I would recommend (although I may be somewhat biased). It makesM >sense to use the same vender for RMS and DCL and there is cross play between H >them. Our RMS uses the exact same FAB/RAB/XAB/NAM interface as VMS. TheL >Filename fields can be specified with VMS filespecs. logical:[dir]name.ext.I >For the name translation to convert on the fly between VMS spec and UNIX J >spec it must tie into the logical name handler which although can be usedF >standalone (through UNIX utilities to get/set logical name) is easier
 >through DCL. J Perhaps a bit biased, considering the same description would apply equally well to Accelr8's toolset. --I  Jim Reiss (jim@accelr8.com)    Accelr8 Technology, Denver, Colorado, USA    ------------------------------  % Date: Fri, 02 Jun 2000 15:22:57 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> 5 Subject: Re: None dare call it Legacy (ISE goes live) , Message-ID: <39380990.864A0423@videotron.ca>   Bill Todd wrote: > N > Sorry for the confusion.  Of course Rdb uses the underlying file system, butF > in a manner that by-passes all of the issues relating to its defaultN > performance deficiencies compared to Unix's (and hence in a manner that made% > Rob's comment entirely irrelevant).    Excuse me ?   N RMS provides optional services above and beyond the basic unix file system. IfN you create the file properly, you can get the raw data without having to fightG againt RMS. What you are saying is the same as someone complaining that K editing a document stored as an ISAM files on an IBM mainframe is very long  and tedious.  H So, RDB creates a raw file with its own structures inside. But a simplerM application can rely on the RMS system to give it access to records. And that " is a function unavailable on Unix.  * Check out what the RMS acronym stands for.   ------------------------------  $ Date: Fri, 2 Jun 2000 17:00:16 -0400' From: "Bill Todd" <billtodd@foo.mv.com> 5 Subject: Re: None dare call it Legacy (ISE goes live) ( Message-ID: <8h972s$opo$1@pyrite.mv.net>  G Perhaps you should become acquainted with the content of a conversation ! before making an ass of yourself.   G I worked on RMS (mostly RMS-11, though in fairly close cooperation with F VAX/RMS, and I was pleasantly surprised recently to find out that theyC adopted my design for key compression in Prologue Version 3 largely J unchanged) in one manner or another for most of the 11 years I was at DEC,K most of that time as its lead (and often sole) access methods developer, so L I suspect that I understand it rather better than you do.  Not that this hasJ much relevance to the subject in question, which was imported from anotherJ thread discussing VMS's relatively poor file system performance (includingK RMS, since that's how most applications access the file system) for typical F applications that just accept default settings compared with Unix fileH system performance performance for similar applications that just accept default settings.    - bill  8 JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message& news:39380990.864A0423@videotron.ca... > Bill Todd wrote: > > L > > Sorry for the confusion.  Of course Rdb uses the underlying file system, but H > > in a manner that by-passes all of the issues relating to its defaultK > > performance deficiencies compared to Unix's (and hence in a manner that  made' > > Rob's comment entirely irrelevant).  > 
 > Excuse me ?  > E > RMS provides optional services above and beyond the basic unix file 
 system. IfJ > you create the file properly, you can get the raw data without having to fight I > againt RMS. What you are saying is the same as someone complaining that H > editing a document stored as an ISAM files on an IBM mainframe is very long > and tedious. > J > So, RDB creates a raw file with its own structures inside. But a simplerJ > application can rely on the RMS system to give it access to records. And that$ > is a function unavailable on Unix. > , > Check out what the RMS acronym stands for.   ------------------------------  # Date: Sat, 03 Jun 2000 01:20:04 GMT + From: Jonas Lindholm <jlindholm@nyc.rr.com> 5 Subject: Re: None dare call it Legacy (ISE goes live) * Message-ID: <39385D02.C6EB80B7@nyc.rr.com>   OK, here is the facts:  N The OMex Click Exchange System is running of OpenVMS and use Oracle Rdb as the% database engine on the backend nodes. M The backend nodes run in a cluster, normaly 2 nodes at each site, This is the 2 normal configuration we tell the customers to use.  Q The other OpenVMS nodes are running as single nodes. These nodes are used for the 2 Market Place, MP, the MP nodes are matching ordersQ from traders. There are also Transaction Routers, TR, nodes that is for RTR. OMex ; use RTR for all transactions from the frontends to backend.   J There are also nodes to receive external informations such OPRA and NASDAQ	 messages.   R All frontends are connected to the backend through Network Gateways, NG,, WNT, and Tunnel Servers, TS, also WNT.   R The NG's will transform incoming trades to RTR transactions and send it to the RTR routers.O The response back is using broadcasts, depending of OMex version but the latest 4 version is using IP broadcasts. These broadcasts areP then transfered to the NG by the TS. The TS will filtered the broadcasts so each( NG only receive the broadcasts it needs.  < These is a very very short description of the functionality.   /Jonas Lindholm 
 OM Technology  Consultant at ISE            JF Mezei wrote:    > Rob Young wrote: > > P > >         Came across this article about the ISE in this week's ComputerWorld: > > > > > http://www.computerworld.com/home/print.nsf/all/000529E3FA >  > No mention of VMS. > O > > http://www.success-stories.compaq.com/css/cgi-bin/cssextusr/s=display/i=690  >  > Nobody sees those.   ------------------------------  % Date: Fri, 02 Jun 2000 15:13:57 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> J Subject: Re: OpenVMS commentaries (was Re: Gartner commentary on Wildfire), Message-ID: <39380774.64C0C329@videotron.ca>  ! steven.reece@quintiles.com wrote: Q > excellent stuff that they're doing, there are significant numbers of people out M > there in "User Land" that won't believe that there's any committment to VMS Q > anyway and it's all just a big show by Compaq before they pull the rug out from  > under all of their customers.     A That is not the way I am seing it. The way I have seen it is that L Digital/Compaq are fully committed to maintaining VMS and continue improvingM it for more time than Microsoft is willing to comming to Windows NT. However, M Decpaq lacks the marketing commitment and lacks the right pricing commitment.   F Compaq wants to play it safe by keeping only the richest of customers,E especially the ones locked into VMS and who can afford to pay for the N development costs for a few rich customers. This excludes the vast majority ofN potential customers. The ones who are locked into VMS don't care about lack ofL applications because they either have their own very specialised package, or! their software is built in-house.     F The problem is that due to the lack of marketing commitment by Compaq,N application developpers do not see a reason to start porting their software toI VMS because there is still very indication that VMS is a shrinking market P where only a small niche is being protected, too small to warrant porting costs.  N Compaq blew a perfect opportunity to start pushing VMS at the Wildfire launch.K For instance, if they had Etrade and that new exchange instead of Oracle on M the stand, that would have been much more impressive and attracted a lot more   media attention and credibility.   ------------------------------  # Date: Sat, 03 Jun 2000 01:32:33 GMT $ From: Stephen Cook <smcook@home.com>: Subject: OpenVMS Date Change software for application only( Message-ID: <3938600E.E25D3A8F@home.com>   Hi All,   Z I'm hoping someone can remember the name or location of some software that I saw last year^ (before Y2K).  This software allowed the user to change the date and time that the application8 saw, but did not change the date and time of the system.  X I did not need it at that time, however it has now become necessary, but I can no longer located it.    Any help would be appreciated.  	 Thank You 
 Stephen C.   ------------------------------  % Date: Fri, 02 Jun 2000 19:03:35 +0111 9 From: Mark Iline - Info-VAX account <ivax@meng.ucl.ac.uk> 3 Subject: Re: OpenVMS vs Tru64 Pathworks performance 1 Message-ID: <009EB04D.B782E205.42@meng.ucl.ac.uk>   K From:	MX%"djweath@attglobal.net"  "Dave Weatherall"  1-JUN-2000 16:59:35.35   I >> This isn't an accurate comparison, but installing Office97 off a file    >> service gives something like: >>   >> Elapsed time	Server platform + >> 5 minutes	NTAS 4 (450 MHz Proliant 1600)  >> 2 minutes	ASDU (PW500AU) , >> 10 minutes	PW 6.0 dual CPU DS20, VMS7.1-2  H >            just to get the scenario straight. Are you installing from & > a PW share _to_ a PW shared disk or 5 > from PW to an NT/W98 local disk or something else?    ! Local hard drive on an NT client.   + > And where does that lovely registry live?    Same.    Mark    Mark Iline	system@meng.ucl.ac.uk- Dept Mech Eng, University College, London. UK    			Read at your own risk.    ------------------------------  % Date: Fri, 02 Jun 2000 19:37:05 +0111 9 From: Mark Iline - Info-VAX account <ivax@meng.ucl.ac.uk> 3 Subject: Re: OpenVMS vs Tru64 Pathworks performance 1 Message-ID: <009EB052.651C5695.53@meng.ucl.ac.uk>   @ From:	MX%"mathog@seqaxp.bio.caltech.edu" 31-MAY-2000 21:04:32.88  L > Advanced Server was licensed from somebody else (IBM?) and originally ran  > on their Unix.  L I think it's something like Microsoft gives the sources to AT&T who port it I to Unix, then pass it to Compaq. So the VMS version is a portof teh Unix   code.   4 >>But it's more than that. This stuff (Pathworks) isA >>real slow on reads, and VMS doesn't read disks all that slowly.   L As someone has suggested, if you take an application that expects a caching G file system underneath it, and stick it on a system that 'does as it's  3 told' (eg VMS), it's likely it won't fly that fast.   I Some more figures for interest. PW 6.0b, Dual CPU DS20, Gbit ethernet to  K Cabletron SS600, 100Mbit FD ethernet to Proliant 1600 NT server. NT server  , reads a 100 Mbyte file off PW share *twice*.  F The second time it reads it, it seems to come straight out of a memoryI cache (little/no activity on RA3000 host bus lights). This cache could be I VIOC, or PW cache. We managed to copy the file in 12 seconds, repeatably. J This is 8+ Mbytes per second average (over a ~10 Mbyte/s link, so the peakG rate is constrained.) First copy was slower. We've seen faster for a 10  Mbyte file.   H Also appears that it may be able to sustain comparable rates to several E clients simultaneously. Though this hasn't been adequately confirmed.   ; My loose guess is that PW isn't doing adequate read-aheads.   L However, we've got another weird problem, in that performance seems to slow K down, having been fastest after a re-boot. It seems that non-paged pool is  I getting tied up. I've heard that 6.0c fixed a memory leak problem - this  K may be related. We've also had a couple of CPUSPINWAIT bugchecks (seems to  G be on a spinlock to do with npaged pool allocation - though it's weird  < that the CPU holding the spinlock was at IPL31 at the time.)  C At least I now feel that PW on VMS is at least capable of a decent  L throughput, so it's worth investigating what's going on here. If it is down K to PW not doing an effective read-ahead, it's hard to see what can be done   though.   J BTW, PW refuses to create shares on DECram RAM disks. We thought we'd see ) how it performed off these, but couldn't.          Mark    Mark Iline	system@meng.ucl.ac.uk- Dept Mech Eng, University College, London. UK    			Read at your own risk.    ------------------------------   Date: 2 Jun 2000 18:35:01 GMT 2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)' Subject: Pipe throughput, sigh, slow... , Message-ID: <8h8uol$9og@gap.cco.caltech.edu>  , This is getting to be a little ridiculous...  4  COMPAQ AlphaServer DS10 466 MHz, OpenVMS AXP V7.2-1  DEC C V6.0-001   I I've got a bunch of programs that do "filter" like operations, read from  7 stdin, write to stdout, and so can be stacked up like:    3 $ pipe type filename | filter1 | filter2 | filter3    # When each is run individually like:     $ define/user sys$Input filename  $ define/user sys$output outfile	 $ filter1   H They run as fast as any other program on OpenVMS, at the disk/read writeL limits as controlled by RMS.  However, when I stack them up in a moderately I deep pipe, the whole thing runs like molasses.  At first I thought it was J my code, but now I think that each "|" puts a huge overhead in somewhere.   * Here's a tiny program to demonstrate this:   $ create echo.c @ /* ECHO.C    02-JUN-200, David Mathog, Biology Division, Caltech? Program to demonstrate molasses like behavior of OpenVMS pipes.  */ #include <stdlib.h>  #include <stdio.h> #define MYMAXSTRING 1000000 ! int main(int argc, char *argv[]){    char bigstring[MYMAXSTRING];a   while( fgets(bigstring,MYMAXSTRING,stdin) != NULL){    (void) fprintf(stdout,"%s",bigstring); }  } 	 $ cc echo  $ link echo  $ echo:==$sys$login:echo  C test.pfa is a text file containing 57759 records in streamlf format  and having 7200 blocks.    Here's a procedure to test it:   $ write sys$output "plain"
 $ sho time  $ define/user sys$Input test.pfa $ define/user sys$output nla0: $ echo
 $ sho time $!# $ write sys$output "one level pipe" 
 $ sho time $ pipe echo < test.pfa  > nla0: 
 $ sho time $!# $ write sys$output "two level pipe" 
 $ sho time% $ pipe echo < test.pfa | echo > nla0: 
 $ sho time $!% $ write sys$output "three level pipe" 
 $ sho time+ $ pipe echo < test.pfa | echo| echo > nla0: 
 $ sho time  < Which gives these results (with the default RMS parameters):  9 $ testecho                                        Elapsed  plain     2-JUN-2000 10:55:06:    2-JUN-2000 10:55:07                            00:00:01 one level pipe    2-JUN-2000 10:55:07:    2-JUN-2000 10:55:08                            00:00:01 two level pipe    2-JUN-2000 10:55:08:    2-JUN-2000 10:55:21                            00:00:13 three level pipe    2-JUN-2000 10:55:21:    2-JUN-2000 11:04:05                            00:08:44  G Sho process/topcpu shows none of the processes using very much CPU time 4 and  $ sho sys/sub  midway through a run looks like:  P 000069DC MATHOG_2        RWMBX    4    34339   0 00:00:01.05       350    130  SP 00006DDD MATHOG_3        LEF      4    66877   0 00:00:02.46       322    183  SP 00006DDE MATHOG_4        LEF      5    35111   0 00:00:01.60       351    129  S  A with either the first, or the first and second, usually in RWMBX.   K When I ran the exact same test on Linux on identical hardware (DS10) these  J commands (using my echo program, not the native Linux one, so I gave it a  different name there):   # cc -o eee echo.c # ./eee <test.pfa >/dev/null% # ./eee <test.pfa | ./eee  >/dev/null . # ./eee <test.pfa | ./eee  | ./eee  >/dev/null8 # ./eee <test.pfa | ./eee  | ./eee  | ./eee   >/dev/null2 # ./eee <test.pfa | ./eee  | ./eee  | ./eee   >foo  J all completed too quickly to time.  The linux test really did work too, as   #  diff foo test.pfa   showed no differences.  6 Conclusion - PIPE performance on OpenVMS is horrible.   K Hypothesis - This is the result of a bad interaction between the lack of IO H buffering and context switching among the various subprocesses. That is,J when each subprocess is active it can only move a small chunk of data fromI in to out and then it stalls.  But then every other process on the system H gets a shot.  Net result, when the pipe is long the various subprocessesK end up with almost no CPU time, and even when they get it, they can't move  E much data.   As a result, PIPE runs like a slug for one "|" and like  H flowing tar for two "|".  One suspects that protons might decay before a# three "|" pipe would complete. :-(.          David Mathog mathog@seqaxp.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech  J **************************************************************************J *                                RIP VMS                                 *J **************************************************************************   ------------------------------  $ Date: Fri, 2 Jun 2000 14:45:03 -0400& From: "Syltrem" <syltrem@videotron.ca>+ Subject: Re: Pipe throughput, sigh, slow... 5 Message-ID: <LfTZ4.2319$rr.47190@weber.videotron.net>   I Remember that ever | mean "SPAWN" so if you've got 5 of them you create 5 H subprocesses. This is not efficient at all. It is a lot faster to handleI your own temporary files with /output= although it involves more lines of  code   Syltrem   = David Mathog <mathog@seqaxp.bio.caltech.edu> wrote in message & news:8h8uol$9og@gap.cco.caltech.edu.... > This is getting to be a little ridiculous... > 6 >  COMPAQ AlphaServer DS10 466 MHz, OpenVMS AXP V7.2-1 >  DEC C V6.0-001  > J > I've got a bunch of programs that do "filter" like operations, read from8 > stdin, write to stdout, and so can be stacked up like: > 4 > $ pipe type filename | filter1 | filter2 | filter3 > % > When each is run individually like:  > " > $ define/user sys$Input filename" > $ define/user sys$output outfile > $ filter1  > J > They run as fast as any other program on OpenVMS, at the disk/read writeB > limits as controlled by RMS.  However, when I stack them up in a
 moderatelyK > deep pipe, the whole thing runs like molasses.  At first I thought it was K > my code, but now I think that each "|" puts a huge overhead in somewhere.  > , > Here's a tiny program to demonstrate this: >  > $ create echo.c B > /* ECHO.C    02-JUN-200, David Mathog, Biology Division, CaltechA > Program to demonstrate molasses like behavior of OpenVMS pipes.  > */ > #include <stdlib.h>  > #include <stdio.h> > #define MYMAXSTRING 1000000 # > int main(int argc, char *argv[]){   >   char bigstring[MYMAXSTRING];A >   while( fgets(bigstring,MYMAXSTRING,stdin) != NULL){    (void) ! fprintf(stdout,"%s",bigstring); }  > }  > $ cc echo 
 > $ link echo  > $ echo:==$sys$login:echo > E > test.pfa is a text file containing 57759 records in streamlf format  > and having 7200 blocks.  >   > Here's a procedure to test it: >  > $ write sys$output "plain" > $ sho time" > $ define/user sys$Input test.pfa  > $ define/user sys$output nla0: > $ echo > $ sho time > $!% > $ write sys$output "one level pipe"  > $ sho time! > $ pipe echo < test.pfa  > nla0:  > $ sho time > $!% > $ write sys$output "two level pipe"  > $ sho time' > $ pipe echo < test.pfa | echo > nla0:  > $ sho time > $!' > $ write sys$output "three level pipe"  > $ sho time- > $ pipe echo < test.pfa | echo| echo > nla0:  > $ sho time > > > Which gives these results (with the default RMS parameters): > ; > $ testecho                                        Elapsed  > plain  >    2-JUN-2000 10:55:06< >    2-JUN-2000 10:55:07                            00:00:01 > one level pipe >    2-JUN-2000 10:55:07< >    2-JUN-2000 10:55:08                            00:00:01 > two level pipe >    2-JUN-2000 10:55:08< >    2-JUN-2000 10:55:21                            00:00:13 > three level pipe >    2-JUN-2000 10:55:21< >    2-JUN-2000 11:04:05                            00:08:44 > I > Sho process/topcpu shows none of the processes using very much CPU time 6 > and  $ sho sys/sub  midway through a run looks like: > H > 000069DC MATHOG_2        RWMBX    4    34339   0 00:00:01.05       350 130  SH > 00006DDD MATHOG_3        LEF      4    66877   0 00:00:02.46       322 183  SH > 00006DDE MATHOG_4        LEF      5    35111   0 00:00:01.60       351 129  S > C > with either the first, or the first and second, usually in RWMBX.  > L > When I ran the exact same test on Linux on identical hardware (DS10) theseK > commands (using my echo program, not the native Linux one, so I gave it a  > different name there): >  > # cc -o eee echo.c > # ./eee <test.pfa >/dev/null' > # ./eee <test.pfa | ./eee  >/dev/null 0 > # ./eee <test.pfa | ./eee  | ./eee  >/dev/null: > # ./eee <test.pfa | ./eee  | ./eee  | ./eee   >/dev/null4 > # ./eee <test.pfa | ./eee  | ./eee  | ./eee   >foo > L > all completed too quickly to time.  The linux test really did work too, as >  > #  diff foo test.pfa >  > showed no differences. > 7 > Conclusion - PIPE performance on OpenVMS is horrible.  > J > Hypothesis - This is the result of a bad interaction between the lack of IOJ > buffering and context switching among the various subprocesses. That is,L > when each subprocess is active it can only move a small chunk of data fromK > in to out and then it stalls.  But then every other process on the system J > gets a shot.  Net result, when the pipe is long the various subprocessesL > end up with almost no CPU time, and even when they get it, they can't moveF > much data.   As a result, PIPE runs like a slug for one "|" and likeJ > flowing tar for two "|".  One suspects that protons might decay before a% > three "|" pipe would complete. :-(.  >  >  >  >  > David Mathog > mathog@seqaxp.bio.caltech.edu @ > Manager, sequence analysis facility, biology division, CaltechL > **************************************************************************L > *                                RIP VMS                                 *L > **************************************************************************   ------------------------------   Date: 2 Jun 2000 19:50:13 GMT 2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)+ Subject: Re: Pipe throughput, sigh, slow... , Message-ID: <8h935l$9og@gap.cco.caltech.edu>  ^ In article <LfTZ4.2319$rr.47190@weber.videotron.net>, "Syltrem" <syltrem@videotron.ca> writes:J >Remember that ever | mean "SPAWN" so if you've got 5 of them you create 5I >subprocesses. This is not efficient at all. It is a lot faster to handle J >your own temporary files with /output= although it involves more lines of >code   H The spawn time would be a factor if I was running the same pipe command K over and over.  In this instance though it's a rarely given command and the I IO time between the subprocesses is the limiting factor.  Using temporary H files in the cases where I'm forced to use PIPE is going to be difficultE because the program at the end of the pipe typically expects all of a G multigigabyte database to be fed into it as one big stream of data, all K of the filter programs will exit at the end of their input streams, and the J users do not have enough free disk space to hold a temporary file anywhere3 near the size of the data passing through the pipe.    Regards,   David Mathog mathog@seqaxp.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech  J **************************************************************************J *                                RIP VMS                                 *J **************************************************************************   ------------------------------   Date: 2 Jun 2000 22:17:18 GMT 2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)+ Subject: Re: Pipe throughput, sigh, slow... , Message-ID: <8h9bpe$k1m@gap.cco.caltech.edu>  a In article <8h8uol$9og@gap.cco.caltech.edu>, mathog@seqaxp.bio.caltech.edu (David Mathog) writes:  >three level pipe  >   2-JUN-2000 10:55:21 ; >   2-JUN-2000 11:04:05                            00:08:44    SYSMAN> param use active$ SYSMAN> param set defmbxbufquo 32000$ SYSMAN> param set defmbxmxmsg  32000 SYSMAN> write active  D drops this to 00:02:30.  Oddly, it seems to go slightly faster to a H real file, more like 00:01:50, but the times vary a bit.  I guess it mayF make sense though that it might go to a real file faster because NLA0:  is probably yet another mailbox.   $ copy test.pfa killme3.txt   ! Only takes a bit under 3 seconds.   = I tried increasing both system parametsr to the their maximum ' values of 64000, but doing so got me a:   , $ pipe echo < test.pfa | echo | echo > nla0:, %SYSTEM-F-EXBYTLM, exceeded byte count quota  J So I backed off that _real fast_ before something important came crashing  down.   H I guess an improvement by 3X is better than nothing, but it doesn't seemL like much compared to the 150X difference between the VMS pipe test and the  Linux one :-(.   Regards,   David Mathog mathog@seqaxp.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech     ------------------------------  $ Date: Fri, 2 Jun 2000 11:40:18 -0700- From: "Dann Corbit" <dcorbit@solutionsiq.com> 9 Subject: Re: Please help a beginner with puzzling code... ( Message-ID: <ZpTZ4.1093$R71.1315@client>  ? "JackM" <jacksomwNOjaSPAM@axa.slu.edu.invalid> wrote in message 3 news:04e9bbf8.2cd2b89f@usw-ex0108-062.remarq.com... < > I am starting to try and interpret some legacy code that I; > have inherited and I do not know what the following lines  > are doing: > : > static readonly $DESCRIPTOR(x_inuse    , "DISKQ_INUSE");> > static readonly $DESCRIPTOR(x_permquota, "DISKQ_PERMQUOTA");> > static readonly $DESCRIPTOR(x_overdraft, "DISKQ_OVERDRAFT"); > = > I understand the 'static' and 'readonly' portions but after < > that, what is variable type, variable name, etc.  Any help+ > would be appreciated.  Thanks in advance.   L You want news:comp.os.vms rather than news:comp.lang.c since standard C does not have OpenVMS descriptors.   = The code you are looking at was designed for an OpenVMS host.   I Follow-ups to news:comp.os.vms please.  My brain-dead newsreader does not  offer a set follow-ups option. --0 C-FAQ: http://www.eskimo.com/~scs/C-faq/top.html$  "The C-FAQ Book" ISBN 0-201-84519-91 C.A.P. Newsgroup   http://www.dejanews.com/~c_a_p I C.A.P. FAQ: ftp://38.168.214.175/pub/Chess%20Analysis%20Project%20FAQ.htm    ------------------------------  % Date: Fri, 02 Jun 2000 15:35:16 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> 9 Subject: Re: Please help a beginner with puzzling code... , Message-ID: <39380C72.23AFE5B6@videotron.ca>  < > > static readonly $DESCRIPTOR(x_inuse    , "DISKQ_INUSE");@ > > static readonly $DESCRIPTOR(x_permquota, "DISKQ_PERMQUOTA");@ > > static readonly $DESCRIPTOR(x_overdraft, "DISKQ_OVERDRAFT");    J Find the descrip.h module description where the macro $DESCRIPTOR is defi= ned.   essentially:  J struct dsc$desccriptor_s x_inuse =A0 where xinuse.dsc$a_pointer =3D "DISK=	 Q_INUSE", 5 								xinuse.dsc$w_length =3D strlen("DISKQ_INUSE") J 				and the other two members of the struct set to define a string descri= ptor.   J VMS uses descriptors to pass strings which makes it a language independan= t way J to call system services. It also allows the inclusion of the null charact= er in strings.   ------------------------------  # Date: Fri, 02 Jun 2000 20:24:43 GMT  From: mlyczewski@ddpwa.com9 Subject: Re: Please help a beginner with puzzling code... 3 Message-ID: <39381511.97529957@client.news.psi.net>   + You basically have some C code that returns , 3 values (by descriptor) of system services.- I'm not 100% sure but it looks like the calls  are to get DISK QUOTA values.    Do this: $RUN SYS$SYSTEM:DISKQUOTA   0 then type HELP and poke around in the help file.* VMS help files are usually pretty good and! this might give you more insight.       0 On Fri, 2 Jun 2000 11:40:18 -0700, "Dann Corbit"  <dcorbit@solutionsiq.com> wrote:  @ >"JackM" <jacksomwNOjaSPAM@axa.slu.edu.invalid> wrote in message4 >news:04e9bbf8.2cd2b89f@usw-ex0108-062.remarq.com...= >> I am starting to try and interpret some legacy code that I < >> have inherited and I do not know what the following lines
 >> are doing:  >>; >> static readonly $DESCRIPTOR(x_inuse    , "DISKQ_INUSE"); ? >> static readonly $DESCRIPTOR(x_permquota, "DISKQ_PERMQUOTA"); ? >> static readonly $DESCRIPTOR(x_overdraft, "DISKQ_OVERDRAFT");  >>> >> I understand the 'static' and 'readonly' portions but after= >> that, what is variable type, variable name, etc.  Any help , >> would be appreciated.  Thanks in advance. > M >You want news:comp.os.vms rather than news:comp.lang.c since standard C does  >not have OpenVMS descriptors. > > >The code you are looking at was designed for an OpenVMS host. > J >Follow-ups to news:comp.os.vms please.  My brain-dead newsreader does not >offer a set follow-ups option.  >-- 1 >C-FAQ: http://www.eskimo.com/~scs/C-faq/top.html % > "The C-FAQ Book" ISBN 0-201-84519-9 2 >C.A.P. Newsgroup   http://www.dejanews.com/~c_a_pJ >C.A.P. FAQ: ftp://38.168.214.175/pub/Chess%20Analysis%20Project%20FAQ.htm >  >    ------------------------------  # Date: Fri, 02 Jun 2000 18:28:45 GMT  From: itjck01@my-deja.com < Subject: Re: POS/credit card verification with OpenVMS Alpha) Message-ID: <8h8uc7$2e2$1@nnrp1.deja.com>   ) In article <8h3hf0$gfb$1@nnrp1.deja.com>,    itjck01@my-deja.com wrote:9 > Does anyone know what credit card verification software H > packages/services are being used on OpenVMS Alpha?  I would appreciateG > hearing the good, bad, and ugly of what folks are using in this area.  > E > Our desire is interactive credit card verification (on the order of B > seconds) that has some sort of API that we can interface to.  WeF > believe the following are available for OpenVMS in the way of credit% > card verfication software/services:  > 	 > PAYLINX 
 > ICVERIFY > PAYMENTECH > 8 > Anyone know of any others or have comment on the above > software/services? >  > :) jck > --& > Free personal opinion is what I post > ( > Sent via Deja.com http://www.deja.com/ > Before you buy.  >   F For what it is worth, we decided on PAYLINX for assorted reasons which/ appeared to us be advantageous over the others.    --$ Free personal opinion is what I post    & Sent via Deja.com http://www.deja.com/ Before you buy.    ------------------------------  $ Date: Fri, 2 Jun 2000 22:52:37 +0400+ From: "Ruslan R. Laishev" <laishev@dls.net> < Subject: Re: POS/credit card verification with OpenVMS Alpha? Message-ID: <ejTZ4.17117$XX4.273703@news-east.usenetserver.com>    Hi !  E itjck01@my-deja.com wrote in message <8h8uc7$2e2$1@nnrp1.deja.com>... * >In article <8h3hf0$gfb$1@nnrp1.deja.com>, >  itjck01@my-deja.com wrote: : >> Does anyone know what credit card verification softwareI >> packages/services are being used on OpenVMS Alpha?  I would appreciate H >> hearing the good, bad, and ugly of what folks are using in this area. >>F >> Our desire is interactive credit card verification (on the order ofC >> seconds) that has some sort of API that we can interface to.  We G >> believe the following are available for OpenVMS in the way of credit & >> card verfication software/services: >>
 >> PAYLINX >> ICVERIFY 
 >> PAYMENTECH  >>9 >> Anyone know of any others or have comment on the above  >> software/services?  >>	 >> :) jck  >> --   G >For what it is worth, we decided on PAYLINX for assorted reasons which 0 >appeared to us be advantageous over the others.=     This product run under VMS ? Can you provide any URL(s) ?      --I + C U, SysMan at DLS ...................................................+ G  http://www.radiusvms.com         |             Cel:  +7 (901) 971-3222 G  http://www.levitte.org/~rlaishev |             Fax:  +7 (812) 115-1035 I + Flying by Su-27........................................ Frying on VMS +    ------------------------------   Date: 2 Jun 2000 22:59:57 GMT 2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)" Subject: Re: SImple graphs fro VMS, Message-ID: <8h9e9d$k1m@gap.cco.caltech.edu>  { In article <031b6fd2.b8662d97@usw-ex0106-046.remarq.com>, Mike Price <mike.priceNOmiSPAM@littlewoods.co.uk.invalid> writes: ? >It would be really useful to us if we had something that could > >draw simpel line graphs from VMS - prefferably something that; >could be sent data at intervals (via mailboxs, ip sockets, ; >DecNet etc.) and would produce a graph continually or when  >requested. ; >The graph could be on a VT terminal or a decwindows screen  > : >Does anyone know of anyhthing simple that can do this. We= >definatly don't need 3D, surfaces or anything complex - just 9 >something that can take output from a piece of DCL (or a " >program) and keep a pretty graph. >   K I'm not sure what you mean by "send" a graph, but gnuplot runs on OpenVMS,  + as does GLE.  The latter is available from:   D   http://seqaxp.bio.caltech.edu/pub/SOFTWARE/GLE_3_3H_VMS_ALPHA.ZIP>  9 My copy of gnuplot is very old, but in 1995 it came from:   0   http://axp616.gsi.de:8080/www/vms/gnuplot.html   Regards,   David Mathog mathog@seqaxp.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech     ------------------------------   Date: 2 Jun 2000 19:51:57 GMT / From: morris@iridium.mv.net (Skipper W. Morris)  Subject: Tadpole Tuning ( Message-ID: <8h938t$aek$1@pyrite.mv.net>  > I'm still trying to get my Tadpole (Alphabook) laptop running.  D I hacked the Phase V install so it installs and runs, UCX even runs.M The system even occasionally will make it all the way thru the boot sequence.   H Usually if I set startup = opa0: in sysboot I can bring the whole systemE up a piece at a time.  But if I just try and boot normally the system E just stops part way up.  Any processes logged in just appear to hang, $ and the system console appears hung.  E It really isn't hung.  I can still ping the system over the net.  And H DECnet hello messages are also going out.  It's just that user processes aren't getting any resources.   L The poor little system only has 32 meg of memory.  When it is running almostH half of the processes are swapped out.  I've run autogen about 5-6 timesA trying to get it so it will try and work in a reasonable fashion.   G Only unusual thing I noticed in monitor that the page fault rate ranged  from 70 to 200.   H Any suggestions as to what I should look at to improve the situation? (I3 know, I know... but I *need* Phase V and TCP/IP...)    thanks /Skip Morris   ------------------------------  % Date: Fri, 02 Jun 2000 17:47:30 -0400   From: norm.raphael@jamesbury.com Subject: Re: Tadpole Tuning 4 Message-ID: <C22568F2.0076DA1B.00@jklh21.valmet.com>  @ I have a text file that differs from its predecessor (mostly) in@ that the old one has lines that start with a single space, while> in the new one, that space has been edited out.  Nevertheless,= when I do $DIFF/IGNORE=(SPACING) newfile.txt;<cr> the leading ' space in the new file is _not_ ignored.    Is this intended behavior?  ; Is there a switch to get DIFF to see them as matched lines?    Here is the help file entry.   DIFFERENCES   	   /IGNORE            /IGNORE=(keyword[,...])   B      Inhibits the comparison of the specified characters, strings,A      or records; also controls whether the comparison records are D      output to the listing file as edited records or exactly as theyE      appeared in the input file. If you specify only one keyword, you E      can omit the parentheses. The keyword parameter refers to either C      a character or a keyword. The first set of keywords determines E      what, if anything, is ignored during file comparison; the second E      set of keywords determines whether or not ignored characters are E      included in the output. The following keywords are valid options       for the /IGNORE qualifier:   "      Keyword          Item Ignored   [snip]  C      SPACING          Extra blank spaces or tabs within data lines.    [snip]   -Norm    ------------------------------  % Date: Fri, 02 Jun 2000 23:34:58 -0400 2 From: "Richard B. Gilbert" <DRAGON@compuserve.com> Subject: Tadpole Tuning 7 Message-ID: <200006022335_MC2-A755-BD91@compuserve.com>   J         Can't you add another 32Mb of memory?   Another 64Mb would be sti= ll better.   .         32Mb is sub-minimum for OpenVMS/Alpha.  ) Message text written by Skipper W. Morris ? >I'm still trying to get my Tadpole (Alphabook) laptop running.   D I hacked the Phase V install so it installs and runs, UCX even runs.C The system even occasionally will make it all the way thru the boot 	 sequence.   J Usually if I set startup =3D opa0: in sysboot I can bring the whole syste= m E up a piece at a time.  But if I just try and boot normally the system E just stops part way up.  Any processes logged in just appear to hang, $ and the system console appears hung.  E It really isn't hung.  I can still ping the system over the net.  And H DECnet hello messages are also going out.  It's just that user processes aren't getting any resources.   E The poor little system only has 32 meg of memory.  When it is running  almostH half of the processes are swapped out.  I've run autogen about 5-6 timesA trying to get it so it will try and work in a reasonable fashion.   G Only unusual thing I noticed in monitor that the page fault rate ranged  from 70 to 200.   H Any suggestions as to what I should look at to improve the situation? (I3 know, I know... but I *need* Phase V and TCP/IP...)    thanks /Skip Morris <    ------------------------------   Date: 2 Jun 2000 18:42:11 GMT ) From: paw@coos.dartmouth.edu (Pat Wilson) 6 Subject: Re: Tru64 poised to open source DLM, AFS, LVM2 Message-ID: <8h8v63$h1v$1@merrimack.Dartmouth.EDU>  , young_r@eisner.decus.org (Rob Young) writes:   >Here's one for you:  C >http://www.zdnet.com/eweek/stories/general/0,11011,2580352,00.html   I >"A source close to the discussions said the first component likely to be O >released is the Distributed Lock Manager, which is an integral part of Tru64's I >clustering. Developers said other interesting parts of Tru64 include its I >clustering, Advanced File System and Logical Volume Manager components."   D That, then, is "AdvFS", DECpaq's own bag-on-the-side fs, not AFS, a 7 product of Transarc (a wholly-owned subsidiary of IBM).   B Gosh, if they open source AdvFS and LVM, maybe someone can fix it!     --
 Pat Wilson paw@dartmouth.edu    ------------------------------  # Date: Fri, 02 Jun 2000 21:13:57 GMT  From: mlyczewski@ddpwa.com$ Subject: Re: UCX$POP_SERVER problems4 Message-ID: <393821bd.100774690@client.news.psi.net>  = Well, I tried to keep VMS going here but I need to be able to > have POP3 services and after tons of hours I still can't make A it work so I am giving up. This Alpha is getting VMS taken off it ; and getting Linux, Another small shop falls to the Penguin.     C On Wed, 31 May 2000 19:51:10 GMT, "mark" <bmlyczewskib@bddwpab.com>  wrote:  9 >I am trying to set up a VMS box running 7.1 and UCX 4.1.  > H >When I try to connect to the host using a POP client (running on Win95)9 >the server dies, sending an email to the System account, C >and the UCX$POP_RECV_STARTUP.LOG file contains the following text: K >(I replaced the actual domain name with mydomain for security reasons...ie $ >I haven't locked this poppy up yet) >  >  >  >$ !+ L >$ !             UCX$POP_RECV_STARTUP.COM -- DEC TCP/IP Services for OpenVMS
 >Software,: >$ !                        POP receiver startup procedure >$ !1 >$ !                        COPYRIGHT (C) 1996 BY > >$ !                    DIGITAL EQUIPMENT CORPORATION, MAYNARD= >$ !                     MASSACHUSETTS.  ALL RIGHTS RESERVED.  >$ !K >$ !  THIS SOFTWARE IS FURNISHED UNDER A LICENSE AND MAY BE USED AND COPIED D >$ !  ONLY IN ACCORDANCE WITH THE TERMS OF SUCH LICENSE AND WITH THE
 >INCLUSIONG >$ !  OF THE ABOVE COPYRIGHT NOTICE.  THIS SOFTWARE OR ANY OTHER COPIES J >$ !  THEREOF MAY NOT BE PROVIDED OR OTHERWISE MADE AVAILABLE TO ANY OTHERB >$ !  PERSON.  NO TITLE TO AND OWNERSHIP OF THE SOFTWARE IS HEREBY
 >TRANSFERRED.  >$ !J >$ !  THE INFORMATION IN THIS SOFTWARE IS SUBJECT TO CHANGE WITHOUT NOTICE >ANDB >$ !  SHOULD NOT BE CONSTRUED AS A COMMITMENT BY DIGITAL EQUIPMENT
 >CORPORATION.  >$ !I >$ !  DIGITAL ASSUMES NO RESPONSIBILITY FOR THE USE OR RELIABILITY OF ITS < >$ !  SOFTWARE ON EQUIPMENT THAT IS NOT SUPPLIED BY DIGITAL. >$ ! >$M >!***************************************************************************  >**  >$ ! >$ ! Modifications >$ !# >$ ! 01  Karol Zielonko Feb-29-1996 6 >$ ! Cloned and munged from original IUPOP .com files. >$M >!***************************************************************************  >**  >$ !- 
 >$ SET VERIFY $ >$ ON WARNING   THEN GOTO ERROR_TRAP$ >$ ON CONTROL_Y THEN GOTO FINAL_EXIT >$!  >$!************************  >$! Purge log files  >$!************************  >$! % >$ SAVE=5+2*f$getsyi("CLUSTER_NODES")  >$ PURGE/KEEP=5  >$!************************  >$! Initialize >$!************************  >$! J >$! If a POP postmaster is defined then send mail to them if error occurs.% >$! Otherwise send to SYSTEM account. / >$ PROGRAMMERS = F$TRNLNM("UCX$POP_POSTMASTER") 6 >$ IF PROGRAMMERS .EQS. "" THEN PROGRAMMERS = "SYSTEM"! >$ NODE    = F$GETSYI("NODENAME")  >$! . >$ DEFINE SYS$SCRATCH  SYS$SYSDEVICE:[UCX$POP] >$! 3 >$! Message maximum = 0 means *no* message maximum. # >$ DEFINE UCX$POP_MESSAGE_MAXIMUM 0  >$! ? >$! Let a link remain idle for two mintes before timing it out. 3 >$ DEFINE UCX$POP_LINK_IDLE_TIMEOUT "0 00:02:00.00" M >$!**************************************************************************  >**** K >$! To turn on any of the following options simply uncomment. Remember that C >$! "boolean" logicals are turned on when the logical is defined to 
 >anything. We L >$! only set them to "1" by convention. If you want a boolean logical to be H >$! turned off then do not define it. If you define it to "0" or "FALSE" >you'll  >$! be turning it on. M >$!**************************************************************************  >****  >$! J >$! Do not include any personal name from the VMS mail "From:" line in the7 >$! "From:" header line that we send to the POP client. " >$! DEFINE UCX$POP_PERSONAL_NAME 1 >$! < >$! Do not do a PURGE/RECLAIM when closing user's mail file." >$! DEFINE UCX$POP_PURGE_RECLAIM 1 >$! ! >$! Do not ignore mail11 headers. * >$! DEFINE UCX$POP_IGNORE_MAIL11_HEADERS 1 >$! * >$! Take UCX POP server default log level.I >$! DEFINE UCX$POP_LOG_LEVEL "enter ERROR, INFORMATIONAL, THREAD or DEBUG  >here" >$!  >$!************************  >$! Start the Server >$!************************ + >$ UCXPOP := $SYS$SYSTEM:UCX$POP_SERVER.EXE 	 >$ UCXPOP 2 >19100-05-31 05:50:19 sizeof(block_wait_times) 1604 >19100-05-31 05:50:19 sizeof(struct vms_time_rec) 32! >19100-05-31 05:50:19 num_elems 5 G >19100-05-31 05:50:19 starting UCX POP server UCX V4.1-12, OpenVMS V7.1 1 >Alpha on host nodename.mydomain.com and port 110  >19100-05-31 05:50:19 - >19100-05-31 05:50:19 POP configuration info: 5 >19100-05-31 05:50:19    ignore_mail11_headers: FALSE / >19100-05-31 05:50:19    send_id_headers: FALSE + >19100-05-31 05:50:19    disuserpass: FALSE - >19100-05-31 05:50:19    personal_name: FALSE - >19100-05-31 05:50:19    purge_reclaim: FALSE . >19100-05-31 05:50:19    max_messages: No max.8 >19100-05-31 05:50:19    master_log_level: INFORMATIONAL+ >19100-05-31 05:50:19    security: FRIENDLY 2 >19100-05-31 05:50:19    decnet_rewrite: TRANSFORM9 >19100-05-31 05:50:19    quoted_decnet_rewrite: TRANSFORM - >19100-05-31 05:50:19    sndbuf: UCX default. < >19100-05-31 05:50:19    link_idle_timeout:    0 00:02:00.00 >19100-05-31 05:50:19 ) >19100-05-31 05:50:19 Miscellaneous info: I >19100-05-31 05:50:19    UCX POP version: UCX V4.1-12, OpenVMS V7.1 Alpha + >19100-05-31 05:50:19    POP server PID: A7 ; >19100-05-31 05:50:19    POP server process name: UCX$POP_1  >19100-05-31 05:50:19 . >19100-05-31 05:50:19 SMTP configuration info:, >19100-05-31 05:50:19    substitute_domain:  >19100-05-31 05:50:19  >19100-05-31 05:50:21 E >%%%%%%%%%%%%                   31-MAY-2000 05:50:21.68  %%%%%%%%%%%% < >%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual; >address=FFFFFFFFF64E733F, PC=000000000003F37C, PS=0000001B  > < >%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual; >address=00000000F64E733F, PC=000000000003F37C, PS=0000001B  > 3 >  Improperly handled condition, image exit forced. 2 >    Signal arguments:   Number = 00000000000000052 >                        Name   = 000000000000000C2 >                                 00000000000100002 >                                 00000000F64E733F2 >                                 000000000003F37C2 >                                 000000000000001B >  >    Register dump: K >    R0  = 00000000007EFD09  R1  = 000000007B520017  R2  = 0000000000011DC0 K >    R3  = 0000000000050410  R4  = 000000007AFC79C8  R5  = 0000000000509140 K >    R6  = 0000000000023830  R7  = 0000000000050170  R8  = 0000000000050250 K >    R9  = 0000000000050370  R10 = 00000000000503F0  R11 = 00000000000507D6 K >    R12 = 000000007AFC7308  R13 = 0000000000050260  R14 = 0000000000000000 K >    R15 = 00000000009EAE4C  R16 = 000000007AFC7328  R17 = 00000000F64E72F7 K >    R18 = 0000000000000020  R19 = 0000000000000000  R20 = 00000000000005A8 K >    R21 = 0000000000023BE8  R22 = 0000000000000017  R23 = 000000007AFC7350 K >    R24 = 0000000000000000  R25 = 0000000000000588  R26 = 000000000003F33C K >    R27 = 0000000000011CE0  R28 = 0000000000000000  R29 = 000000007AFC72E0 K >    SP  = 000000007AFC72E0  PC  = 000000000003F37C  PS  = 200000000000001B  >$ ERROR_TRAP: >$ SS_STATUS = $STATUS >$ ON WARNING THEN CONTINUE ' >$ ERROR_MESSAGE = F$MESSAGE(SS_STATUS)  >$ NODE = F$GETSYI("NODENAME") >$! C >$ MAIL SYS$INPUT "SYSTEM"/SUBJECT="KIRK - %SYSTEM-F-ACCVIO, access A >violation, reason mask=!XB, virtual address=!XH, PC=!XH, PS=!XL" I >Enter your message below. Press CTRL/Z when complete, or CTRL/C to quit: @ >The UCX POP server has experienced a runtime error.  The reasonA >for the error should appear on the subject line of this message.  > 8 >Please investigate this problem as quickly as possible. >Thank you.  >$!  >$!************************  >$! Final Exit >$!************************  >$ FINAL_EXIT: >$ EXIT 9 >  UCX$POP      job terminated at 31-MAY-2000 05:50:22.92  >  >  Accounting information:K >  Buffered I/O count:                267      Peak working set size:        >4528 I >  Direct I/O count:                  197      Peak virtual size:          >178240 L >  Page faults:                      1448      Mounted volumes:              >  0D >  Charged CPU time:        0 00:00:00.98      Elapsed time:       0 >00:00:05.59 >  >  >****************  >****************  >****************  >  > > >and here is the list of all the UCX images I got from INSTAL> >  >  > 2 >   UCX$ACCESS_SHR;1 Open     Shar          Lnkbl 2 >   UCX$CFS_SHR;1    Open     Shar     Prot Lnkbl 2 >   UCX$ESNMP_SHR;1  Open Hdr Shar Prv      Lnkbl 2 >   UCX$IPC_SHR;1    Open     Shar          Lnkbl 2 >   UCX$RPCXDR_SHR;1 Open     Shar          Lnkbl  >   UCX$SMTP_MAILSHR;12 >                    Open     Shar          Lnkbl  >   UCX$SMTP_PARSESHR_TV;12 >                    Open     Shar          Lnkbl  >  >  >  >  >   UCX$ESNMP_SERVER;1' >                    Open Hdr Shar Prv  ' >   UCX$FTP;1        Open Hdr Shar Prv  ' >   UCX$FTPC;1       Open Hdr Shar Prv  ' >   UCX$FTPD;1       Open Hdr Shar Prv  ' >   UCX$OS_MIBS;1    Open Hdr Shar Prv  ' >   UCX$POP_SERVER;1               Prv  ' >   UCX$RLOGIN;1     Open Hdr Shar Prv   >   UCX$SMTP_RECEIVER;1 ' >                                  Prv  ' >   UCX$TELNET;1     Open Hdr Shar Prv  # >   UCX$UCP;1        Open Hdr Shar     ------------------------------  $ Date: Fri, 2 Jun 2000 19:26:41 +0100. From: David Skinner <david3@drspc.demon.co.uk> Subject: Re: VAX on Intel?0 Message-ID: <5W58vDAhx$N5Ewjg@drspc.demon.co.uk>  D In article <3934fd1b@news.toast.net>, Jeff Killeen <Jeff@Killeen.cc> writes >http://www.charon-vax.com/   6 Downloaded it this morning, and it seems to work well.  G However, I'm a VAX/VMS newbie and what I really want it for is to learn 
 about VMS.  E I have a question: When I finished playing with it, I needed to bring < the system down. As a guess, I typed SHUTDOWN, which worked.  F However, the last message told me to halt the system at the console. I? couldn't get a prompt, so I ended up just killing the emulator.   H Big mistake. The next time I booted, I got 7...6...5...4...3... OH SHIT,F Faiure (okay, I'm paraphrasing). Pesumably I screwed the filesystem or5 something. I had to wipe the emulator and re-install.   F Is there a magic key combination to get back to the console prompt andE halt the system properly? Or could I have cleaned up the mess by some 3 other means? Have I missed something in the manual?    --  
 David Skinner    ------------------------------  % Date: Fri, 02 Jun 2000 17:59:45 -0400 * From: David A Froble <davef@tsoft-inc.com> Subject: Re: VAX on Intel?- Message-ID: <39382E51.1C5499BA@tsoft-inc.com>    Eric Dittman wrote:  > @ > In comp.org.decus Tim Shoppa <shoppa@trailing-edge.com> wrote: > : David J. Dachtera wrote:; > :> I got a code for the download. Will try real soon now.  > :>L > :> If my "Affordable OpenVMS BOF" gets accepted for CETS-2000, and if RichF > :> Marcello can be there, how 'bout we demo it for him at that time?E > :> (Remember to have it on a Compaq PC so no one bitches about it.)  > G > : Maybe I'm just being dense, but how does a VAX emulator on an Intel D > : platform make anything more affordable?  Let's add up the costs,< > : not including a VMS license, for a 4 VUP implementation: >  > : The non-emulator way:   > :       VAXStation 3100/38:$35  > :       1 Gig SCSI drive:  $35  > :       Console terminal:  $30  > :                         ----  > :       Total:            $100 >  > : The emulator way: & > :       500 MHz cheapo Pentium: $600, > :       Windows 95/98/NT:       $100-$500.E > :       Emulator license:       unknown, but based on the Charon-11 C > :                               pricing probably a few thousand $ & > :                               ----; > :       Total:                  Several thousand dollars.  > H > : It sounds to me like the only way the VAX emulator makes sense is ifH > : the IT boss insists on spending lots of money on the Intel platform,G > : out of an irrational fear of older reliable DEC hardware.  And this I > : is all assuming that a VMS license will be available at any price for  > : the emulator.  > E > What you don't have listed for the non-emulator way is the hardware @ > support costs for the old VS3100.  If 24/7 hardware support isB > required that $100 VS3100/38 is going to get expensive fast.  IfC > you can convince management and/or the customer that having a few D > spare units around to swap out is acceptable then you will stil beE > able to save quite a bit of money (heck, five total VS3100/38 units B > at $100/each would still cost less than the base 500MHz PC), butB > sometimes you have to have a support contract to satisfy certain1 > requirements set by management or the customer.  > -- > Eric Dittman > dittman@dittman.net   E Just out of curiosity, what's a 24x7, Compaq service type of quality, C maintenance contract cost for the 500 MHz PC, NT, and the emulator?    Dave   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596; 170 Grimplin Road               E-Mail: davef@tsoft-inc.com  Vanderbilt, PA  15486    ------------------------------  # Date: Sat, 03 Jun 2000 04:01:04 GMT 3 From: Eric Dittman <dittman@narnia.int.dittman.net>  Subject: Re: VAX on Intel?? Message-ID: <4q%Z4.52918$v7.2116040@news-west.usenetserver.com>   = In comp.org.decus David A Froble <davef@tsoft-inc.com> wrote:  : Eric Dittman wrote:  :>  F :> What you don't have listed for the non-emulator way is the hardwareA :> support costs for the old VS3100.  If 24/7 hardware support is C :> required that $100 VS3100/38 is going to get expensive fast.  If D :> you can convince management and/or the customer that having a fewE :> spare units around to swap out is acceptable then you will stil be F :> able to save quite a bit of money (heck, five total VS3100/38 unitsC :> at $100/each would still cost less than the base 500MHz PC), but C :> sometimes you have to have a support contract to satisfy certain 2 :> requirements set by management or the customer. :> --  :> Eric Dittman  :> dittman@dittman.net  G : Just out of curiosity, what's a 24x7, Compaq service type of quality, E : maintenance contract cost for the 500 MHz PC, NT, and the emulator?   B There's where the inconsistency lies:  Management and the customerC usually have different expectations for hardware maintenance on the E NT box even though it may be mission critical since, after all, "it's F just a PC".  If the application runs on a midrange or mainframe systemB they want to see 24x7 hardware support but if it runs on a PC thenA they don't really worry as long as there is an extra box or parts E somewhere.  I'm not even going to get into the "dependability" of the  PC running NT. --   Eric Dittman dittman@dittman.net    ------------------------------  % Date: Fri, 02 Jun 2000 18:31:55 -0400 2 From: "Richard B. Gilbert" <DRAGON@compuserve.com> Subject: Re: VAX VMS 7.2 Bug? 7 Message-ID: <200006021832_MC2-A755-8A85@compuserve.com>   F         Some problems, either the same or related, are apparently wellC known to VMS Engineering.  I have in my possesion a document headed J "Employees Only" which says (paraphrasing) that problems with BACKUP /IMA= GEF /ALIAS being unable to restore files are known and that the product isJ performing within design specifications and that there is no intention to=   change the specifications!!   J         This has got to be the work of a saboteur!!!   No true VMS person=  H would even consider doing such a thing!  I suspect it's one of these new! "make it work like Unix weenies".       $ Message text written by "Neil Rieck"  >VAX VMS 7.2 Update (2000.06.02)   Folks,  E I was finally able to find some time to address this problem. I'm 90% D certain that there's a bug in "VAX VMS 7.2 BACKUP" from "VAX VMS 7.2D UPDATE_1" patch forward. (only the programmer would be 100% certain)  J The original "VAX VMS 7.2 BACKUP" from the CD-ROM seems to work and so do= es "Alpha VMS 7.2 BACKUP".    My Tests =3D=3D=3D=3D=3D=3DJ Last night I decided to do a new installation of "VAX VMS 7.2" from CD-RO= M J onto a spare disk. This would yield a system disk that had no questionabl= e J alias or back link problems. Next, I booted the new system disk, then did=  a= hot-backup of it to a third disk. Everything worked properly.   H Next, I rebooted the original "VAX VMS 7.2" (which contains a new BACKUPH image from VAX_UPDATE_1) and then repeated the previous backup operationJ from the new system disk to a third disk. The backup failed with the same=  9 failure mode mentioned previously in this message thread.    Summary  =3D=3D=3D=3D=3D=3D; Maybe OpenVMS Engineering at Compaq can now look into this.   F p.s. If memory serves; I think we installed the "VAX VMS 7.2 UPDATE_1" patch D because we were getting incomplete backups (interrupted by an access1 violation when ever the /JOURNAL switch was used)  <    ------------------------------  % Date: Fri, 02 Jun 2000 20:37:06 -0400 ' From: Neil Rieck <n.rieck@sympatico.ca>  Subject: Re: VAX VMS 7.2 Bug? 3 Message-ID: <000901bfccf3$d93c7340$b4ceacce@nrieck>   # Please tell me that you are joking. N Seriously, if you are not joking you should scan the document and then post it on news group "comp.os.vms"   
 Neil Rieck* Kitchener(New Berlin?)/Waterloo/Cambridge, Ontario, Canada.! http://www3.sympatico.ca/n.rieck/    ----- Original Message -----0 From: Richard B. Gilbert <DRAGON@compuserve.com>H To: Neil Rieck <n.rieck@sympatico.ca>; [unknown] <Info-VAX@Mvb.Saic.Com>! Sent: Friday, June 02, 2000 18:31  Subject: Re: VAX VMS 7.2 Bug?     F         Some problems, either the same or related, are apparently wellC known to VMS Engineering.  I have in my possesion a document headed K "Employees Only" which says (paraphrasing) that problems with BACKUP /IMAGE F /ALIAS being unable to restore files are known and that the product isI performing within design specifications and that there is no intention to  change the specifications!!   I         This has got to be the work of a saboteur!!!   No true VMS person H would even consider doing such a thing!  I suspect it's one of these new! "make it work like Unix weenies".       $ Message text written by "Neil Rieck"  >VAX VMS 7.2 Update (2000.06.02)   Folks,  E I was finally able to find some time to address this problem. I'm 90% D certain that there's a bug in "VAX VMS 7.2 BACKUP" from "VAX VMS 7.2D UPDATE_1" patch forward. (only the programmer would be 100% certain)  K The original "VAX VMS 7.2 BACKUP" from the CD-ROM seems to work and so does  "Alpha VMS 7.2 BACKUP".    My Tests ======J Last night I decided to do a new installation of "VAX VMS 7.2" from CD-ROMJ onto a spare disk. This would yield a system disk that had no questionableK alias or back link problems. Next, I booted the new system disk, then did a = hot-backup of it to a third disk. Everything worked properly.   H Next, I rebooted the original "VAX VMS 7.2" (which contains a new BACKUPH image from VAX_UPDATE_1) and then repeated the previous backup operationI from the new system disk to a third disk. The backup failed with the same 9 failure mode mentioned previously in this message thread.    Summary  ======; Maybe OpenVMS Engineering at Compaq can now look into this.   F p.s. If memory serves; I think we installed the "VAX VMS 7.2 UPDATE_1" patch D because we were getting incomplete backups (interrupted by an access1 violation when ever the /JOURNAL switch was used)  <    ------------------------------  % Date: Fri, 02 Jun 2000 23:49:43 -0400 2 From: "Richard B. Gilbert" <DRAGON@compuserve.com> Subject: Re: VAX VMS 7.2 Bug? 7 Message-ID: <200006022350_MC2-A755-9357@compuserve.com>   J         I'm not joking.  I'm far from certain that it would be wise to po= stJ the original document.  First, I believe it's copyright by Compaq.  Secon= d,. I feel that I should not compromise my source!    $ Message text written by "Neil Rieck"$ >Please tell me that you are joking.J Seriously, if you are not joking you should scan the document and then po= st it on news group "comp.os.vms"  <    ------------------------------  % Date: Fri, 02 Jun 2000 15:03:53 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca>  Subject: Re: VEST / DECmigrate, Message-ID: <39380519.6B0A3BCC@videotron.ca>   Paul Hardy wrote: J > 1)  VEST requires a VAX image linked no later than VMS V5.5-2, so if youN > have built VAX images under V6 or V7, you have to find an old VMS system and# > recompile them before translating   N Ouch. Technically, couldn't DECpaq recompile VEST on a more modern VMS so that4 it would be able to translate more modern programs ?   ------------------------------  " Date: Fri, 2 Jun 2000 21:53:12 GMT9 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen)  Subject: Re: VEST / DECmigrate+ Message-ID: <X8KUDpFbJ2ar@eisner.decus.org>   \ In article <39380519.6B0A3BCC@videotron.ca>, JF Mezei <jfmezei.spamnot@videotron.ca> writes: > Paul Hardy wrote: K >> 1)  VEST requires a VAX image linked no later than VMS V5.5-2, so if you O >> have built VAX images under V6 or V7, you have to find an old VMS system and $ >> recompile them before translating > P > Ouch. Technically, couldn't DECpaq recompile VEST on a more modern VMS so that6 > it would be able to translate more modern programs ?   No.   < DECpaq could put a lot of effort into updating VEST and TIE,< but it has nothing to do with what version VEST was compiledC on and the type of work is _far_ more elaborate than recompilation.   9 VEST was a worthwhile investment back when Alpha started, 8 and they only had C and Fortran compilers ported.  These9 days, the only audience is those who haven't been able to  port their code over 7 years.    ------------------------------  $ Date: Fri, 2 Jun 2000 14:10:31 -0400/ From: "IanPercival" <IanPercival@email.msn.com> @ Subject: Re: VMS File Caching Futures (Was: Re: Andrew whatever)) Message-ID: <#WeqE3Lz$GA.347@cpmsnbbsa08>   F I don't read the news group very often as I'm usually busy doing otherK things - but there does seem to be an awful lot of discussion about VMS and L file data caching.   I don't particularly want to get involved in any of theI various discussions  - but there are an awful lot of misconceptions being L bandied about.  Feel free to email me at ian.percival@compaq.com if you have  serious questions or discussion.  ! VMS has two sets of data caching. L 1. RMS. It can use Local or Global Buffers.  Some manual setiing and controlE is required in order to use these if the application isn't using them  automatically.   2. System wide data caches. E First of all I agree that the system wide central data cache that VMS I has/had known variously as VCC or VIOC which was first implemented on VAX K V6.0 a number of years ago, is pretty poor by many performance metrics.  It F actually got worse, as functionality such as dynamic memory sizing gotH dropped during the port to Alpha.  This is the reason there are a myriadI third party implementations of caches (OK maybe not quite that many - but 2 quite a few!) all of which worked better than VCC.  F A new system wide file data cache called XFC (stands for extended fileH cache) has been developed.   It is a true 64 bit cache - can store Tb of data if you have the memory.H It can store over 100 CLOSED files (the VCC limitation was that it wouldD only store up to 100 closed files - not files in total!). XFC has noK limitation on number of closed files it will store - apart from the obvious  one of memory usage.D It can cache I/Os larger than 34 blocks (VCC was hard limited to 34)+ It can perform readahead where appropriate. I It provides performance statistics such as absolute I/O response times in 
 microseconds. 9 It is dynamic or static or both in terms of memory usage. L My Alpha boots much quicker when using it! Many benchmarks are significantly better when using it.   C XFC V1.0 is about to enter the field test cycle.  By the time it is F released, it should have some other major performance features added -D making it perform even better for users of medium to large machines.   Hope this helps a bit!   Ian Percival XFC Project Leader OpenVMS Engineering ) (writing from home on my son's birthday!)    ------------------------------  $ Date: Fri, 2 Jun 2000 13:46:36 -05001 From: "Dave Gudewicz" <david.gudewicz@abbott.com> @ Subject: Re: VMS File Caching Futures (Was: Re: Andrew whatever)8 Message-ID: <8h8v7q$7dm$1@fizban.fizban.pprd.abbott.com>  G Thanks for this update Ian and more importantly in the long run,  Happy  Birthday to your son!!   Dave...   8 IanPercival <IanPercival@email.msn.com> wrote in message# news:#WeqE3Lz$GA.347@cpmsnbbsa08... H > I don't read the news group very often as I'm usually busy doing otherI > things - but there does seem to be an awful lot of discussion about VMS  and J > file data caching.   I don't particularly want to get involved in any of the K > various discussions  - but there are an awful lot of misconceptions being I > bandied about.  Feel free to email me at ian.percival@compaq.com if you  have" > serious questions or discussion. > # > VMS has two sets of data caching. F > 1. RMS. It can use Local or Global Buffers.  Some manual setiing and control G > is required in order to use these if the application isn't using them  > automatically. >  > 2. System wide data caches. G > First of all I agree that the system wide central data cache that VMS K > has/had known variously as VCC or VIOC which was first implemented on VAX I > V6.0 a number of years ago, is pretty poor by many performance metrics.  ItH > actually got worse, as functionality such as dynamic memory sizing gotJ > dropped during the port to Alpha.  This is the reason there are a myriadK > third party implementations of caches (OK maybe not quite that many - but 4 > quite a few!) all of which worked better than VCC. > H > A new system wide file data cache called XFC (stands for extended fileJ > cache) has been developed.   It is a true 64 bit cache - can store Tb of > data if you have the memory.J > It can store over 100 CLOSED files (the VCC limitation was that it wouldF > only store up to 100 closed files - not files in total!). XFC has noE > limitation on number of closed files it will store - apart from the  obvious  > one of memory usage.F > It can cache I/Os larger than 34 blocks (VCC was hard limited to 34)- > It can perform readahead where appropriate. K > It provides performance statistics such as absolute I/O response times in  > microseconds. ; > It is dynamic or static or both in terms of memory usage. @ > My Alpha boots much quicker when using it! Many benchmarks are
 significantly  > better when using it.  > E > XFC V1.0 is about to enter the field test cycle.  By the time it is H > released, it should have some other major performance features added -F > making it perform even better for users of medium to large machines. >  > Hope this helps a bit! >  > Ian Percival > XFC Project Leader > OpenVMS Engineering + > (writing from home on my son's birthday!)  >    ------------------------------   Date: 2 Jun 2000 22:51:41 GMT 2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)@ Subject: Re: VMS File Caching Futures (Was: Re: Andrew whatever), Message-ID: <8h9dpt$k1m@gap.cco.caltech.edu>  X In article <PjT1QfU8ohB5@eisner.decus.org>, young_r@eisner.decus.org (Rob Young) writes: > < >"I'm not counting any of those chickens before they hatch." > = >	In fairness to David, maybe he didn't know about filesystem A >	caching futures BUT I do know that has been mentioned more than < >	once in this forum.  So in examination of the evidence hisC >	response more than reveals his bent.  He's a "glass is half full"  >	kinda guy   J Actually, I'm a "I can't load it now, and will believe it only when I can"I kind of guy.   Also a "what in the hell could have been a higher priority E all these years than improving disk IO throughput 10X?" kind of guy.    H Besides, regarding disk IO, my performance tests for the programs we runK here under our typical load indicate that the glass is way below half full, H in some tests, it's barely even moist.  However, for CPU bound jobs it's/ 100% full.  For security it's about 400% full.     > A >	But it is a lose lose proposition for those that want to use it D >	today.  No wonder some of those folks are "glass half full" kinda 
 >	folks.    H It's also going to be a big loser if DII-COE goes in before an improved G caching file system does.  The benchmarks/test programs are all written J to run, essentially, on a Solaris system.  There is little tuning that canI be done to improve them, and direct access to RMS would seem to be out of H the question (in terms of compliance with the standard). Since they willJ only use write() and fprintf() many of the IO intensive ones will run like dogs on OpenVMS as it is now.    Regards,   David Mathog mathog@seqaxp.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech     ------------------------------  " Date: Sat, 3 Jun 2000 01:51:23 GMT9 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen) @ Subject: Re: VMS File Caching Futures (Was: Re: Andrew whatever)+ Message-ID: <RbNRPSXaQfix@eisner.decus.org>   a In article <8h9dpt$k1m@gap.cco.caltech.edu>, mathog@seqaxp.bio.caltech.edu (David Mathog) writes:   J > It's also going to be a big loser if DII-COE goes in before an improved I > caching file system does.  The benchmarks/test programs are all written L > to run, essentially, on a Solaris system.  There is little tuning that canK > be done to improve them, and direct access to RMS would seem to be out of J > the question (in terms of compliance with the standard). Since they willL > only use write() and fprintf() many of the IO intensive ones will run like > dogs on OpenVMS as it is now.   @ VMS has to meet the standard, but I don't think the standard has a performance requirement.  ? That is not to say that performance is irrelevant, but that the > order of file caching vs. DII-COE support may not be critical.  A We did have a post from someone working on file caching, but none $ from someone working on DII-COE :-).   ------------------------------  % Date: Fri, 02 Jun 2000 17:18:35 -0700 3 From: Jojimbo <jgesslingNOjgSPAM@yahoo.com.invalid> @ Subject: Re: VMS File Caching Futures (Was: Re: Andrew whatever)9 Message-ID: <020eceba.1df14787@usw-ex0102-084.remarq.com>   6 at the risk of getting back to the original subject...  = Having just spent the best part of a week understanding, so I @ can modify it, a completely async AST driven IO program, I think0 I understand a bit more why the kids like U***X.   Because it's simple!  ; No need to do async IO, the operating system does it.  Just ? read and write, so race conditions, buffers to manage until the 7 IO completes, etc... are not an issue.  At least to the > programmer!  Now the users and the customers and the operators= might be in big trouble, but the programmer is done and gone!    Sigh,    Jim    p.s.@ By the way. "I have a Java Class that will do all that, although= you may have to change the interface a bit".  (heard over the * cubicle wall this afternoon)  Double sigh.      L * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network *G The fastest and easiest way to search and participate in Usenet - Free!    ------------------------------  $ Date: Fri, 2 Jun 2000 21:18:37 -0400' From: "Bill Todd" <billtodd@foo.mv.com> @ Subject: Re: VMS File Caching Futures (Was: Re: Andrew whatever)( Message-ID: <8h9m79$oim$1@pyrite.mv.net>  > Jojimbo <jgesslingNOjgSPAM@yahoo.com.invalid> wrote in message3 news:020eceba.1df14787@usw-ex0102-084.remarq.com... 8 > at the risk of getting back to the original subject... > ? > Having just spent the best part of a week understanding, so I B > can modify it, a completely async AST driven IO program, I think2 > I understand a bit more why the kids like U***X. >  > Because it's simple! > 7 > No need to do async IO, the operating system does it.   K That's only true in some cases, automatic multi-buffering being perhaps the  most significant.   H Some more cases are handled, though not transparently, by using multipleK threads (each using synchronous I/O) in a single process where, in the days F before system-supported threads existed, you would have had to use theG single-threaded/asynchronous-I/O whirling dervish approach.  This works G pretty well when the required per-thread overhead is reasonable and the L operations performed by the individual threads are essentially sequential inI nature (but interact in ways that can be synchronized - and shared - more I efficiently within a single process than across multiple processes, which = was the old traditional Unix substitute for asynchrony and/or  multi-threading).   K But when it comes to rats-nests like databases or distributed file systems, L where disk accesses and distributed network operations pop up asynchronouslyL and multiple times within the course of a single operation, there's still noJ comparably-efficient alternative to having just enough threads to keep allK the available physical processors busy and having each such thread work its ? little tail off asynchronously.  Of course, such (often kernel) L 'applications' nicely hide such complexity from their clients (just like theA Unix file system itself does, though I'm not sure how common real K down-and-dirty asynchrony is even in Unix kernels:  it's so - unUnixy), but E the best (or at least most powerful) implementations pass support for J asynchrony up to the application level so that the applications can do the, same kind of thing when they really need to.  I So the real question is whether your 'asynch AST driven IO program' could J have benefited from using threads to avoid its asynchrony or avoided it byF virtue of Unix's automatic system support for multi-buffering.  If theJ former, doesn't VMS also support kernel-based threads?  If the latter, RMS9 provides such mechanisms, albeit you have to invoke them.   K In either case, VMS gives you the means to avoid those nasty ASTs (at least K any you can see) and still get the same performance you can get with Unix - K it just doesn't do this *by default*, which, while important for the casual K user (who no way is going to act differently anyway, but will just complain G about performance), is not the same as *forcing* you into the depths of I complexity you've been exploring.  And if the application you're fiddling J with *really needed* to be AST-driven on VMS, then it would have needed toJ be coded in just about the same manner in a Unix environment - except thatF you'd be hard-pressed to find a Unix environment with as comprehensive< support for asynchrony (when you really need it) as VMS has.   - bill     JustA > read and write, so race conditions, buffers to manage until the 9 > IO completes, etc... are not an issue.  At least to the @ > programmer!  Now the users and the customers and the operators? > might be in big trouble, but the programmer is done and gone!  >  > Sigh,  >  > Jim  >  > p.s.B > By the way. "I have a Java Class that will do all that, although? > you may have to change the interface a bit".  (heard over the , > cubicle wall this afternoon)  Double sigh. >  >  > L > * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * I > The fastest and easiest way to search and participate in Usenet - Free!  >    ------------------------------  % Date: Fri, 02 Jun 2000 22:39:11 -0500 * From: Keith Brown <kbrown780@usfamily.net>@ Subject: Re: VMS File Caching Futures (Was: Re: Andrew whatever), Message-ID: <39387DDE.47954141@usfamily.net>   Bill Todd wrote: > 7 > Rob Young <young_r@eisner.decus.org> wrote in message ' > news:Uz8oKEwyXUi3@eisner.decus.org... N > > In article <8h7l78$jpg$1@pyrite.mv.net>, "Bill Todd" <billtodd@foo.mv.com>	 > writes: M > > > Talking to a brick wall sometimes seems more effective:  at least one's 2 > > > expectations can be set reasonably up front. > > > I > > > David's comment "I'm not counting any of those chickens before they  > hatch"D > > > seems, at least to me, to have a very obvious, and reasonable,I > > > interpretation:  Spiralog was planned, was hyped, actually shipped,  > and - K > > > flopped.  The same thing could happen with EFC, and its appearance on  > the M > > > road map, in which you place so much faith, means approximately as much  > asJ > > > Spiralog's did (at least I assume Spiralog appeared on a road map at > some
 > > > point).  > > >  > > 9 > > Ah, now we are in the realm of "it happened before it > > > can happen again."   Sure.  But then the question becomes: > >  > > "Do you think it will?"  > >  > > Is it worth a wager to you?  > G > I've already said that I don't *expect* EFC to be the kind of failure N > Spiralog was (though it could happen).  And I can't remember the last time IH > wagered on anything, though I'm reasonably sure it was well before VMSK > appeared on the scene:  I don't have anything against wagering, it's just L > not of interest to me (if it were, I suspect I'd spend my time working the( > market instead of this kind of thing). > J > However, my expectation is that, while it should improve current defaultL > performance considerably, default file system performance with EFC V2 willN > likely still fall somewhat short of Unix default file system performance (atI > least for some of the better implementations, like SGI's):  some of the M > optimizations involve things like cooperating with the file system to defer G > specific space allocation until time of actual disk write (there's no N > indication I know of that EFC is getting into stuff like that), while othersD > aren't cache-related at all (e.g., the use of a log to allow smallF > synchronous writes to be made persistent immediately in an efficient
 > manner). >  > - bill >  > >  > > Rob  > >    >EFC V2 willN > likely still fall somewhat short of Unix default file system performance (at; > least for some of the better implementations, like SGI's)   > As you are fond of saying about NT, "It may not be perfect but it is good enough"   --   Keith Brown  kbrown780@usfamily.net   ------------------------------  % Date: Fri, 02 Jun 2000 22:54:59 -0500 * From: Keith Brown <kbrown780@usfamily.net>@ Subject: Re: VMS File Caching Futures (Was: Re: Andrew whatever), Message-ID: <39388193.FBE81237@usfamily.net>   Bill Todd wrote: > @ > Jojimbo <jgesslingNOjgSPAM@yahoo.com.invalid> wrote in message5 > news:020eceba.1df14787@usw-ex0102-084.remarq.com... : > > at the risk of getting back to the original subject... > > A > > Having just spent the best part of a week understanding, so I D > > can modify it, a completely async AST driven IO program, I think4 > > I understand a bit more why the kids like U***X. > >  > > Because it's simple! > > 9 > > No need to do async IO, the operating system does it.  > M > That's only true in some cases, automatic multi-buffering being perhaps the  > most significant.  > J > Some more cases are handled, though not transparently, by using multipleM > threads (each using synchronous I/O) in a single process where, in the days H > before system-supported threads existed, you would have had to use theI > single-threaded/asynchronous-I/O whirling dervish approach.  This works I > pretty well when the required per-thread overhead is reasonable and the N > operations performed by the individual threads are essentially sequential inK > nature (but interact in ways that can be synchronized - and shared - more K > efficiently within a single process than across multiple processes, which ? > was the old traditional Unix substitute for asynchrony and/or  > multi-threading).  > M > But when it comes to rats-nests like databases or distributed file systems, N > where disk accesses and distributed network operations pop up asynchronouslyN > and multiple times within the course of a single operation, there's still noL > comparably-efficient alternative to having just enough threads to keep allM > the available physical processors busy and having each such thread work its A > little tail off asynchronously.  Of course, such (often kernel) N > 'applications' nicely hide such complexity from their clients (just like theC > Unix file system itself does, though I'm not sure how common real M > down-and-dirty asynchrony is even in Unix kernels:  it's so - unUnixy), but G > the best (or at least most powerful) implementations pass support for L > asynchrony up to the application level so that the applications can do the. > same kind of thing when they really need to. > K > So the real question is whether your 'asynch AST driven IO program' could L > have benefited from using threads to avoid its asynchrony or avoided it byH > virtue of Unix's automatic system support for multi-buffering.  If theL > former, doesn't VMS also support kernel-based threads?  If the latter, RMS; > provides such mechanisms, albeit you have to invoke them.  > M > In either case, VMS gives you the means to avoid those nasty ASTs (at least M > any you can see) and still get the same performance you can get with Unix - M > it just doesn't do this *by default*, which, while important for the casual M > user (who no way is going to act differently anyway, but will just complain I > about performance), is not the same as *forcing* you into the depths of K > complexity you've been exploring.  And if the application you're fiddling L > with *really needed* to be AST-driven on VMS, then it would have needed toL > be coded in just about the same manner in a Unix environment - except thatH > you'd be hard-pressed to find a Unix environment with as comprehensive> > support for asynchrony (when you really need it) as VMS has. >  > - bill >  >   JustC > > read and write, so race conditions, buffers to manage until the ; > > IO completes, etc... are not an issue.  At least to the B > > programmer!  Now the users and the customers and the operatorsA > > might be in big trouble, but the programmer is done and gone!  > > 	 > > Sigh,  > >  > > Jim  > >  > > p.s.D > > By the way. "I have a Java Class that will do all that, althoughA > > you may have to change the interface a bit".  (heard over the . > > cubicle wall this afternoon)  Double sigh. > >  > >  > > N > > * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network > * K > > The fastest and easiest way to search and participate in Usenet - Free!  > > 4 > > I understand a bit more why the kids like U***X. > > Because it's simple!  3 How difficult is this in PL/I. What can be simpler?   @ OPEN FILE          (MY_OUTFILE)  TITLE ('MY_OUTFILE.dat') OUTPUT.      ENVIRONMENT   (BUFFERS(255),WRITEBEHIND);  ' WRITE FILE (MY_OUTFILE) FROM (BUFFER));    CLOSE (MY_OUTFILE);        --   Keith Brown  kbrown780@usfamily.net   ------------------------------  $ Date: Sat, 3 Jun 2000 01:23:34 -0400' From: "Bill Todd" <billtodd@foo.mv.com> @ Subject: Re: VMS File Caching Futures (Was: Re: Andrew whatever)( Message-ID: <8ha4ig$jbe$1@pyrite.mv.net>  5 Keith Brown <kbrown780@usfamily.net> wrote in message & news:39388193.FBE81237@usfamily.net...   ...   5 > How difficult is this in PL/I. What can be simpler?   J 1) Using a language that the average developer (who, after all, is exactlyL the developer who *won't* do much if anything in the way of optimization) is likely to select.   K 2) Making the desired behavior transparent rather than requiring that it be K specified (though making it easy to obtain, albeit explicitly, is certainly & better than keeping it truly obscure).  L As for your other comment elsewhere, if VMS performance is 'good enough' forI you even if it doesn't match Unix performance, rejoice and be happy.  But D don't assume that's 'good enough' for everyone, or even a majority -H especially given purchasing situations where VMS is typically the systemD that must justify being considered against competition that sets the2 expectations by virtue of its industry acceptance.   - bill   > B > OPEN FILE          (MY_OUTFILE)  TITLE ('MY_OUTFILE.dat') OUTPUT0 >      ENVIRONMENT   (BUFFERS(255),WRITEBEHIND); > ) > WRITE FILE (MY_OUTFILE) FROM (BUFFER));  >  > CLOSE (MY_OUTFILE);  >  >  >  > --
 > Keith Brown  > kbrown780@usfamily.net   ------------------------------  % Date: Fri, 02 Jun 2000 17:15:42 -0400 . From: Dan Gahlinger <dan@itac11-sun.sprint.ca>= Subject: What is the simplest way to network two VMS systems? 4 Message-ID: <393823FE.BBE6E492@itac11-sun.sprint.ca>  F     I've looked at the openvms site for DECnet info (over 500 pages of it) - not read in detail yet. C I've also tried running cluster_config myself, but it asks some odd " questions and I'm not sure what it
 really wants.   H What is the quickest, easiest way to get two VMS systems to talk to each other?@ Assuming you set up the decnet node numbers like system1-3.1 and system2-3.2 or somethingG the tcp/ip docs look very foreboding - I've done tcp/ip on unix and pcs 0 and nowhere near as complex as this seems to be.  H I'd like users to be able to log onto one system and access resources onH the other one as well, also be able to use "phone" and notes and such toG communicate. Mail also. This seems to require full decnet connectivity. F or would cluster_config suffice?  would tcp/ip by itself be enough and work ok?   Dan.   --" -There are always possibilities...   ------------------------------   Date: 2 Jun 2000 22:26:53 GMT / From: morris@iridium.mv.net (Skipper W. Morris) A Subject: Re: What is the simplest way to network two VMS systems? ( Message-ID: <8h9cbf$58r$1@pyrite.mv.net>  : Simplest?  Perhaps Kermit.  But it won't do what you want.  F DECnet Phase IV is the easiest, and it's well integrated into the O/S.  / Clusters would work for a lot of stuff as well.   . At home I use DECnet, VMSClusters, and TCP/IP.  F The TCP/IP so the PC's can talk to VMS.  Cluster and DECnet so the VMS systems work well.   /Skip    ------------------------------  # Date: Fri, 02 Jun 2000 23:15:05 GMT  From: Dan <dan@vrx.net> A Subject: Re: What is the simplest way to network two VMS systems? ' Message-ID: <39383F5C.7970D52A@vrx.net>   ! So then the question becomes how. B I thought cluster-config would be easy, but it asked me some weird
 questions.J something about a cluster ID number (not the node number, it already asked	 me that).   , Maybe I'll go over the tcp/ip docs again. :)   Thanks   "Skipper W. Morris" wrote:  < > Simplest?  Perhaps Kermit.  But it won't do what you want. > H > DECnet Phase IV is the easiest, and it's well integrated into the O/S. > 1 > Clusters would work for a lot of stuff as well.  > 0 > At home I use DECnet, VMSClusters, and TCP/IP. > H > The TCP/IP so the PC's can talk to VMS.  Cluster and DECnet so the VMS > systems work well. >  > /Skip    ------------------------------  % Date: Fri, 02 Jun 2000 19:30:59 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> A Subject: Re: What is the simplest way to network two VMS systems? , Message-ID: <393843B0.2B206171@videotron.ca>   Dan Gahlinger wrote:E > I've also tried running cluster_config myself, but it asks some odd $ > questions and I'm not sure what it > really wants.   J If you have DECNET 4 istalled, then @SYS$MANAGER:netconfig.com is what youN need to configure decnet. You need to configure it on both nodes. And you need) to know the device name of yoru ethernet.   M The "Guide to decnet networking" has examples of how to configure things. (It 5 may be in the section on the old manuals on the net).   J > I'd like users to be able to log onto one system and access resources onJ > the other one as well, also be able to use "phone" and notes and such toI > communicate. Mail also. This seems to require full decnet connectivity. H > or would cluster_config suffice?  would tcp/ip by itself be enough and
 > work ok?    Y DECNET allows MAIL, PHONE, and network file specifications (NODE::device:[dir]file.name).   J CLUSTERING adds to this by sharing resources such as disks and print/batchL queues to create a single entitu across a cluster. (you'd access a file withN device:[dir]file.name in a cluster since the device is mounted on your node as' well, so no need to use remote access).    ------------------------------  % Date: Fri, 02 Jun 2000 14:37:41 -0500 ) From: "John E. Malmberg" <wb8tyw@qsl.net> H Subject: Where to find VXT load image  was - Re: where to find firmware?7 Message-ID: <14de01bfccca$055bb6e0$020a0a0a@xile.realm>   , http://www.openvms.digital.com/freeware/VXT/   ------------------------------  # Date: Sat, 03 Jun 2000 01:44:18 GMT + From: Jonas Lindholm <jlindholm@nyc.rr.com> - Subject: Re: Which is the language of VAX/VMS * Message-ID: <393862B1.95152624@nyc.rr.com>  " It is not Cobol and Unix is an OS.  0 The original VMS was written in Macro and Bliss.  H The current version is written in C and some parts in Macro. There could& still be parts that are using Bliss...   /Jonas Lindholm 
 OM Technology    moises_hernandez wrote:   D > Do you know which is the lenguage of the VAX/VMS operating system? > Is Cobol or Unix? 	 > Regards  > Moises Hernandez   ------------------------------  % Date: Fri, 02 Jun 2000 15:18:22 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> " Subject: Re: Wildfire Announcement, Message-ID: <3938087D.419B9F98@videotron.ca>   Peter Moreton wrote: > G > The fact is that OpenVMS is a fine operating system, in the mainframe F > 24*7*365 class. It will continue to be used in volume as it is rightH > now, for a very long time. OpenVMS is a cash-cow for Compaq (check the: > figures) and shows no signs of ceasing to be profitable.   Wrong and wrong.  M VMS is a fine operating system from desktop to the mainframe,  and is perhaps N the best with that type of scalability. That Digital and Compaq have chosen toS remove VMS from certain markets is only a marketing issue, not a technological one.   J An no, Palmer's efforts to kill VMS still are having an effect. Compaq mayM have stopped actively trying to kill VMS, and they have have shown signs that K they want to push it more, but so far, my opinion is that it has slowed the G bleeding. It still has some distance to go before not only is the wound , healed, but the thing starts to get popular.  K Perhaps Compaq management do not have the credibility on the street to risk N making a long term marketing commitment for VMS. If the street doesn't beleiveH that Compaq can sell PCs competitively, they might think Compaq is crazy putting money in VMS marketing.    ------------------------------  % Date: Sat, 03 Jun 2000 00:33:19 -0400 + From: "Main, Kerry" <Kerry.Main@compaq.com> " Subject: RE: Wildfire AnnouncementJ Message-ID: <910612C07BCAD1119AF40000F86AF0D805284353@kaoexc4.kao.dec.com>  " Sorry, no time for long replies ..  H >>> The fact is that 8i still isn't available on OpenVMS it was due last0 year, then march and it still hasn't happened.<<  B oops, error. Oracle 8.1.6 on OpenVMS is apparently now shipping toI Customers. I do know one of our partners was testing 8.1.6 last week with ; OPS and it went very well. So much for that old argument ..   > >>> I asked you who these unnamed eBusiness partners were. <<<  I Sure, Compaq will just blurt out what potential partners it is talking to E before any final deals are made .. Yup, that would go over real well.    Regards,  
 Kerry Main Senior Consultant,
 Compaq Canada  Professional Services  Voice : 613-592-4660 FAX   : 819-772-7036 Email : kerry.main@compaq.com        -----Original Message-----' From: Andrew Harrison SUNUK Consultancy ! [mailto:andrew.nospam@uk.sun.com] # Sent: Friday, June 02, 2000 5:18 AM  To: Info-VAX@Mvb.Saic.Com " Subject: Re: Wildfire Announcement     "Main, Kerry" wrote:   > Andrew, Andrew ..  > J > So, if you have no faith in what Compaq marketing is stating on officialI > Compaq pages, and if you continue to think that the head of Oracle does  not L > in any way stand behind what he says, and if you continue to ignore recentF > wins (17% of OpenVMS revenue last year was new business, but I don't expectH > you to believe that) and if you continue to try to spread gloom-n-doomJ > information from the past to the readers of this list no matter what new" > information is presented here .. >   F Kerry, what exactly did Larry say. He said nice products, nice companyI and Oracle runs on the nice procucts. Thats it, there isn't anything more B reading Larrys comments as being some sort of long term endorsment? of Compaq/Tru64/OpenVMS is naive in the extreme. Reading Larrys < comments as being an endorsment of Tru64/OpenVMS over any ofF Oracle's other platforms is simply clutching at straws. He turns up to> our launches, HP's launches and IBM's launches and pretty much says the same thing.  ? Its designed to keep the platform vendor doing the launch happy : without offending Oracle's other partners. Why would Larry; want to offend Sun, HP or IBM. Sun has the largest share of 6 Oracles revenues on our systems would he want to place< that revenue stream under any threat. Of course he wouldn't.  = I would place a lot more emphasis on Larrys kind words if the > words were backed up by deads. The fact is that 8i still isn't9 available on OpenVMS it was due last year, then march and ; it still hasn't happened. How many OpenVMS customers do you A think you have lost to other vendors or to Tru64 because of this, 9 I know of two that I have worked with in the UK and thats ? before you start looking at all the other apps that Oracle have  that don't run on OpenVMS.  J > However, when presented with very positive new information, you continueJ > with your "sky is falling" FUD that totally ignores the new information. >   : No sorry, there is no new information in the press release= you posted, there arn't endorsments from any of the eBusiness : partners alluded to in the press release but not named and? this is simply a preservation of the status quo, press releases 5 are cheap, but marketing and market development isn't  just press releases.  9 Come on Kerry where is the new information, I bet that if 8 I could find the 8400 product announcements I would find( an almost identical sentiment expressed.   > > > How can you expect anyone to reasonably take you seriously ? >   4 Sorry but given your postings this is hardly a point2 that is worth making from your standpoint. I asked. you who these unnamed eBusiness partners were.  0 You posted a Cognos URL which simply showed that6 it was possible to put a WEB front end on a Powerhouse/ app running on OpenVMS, this isn't the basis of 0 an eBusiness strategy and if you think it is and5 if your views reflect Compaqs views then to be honest " you may as well retire gracefully.  E > I just came back from a week of futures, new stuff in VMS V7.3, new L > application vendor strategies (including eBusiness), great new performanceJ > enhancements, new marketing strategies (read the magic word GROW OpenVMS > business) etc etc..  >   D Kerry an application vendor strategy is simply that a strategy, what? matters is how that strategy is executed and what resources are A available to assist in the execution. You need to spend money and I lots of it, people like Broadvision and Ariba and all the other eBusiness < apps vendors arn't going to port to Tru64 or OpenVMS withoutI you parting with cash. It is also pretty apparent that Digital and Compaq B have never been prepared to spend the kind of cash required to get them into this market.  B Why do you think Sun has being buying people like JCP and InnosoftF it wasn't because somebody said nice technology it was because someone= in market development said we need this technology to capture 4 a particular market segment lets go buy the company.   Regards  Andrew Harrison  Enterprise IT Architect    ------------------------------  % Date: Sat, 03 Jun 2000 00:47:07 -0400 + From: "Main, Kerry" <Kerry.Main@compaq.com> " Subject: RE: Wildfire AnnouncementJ Message-ID: <910612C07BCAD1119AF40000F86AF0D805284354@kaoexc4.kao.dec.com>   Andrew,   H Of course, Oracle has many partners. So does Compaq. So does HP. So does IBM.    L That is the reality of todays world. While a statement of support from LarryG Ellison is obviously not an exclusive statement, it is a statement that . Compaq is a serious player in Oracle's books.   $ Customers also recognise this point.  L Without a quote from Larry Ellison, you would be running around stating "HowI come Oracle does not say nice things about Compaq's Tru64 UNIX or OpenVMS  like he does for HP or Sun ?"   / With the quote, you are saying its no big deal.   5 You are simply twisting words to spread gloom-n-doom.   % Ah well, you do it very well though..    :-)    Regards,  
 Kerry Main Senior Consultant,
 Compaq Canada  Professional Services  Voice : 613-592-4660 FAX   : 819-772-7036 Email : kerry.main@compaq.com        -----Original Message-----' From: Andrew Harrison SUNUK Consultancy ! [mailto:andrew.nospam@uk.sun.com] # Sent: Friday, June 02, 2000 5:22 AM  To: Info-VAX@Mvb.Saic.Com " Subject: Re: Wildfire Announcement     JF Mezei wrote:    > "Main, Kerry" wrote:L > > So, if you have no faith in what Compaq marketing is stating on officialK > > Compaq pages, and if you continue to think that the head of Oracle does  not ( > > in any way stand behind what he says > E > I don't think that this is what has been said. I think it is more a  questionC > of Compaq having no real special status within the Oracle empire.  >   D Exactly, Kerry seems to be claiming some kind of special endorsement@ and relationship with Oracle based on Larrys comments, which areF almost identical to the ones he has made for Sun launches, HP launches and IBM launches.   ; Sadly OpenVMS/VMS did once have a special status in Oracles ? empire, it was the development platform and first port but that : is long gone with Solaris and in some areas NT taking over	 from VMS.    Regards  Andrew Harrison  Enterprise IT Architect    ------------------------------   End of INFO-VAX 2000.308 ************************