1 INFO-VAX	Sat, 31 May 2003	Volume 2003 : Issue 299       Contents: accessing device registers% Re: advsyscon remote shadow on alpha? D Re: Block IO question for RMS files ... Detecting EOF for empty fileP Re: Block IO question for RMS files ... Detecting EOF for empty file with deleteP Re: Block IO question for RMS files ... Detecting EOF for empty file with delete% Re: Calculating semesters with DCL... * Can't zero lan counters after ALPLAN06_062' Current Alpha and Vax Software for Sale + Re: Current Alpha and Vax Software for Sale + Re: Current Alpha and Vax Software for Sale  Re: ES40 and CPU upgrade Re: ES45 AGP Re: ES45 AGP Re: ES45 AGP Re: ES45 AGP Re: ES45 AGP Re: ES45 AGP eXcursion RAS Mystery  Re: Help with VMS ' Re: Hobbyist seeks generic SIMM vendors # Re: How to sort a list in VAX BASIC  HP gets squeezed missing sys$errorlog file? Re: missing sys$errorlog file? Re: missing sys$errorlog file?, Re: Mystery Alpha CPUS, what do they fit in?, Re: Mystery Alpha CPUS, what do they fit in?, Re: Mystery Alpha CPUS, what do they fit in?$ Re: Needed, history of startup files Re: Portents of VMS death  Re: Portents of VMS death  Re: Portents of VMS death  Re: Portents of VMS death   Re: SYS$QIOW - Performance Issue  Re: SYS$QIOW - Performance Issue( SYS$QIOW - Wtih more Details - Final Try, Re: SYS$QIOW - Wtih more Details - Final Try, Re: SYS$QIOW - Wtih more Details - Final Try  F ----------------------------------------------------------------------  # Date: Sat, 31 May 2003 03:28:00 GMT  From: hoblit@liii.com # Subject: accessing device registers , Message-ID: <3ed81e49.7610192@news.liii.com>  F To access device registers in a device driver VMS supplies 2 differentB methods: use CRAMs or the ioc$read_io & ioc$write_io routines. TheC device driver book discusses both methods but doesn't compare them. D Is one method preferable to the other? Is there a slight performanceH gain using CRAM rather than  IOC routines? Do the ioc$read_io & write_ioA internally convert the request to one using CRAMs/ or vice versa?   : It probably doesn't really matter very much, just curious.   Sam    ------------------------------  % Date: Sat, 31 May 2003 13:23:12 +1000  From: "Pip" <pip@ti.nl0.com>. Subject: Re: advsyscon remote shadow on alpha?< Message-ID: <3ed8200a$0$16255$afc38c87@news.optusnet.com.au>  H Hal, we're mirroring four disk, and are capturing 10 minute snapshots ofH data transferred. At the peak time, the total average data throughput isH something like 200KBytes/sec - so the real peak per second rate would be4 something higher than that - perhaps double or more.L Our data is alot of text, so the compression feature of RSO helps alot ; the0 compressed data is probably 50% of the raw data.B The actual transfer sizes are a bit harder to quantify; RSO can beG configured to either send the data as-is immediately, or wait a defined L period of msec for more data to fill the bucket, or wait until the specifiedK message size is filled. This is the area that takes a bit of monitoring and ; tuning to determine the optimal message size for each disk. K The WAN pipe (which can burst to 20Mbit) is alot fatter than the RSO needs, 8 because there's alot more stuff running over it as well.   HTH  Pip.    - "Hal Kuff" <kuff@tessco.com> wrote in message ' news:bb7h7s$hnc@library1.airnews.net... J > Pip, could you provide some info on the size of transfers and the nature of > the wan link please? >  >  >    ------------------------------  % Date: Fri, 30 May 2003 20:48:39 -0500 1 From: "David J. Dachtera" <djesys.nospam@fsi.net> M Subject: Re: Block IO question for RMS files ... Detecting EOF for empty file ' Message-ID: <3ED809F7.B578369D@fsi.net>    mckinneyj@cpva.saic.com wrote: > 1 > In article <bb7mrq$deg$1@news.fas.harvard.edu>, 9 >  Chip Coldwell <coldwell@frank.harvard.invalid> writes: * > > Dann Corbit <dcorbit@connx.com> wrote:? > >> "Larry Kilgallen" <Kilgallen@SpamCop.net> wrote in message 2 > >> news:4GTqTUCDyFzT@eisner.encompasserve.org...B > >>> In article <bb6016$d3g$1@news1.tilbu1.nb.home.nl>, Dirk Munk > >> <munk@home.nl> writes:  > >>> > Dann Corbit wrote: > >>> P > >>> >> I have searched high and low for the book "RMS Internals" but it cannot > >> be  > >>> >> had any more. > >>> > ; > >>> > The VMS books have been a bit reorganized it seems. O > >>> > Did you check the "OpenVMS Record Management Services Reference Manual"  > >> ? > >>> N > >>> The "OpenVMS Record Management Services Reference Manual" is part of the > >>> VMS Documentation. > > F > >> I have this document.  If my answer is in there, I can't find it. > > A > >>> "RMS Internals" was a Digital Press book (by Kirby McCoy?).  > > ) > >> Yes, I know.  I can't find the book.  > >  > > http://www.amazon.com/exec/obidos/tg/detail/-/1555580564/qid=1054301724/sr=1-1/ref=sr_1_1/002-8236346-3756063?v=glance&s=books > > H > > 3 used copies, starting at $90.  DP is coming out with a new edition > > in Dec 2004. > >  > > Chip > >  > > --( > > Charles  M. "Chip" Coldwell      __o' > > "Turn on, log in, tune out"      \< ) > >                               (@)/(@) ) > > -------------------------------------  > F > McCoy's book is title "VMS File System Iternals". This should not beD > confused with RMS internals. The file system and RMS are differentI > animals. McCoy does not discuss RMS. This book focuses on the structure E > of the file system, volumes, caching, locking, QIOs, etc - but, not  > the record management system.   H Interesting. I thought I was the only one who made a distinction between them.    --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  % Date: Fri, 30 May 2003 10:12:46 -0700 ' From: "Dann Corbit" <dcorbit@connx.com> Y Subject: Re: Block IO question for RMS files ... Detecting EOF for empty file with delete , Message-ID: <bb83eh018a5@enews4.newsguy.com>  = "Chris Sharman" <chris.sharman@sorry.nospam> wrote in message . news:bb7ftf$4h0$1$8300dec7@news.demon.co.uk... > Larry Kilgallen wrote:> > > In article <bb6hbl01s6q@enews3.newsguy.com>, "Dann Corbit" <dcorbit@connx.com> writes:  > >  > > 4 > >>>>But in this case, the seek is returning 65537. > >>> ; > >>>That is a normal status - it did what you asked of it.  > >>L > >>I know what it means.  But it is a lie.  The function says it read a big) > >>block of data, but the file is empty.  > >  > > 1 > > But didn't you say this was an indexed file ?  > > 8 > > On an indexed file, block IO will read the metadata. > B > The easy way (to test for empty) is to use sys$get or sys$find !I > Depending what you want to do, you could alternatively check out Hein's A > rms_tune_check, available as freeware (with source), very nice, H > discussed here recently. Understands RMS indices pretty well - they're> > complex, & it took Hein a few goes to get all bases covered.   Thanks!  I found it. --  0 C-FAQ: http://www.eskimo.com/~scs/C-faq/top.html$  "The C-FAQ Book" ISBN 0-201-84519-9H C.A.P. FAQ: ftp://cap.connx.com/pub/Chess%20Analysis%20Project%20FAQ.htm   ------------------------------  % Date: Fri, 30 May 2003 20:53:35 -0500 1 From: "David J. Dachtera" <djesys.nospam@fsi.net> Y Subject: Re: Block IO question for RMS files ... Detecting EOF for empty file with delete & Message-ID: <3ED80B1F.EF0133D@fsi.net>   David Froble wrote:  > [snip]! > Forget keys, there are no keys.    There is no spoon.   --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------   Date: 30 May 2003 18:16 CDT ' From: carl@gerg.tamu.edu (Carl Perkins) . Subject: Re: Calculating semesters with DCL...- Message-ID: <30MAY200318161664@gerg.tamu.edu>   1 contracer11@uol.com.br (Shiva MahaDeva) writes...  }Giving two dates like this: } 7 }  10-nov-2000 -  23-jul-2001   ->  first semester/2001  } 7 }  15-dec-2002 -  05-aug-2003   ->  first semester/2003  } 8 }  05-may-2002 -  28-jan-2003   ->  second semester/2002 } 8 }  11-jun-2000 -  01-jan-2001   ->  second semester/2000 }  } B } How could I calculate this using a procedure ? (first or second  }   } semester , using two dates ? ) }  } Thanks in advance...  @ The very first thing you need to do is define what, exactly, you9 mean by "semester" and what two dates have to do with it.   A For example, your third line has a "start date" (I assume that is > what the first date is) of 05-MAY-2002 which would normally beD considered to be in the first half of 2002 and thus would presumably@ be considered to be in the "first semester/2002" in your scheme.C Likewise it has an "end date" (I assume that is what the second one G is) that is 28-JAN-2003, which would not normally be considered "second @ semester/2002" but "first semester/2003".  So neither date is in/ the time span that you have decided to call it.   C You seem to want the half of the year that the range fills the most F completely, or perhaps just completely, but it is hard to be certain -A you could be using some deinfined values for start end end dates.   A If you define exactly what you mean in detailed enough terms, the C program is almost written (since about all a program is is an exact I definition of what is meant, excepting bugs, using a specific language).    A I suspect that you can do it mostly with with a comparison on the C dates using only the month part in numerical format, comparing them D to 07 (July), but that is assuming that your dates will always startF before the start of the semester (in a literal "first and second halfsC of the months of a calendar year" definition of the term, where the B 1st semester is 1-Jan through 30-Jun) and end after the end of theI semester. If this is the case, then if the MM parts are both greater than D or equal to 07 then the semester is the first semester, otherwise ifD they are both less then 07 then it is the second semester, otherwiseF it is an error (the assumption that the start and end dates would bothH be outside of the actual semester is then seen to be false, or the datesF are incorrect). Then, assuming you did not hit the error condition, ifE it is the first semester it is in the year of the end date, otherwise D it is the second semester of the year of the start date. The DCL forC this is not complicated. But again, this is using assumptions about 5 what you really mean that I pretty much just made up.   @ The important routine for doing this is likely to turn out to beC the F$CvTime() lexical. For example, you can extract the month from 2 a date string, output in the numerical format, via  0 $ month = F$CvTime( date, "COMPARISON", "MONTH")  ! and the year in a similar manner.    --- Carl   ------------------------------    Date: 30 May 2003 12:02:36 -0700& From: chessmaster1010@hotmail.com (JG)3 Subject: Can't zero lan counters after ALPLAN06_062 = Message-ID: <dd3f0cb7.0305301102.11f216c4@posting.google.com>   E We have a mixed VMS 6.2/7.2 cluster.  On the 6.2 nodes the LANCP SHOW F DEVICE /COUNTERS are 32 bit, just like NCP SHOW LINE COUNTERS, and NCP ZERO LINE COUNTERS zeros both.  F On the 7.2 nodes the LANCP counters are 64 bit and to zero them we useF the command LANCP SET DEVICE /DEVICE_SPECIFIC=FUNCTION="CCOU", which I got from the VMS FAQ.   D I just installed patch ALPLAN06_062 on some of the 6.2 nodes.  I wasE excited to see in the release notes that it backports the support for C 64 bit counters into its version of LANCP.  We log network counters D every hour and we have some nodes that break the 4GB barrier in less time than that.   = The problem is that they didn't backport the /DEVICE_SPECIFIC C qualifier into that LANCP version.  So with no way to zero them the ; LANCP device counters are now very big, but pretty useless.    ------------------------------  # Date: Fri, 30 May 2003 18:29:00 GMT 3 From: "George Samuelson" <samuelsong@insightbb.com> 0 Subject: Current Alpha and Vax Software for Sale1 Message-ID: <MpNBa.1076414$S_4.1082862@rwcrnsc53>    http://cgi6.ebay.com/ws/eBayISAPI.dll?ViewSellersOtherItems&include=0&userid=robins-nest-collectibles&sort=3&rows=25&since=-1&rd=1  K Please check the link for Alpha and Vax OS and SPL software up for auction.    ------------------------------   Date: 30 May 2003 18:41:38 GMT( From: bill@cs.uofs.edu (Bill Gunshannon)4 Subject: Re: Current Alpha and Vax Software for Sale5 Message-ID: <bb88l1$6vdre$1@ID-135708.news.dfncis.de>   1 In article <MpNBa.1076414$S_4.1082862@rwcrnsc53>, 6 	"George Samuelson" <samuelsong@insightbb.com> writes: > http://cgi6.ebay.com/ws/eBayISAPI.dll?ViewSellersOtherItems&include=0&userid=robins-nest-collectibles&sort=3&rows=25&since=-1&rd=1 > M > Please check the link for Alpha and Vax OS and SPL software up for auction.    D Hmmmm.....  Maybe this is what I should be doing with those boxes ofC old layered product sets and stuff.  Heaven knows when your hobbies D include computers like PDP-11's, VAX and Alpha youe can sure use the9 extra money.  And that's just for the electric bill......    bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  % Date: Sat, 31 May 2003 00:06:55 +0200 + From: "Hans Vlems" <hvlems.nieuw@zonnet.nl> 4 Subject: Re: Current Alpha and Vax Software for Sale5 Message-ID: <bb8km3$6rm41$1@ID-143435.news.dfncis.de>   ) The electric bill indeed rings a bell :-) J BTW did you notice that the second person on the list had an email address+ on mindspring.com, sounds familiar somehow.     7 "Bill Gunshannon" <bill@cs.uofs.edu> schreef in bericht / news:bb88l1$6vdre$1@ID-135708.news.dfncis.de... 3 > In article <MpNBa.1076414$S_4.1082862@rwcrnsc53>, 7 > "George Samuelson" <samuelsong@insightbb.com> writes:  > >  http://cgi6.ebay.com/ws/eBayISAPI.dll?ViewSellersOtherItems&include=0&userid=robins-nest-collectibles&sort=3&rows=25&since=-1&rd=1 > > F > > Please check the link for Alpha and Vax OS and SPL software up for auction. > F > Hmmmm.....  Maybe this is what I should be doing with those boxes ofE > old layered product sets and stuff.  Heaven knows when your hobbies F > include computers like PDP-11's, VAX and Alpha youe can sure use the; > extra money.  And that's just for the electric bill......  >  > bill >  > --  L > Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesF > bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. > University of Scranton   |@ > Scranton, Pennsylvania   |         #include <std.disclaimer.h>   ------------------------------  % Date: Fri, 30 May 2003 11:20:35 -0600 ' From: DigiDemon <digidemon@hotmail.com> ! Subject: Re: ES40 and CPU upgrade 8 Message-ID: <pan.2003.05.30.17.20.35.277894@hotmail.com>  4 On Thu, 29 May 2003 09:09:39 +0100, Nic Clews wrote:   > DigiDemon wrote: >>  J >> Real quick...we have just bought a 887 mtz CPU for our spiffy ES40.  WeM >> needed this to support our SmartArray 5304 card.  Anyone ever run into any  >> snags doing this?  Thanks!  > I > Have the latest copy of the firmware CD, or the relevant files from the ? > download site handy, and you'll need VMS 7.2-2 or above IIRC.   I Thanks Nic...had a PALCode/CPU mismatch, but reloading the firmware fixed I that and it went like a champ =)  Too bad the SmartArray 5304 cards STILL 
 don't work =(    James    ------------------------------  % Date: Fri, 30 May 2003 21:47:35 +0100 * From: "John Travell" <john@travell.uk.net> Subject: Re: ES45 AGP 5 Message-ID: <bb8gka$73qga$1@ID-120847.news.dfncis.de>   . "Tom Linden" <tom@kednos.com> wrote in message3 news:CIEJLCMNHNNDLLOOGNJIKENLHEAA.tom@kednos.com...  >  >  > >-----Original Message----- A > >From: Fred Kleinsorge [mailto:my-last-name@stardotzko.dec.com] % > >Sent: Friday, May 30, 2003 8:00 AM  > >To: Info-VAX@Mvb.Saic.Com > >Subject: Re: ES45 AGP > >  > > @ > >Yes on all counts.  But the VX1 will not use any AGP-specific capabilitiesK > >and is still 2D-only.  I have recently made some changes to the VX1 code I > >that optimizes the server for DMA operation - which should make things I > >faster and use less CPU.  But the VX1 is nearing end-of-life.  The ATI 6 > >Radeon 7500 will replace it for both 2D and for 3D. > > J > >Note that to get the AGP configuration of the ES45, you give up several PCI K > >slots.  So unless you are doing some heavy duty visualization with 3D...  > >I'd stick with PCI. >  > Why do you give up PCI slots? 1 > Is it not a separate slot as on typical PC MBs?  > 7 Not in this case. There are two different motherboards. ? One is PCI only, the other has an AGP slot and fewer PCI slots.   < Just imagine an ES45 workstation... That would be stunning !     -- John Travell  VMS crashdump expertise for hire john@travell.uk.net  http://www.travell.uk.net/       --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).A Version: 6.0.484 / Virus Database: 282 - Release Date: 27/05/2003    ------------------------------    Date: 30 May 2003 17:19:38 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen)  Subject: Re: ES45 AGP 3 Message-ID: <Mx07UeREsQqz@eisner.encompasserve.org>   b In article <bb8gka$73qga$1@ID-120847.news.dfncis.de>, "John Travell" <john@travell.uk.net> writes:  > > Just imagine an ES45 workstation... That would be stunning !   That would be noisy.   ------------------------------  % Date: Sat, 31 May 2003 00:39:18 +0200  From: Dirk Munk <munk@home.nl> Subject: Re: ES45 AGP 2 Message-ID: <bb8mpj$kn2$1@news3.tilbu1.nb.home.nl>   VAXman- wrote:  M > Now all we need is a browser...  That works...  That would mean that it can L > access and navigate all of HP's pages which the present offering DOES NOT!  P Well, I'm using Mozilla on OpenVMS (check the headers of this message), and I'm O quite happy with it. The only thing missing is a proper pdf reader, but I have  L reasons to believe we will get one real soon. I think that is all I need to  access the HP site.    >  > --Q > VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COM 
 >             7 >   "Well my son, life is like a beanstalk, isn't it?"   >    ------------------------------  # Date: Sat, 31 May 2003 01:20:29 GMT " From:   VAXman-  @SendSpamHere.ORG Subject: Re: ES45 AGP 0 Message-ID: <00A20A79.C3C222FE@SendSpamHere.ORG>  S In article <bb8mpj$kn2$1@news3.tilbu1.nb.home.nl>, Dirk Munk <munk@home.nl> writes:  >VAXman- wrote:  > N >> Now all we need is a browser...  That works...  That would mean that it canM >> access and navigate all of HP's pages which the present offering DOES NOT!  > Q >Well, I'm using Mozilla on OpenVMS (check the headers of this message), and I'm  P >quite happy with it. The only thing missing is a proper pdf reader, but I have M >reasons to believe we will get one real soon. I think that is all I need to   >access the HP site.  H Great.  I'll send you my username and password to the /DSPP pages.  WhenB you've gone through them and gotten me approved, do let me know.     --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COM              5   "Well my son, life is like a beanstalk, isn't it?"     ------------------------------  % Date: Fri, 30 May 2003 23:26:13 -0400 2 From: rdeininger@mindspring.com (Robert Deininger) Subject: Re: ES45 AGP L Message-ID: <rdeininger-3005032326130001@user-uinj4dr.dialup.mindspring.com>  D In article <bb8gka$73qga$1@ID-120847.news.dfncis.de>, "John Travell" <john@travell.uk.net> wrote:  / >"Tom Linden" <tom@kednos.com> wrote in message 4 >news:CIEJLCMNHNNDLLOOGNJIKENLHEAA.tom@kednos.com... >> >> >> >-----Original Message-----B >> >From: Fred Kleinsorge [mailto:my-last-name@stardotzko.dec.com]& >> >Sent: Friday, May 30, 2003 8:00 AM >> >To: Info-VAX@Mvb.Saic.Com  >> >Subject: Re: ES45 AGP  >> >  K >> >Note that to get the AGP configuration of the ES45, you give up several  >PCIL >> >slots.  So unless you are doing some heavy duty visualization with 3D... >> >I'd stick with PCI.  >>  >> Why do you give up PCI slots?2 >> Is it not a separate slot as on typical PC MBs? >>8 >Not in this case. There are two different motherboards.@ >One is PCI only, the other has an AGP slot and fewer PCI slots.  ? Actually, there are 3 different bus configurations in the ES45:    Model 1 has:    2 33-MHz PCI slots     4 66-MHz PCI slots 
    1 AGP slot    Model 2 has:    4 33-MHz PCI slots     6 66-MHz PCI slots    Model 3 has:    8 33-MHz PCI slots     2 66_MHz PCI slots   I In each case, the 33-MHz slots support 5V cards; the 3.3 V slots do not.  H (The model 3 is targeted at CIPCA and Memory Channel customers, who like lots of 5V slots.)  I The chipset supports 4 PCI busses.  The AGP slot uses a whole bus, so the ! total number of slots is reduced.    ------------------------------  % Date: Fri, 30 May 2003 23:11:47 -0400 ( From: David Froble <davef@tsoft-inc.com> Subject: Re: ES45 AGP , Message-ID: <3ED81D73.4070101@tsoft-inc.com>   Larry Kilgallen wrote:  d > In article <bb8gka$73qga$1@ID-120847.news.dfncis.de>, "John Travell" <john@travell.uk.net> writes: >  > > >>Just imagine an ES45 workstation... That would be stunning ! >> >  > That would be noisy.     Depends.  O How long are the cables from the monitor, keyboard, and whatever, and how many   walls do the cables go through?      Dave     --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com6 T-Soft, Inc.  170 Grimplin Road  Vanderbilt, PA  15486   ------------------------------  % Date: Fri, 30 May 2003 21:29:04 -0700 . From: terrence golden <terrygolden@brandx.net> Subject: eXcursion RAS Mystery) Message-ID: <3ED82F8F.78704C6@brandx.net>   * I'm having a problem access my VMS servers* through eXcursion when using a RAS server.1 At first it was that the net admins didn't config ' a rexec port. Now rexec works fine, but , $cre/term/det doesn't create a window on the? client (PC running eXterm Piece of Crap Home). when I do a $set / display/cre/node=ipadr/trans=tcpip it completes 0 successfully but again no display. $cre/term/det. fails with can't create display error. The tcp. logs don't show any errors. It works fine when0 you're inside the firewall, as in not USING RAS.   Any ideas? Thanks in advance.    ------------------------------   Date: 30 May 2003 18:46 CDT ' From: carl@gerg.tamu.edu (Carl Perkins)  Subject: Re: Help with VMS- Message-ID: <30MAY200318463566@gerg.tamu.edu>   / "Hans Vlems" <hvlems.nieuw@zonnet.nl> writes... K }The 3100-98 is faster than the 3100-80 so it requires licenses with higher  }units for theH }base OS and possibly the layered products. Try $ SHOW LICENSE/CHARGE to! }determine what the system needs.   D This relationship is not universal. For example, an XP900 requires aJ 12 unit VMS license, but the (much) slower DEC 3000m600 requires 20 units.  E It pretty much depends on how the system is classified at the time it 9 is/was new. They rarely, if ever, change them after that.    --- Carl   ------------------------------  % Date: Fri, 30 May 2003 20:41:25 -0500 1 From: "David J. Dachtera" <djesys.nospam@fsi.net> 0 Subject: Re: Hobbyist seeks generic SIMM vendors' Message-ID: <3ED80844.51692392@fsi.net>    Alder wrote: > L > If anyone has links to online vendors known to stock SIMMs compatible withK > "older" AlphaServers (mine is a 1000A 4/266), could you please post links J > here.  I've contacted Island Computers, but the price is still beyond my( > reach.  Best would be sites in Canada.  . I'm pretty sure they're not in Canada, but try http://www.chipmerchant.com/   --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  % Date: Sat, 31 May 2003 00:08:57 +0200 + From: "Hans Vlems" <hvlems.nieuw@zonnet.nl> , Subject: Re: How to sort a list in VAX BASIC5 Message-ID: <bb8kps$71nm0$1@ID-143435.news.dfncis.de>   K Don't be surprised: the other day I was asked (requested) to send an Ultrix L kit to a student who had to build a web page on a VAXstation3100 that had toA run Ultrix. If it is true then this is entirely understandable...   7 "Bill Gunshannon" <bill@cs.uofs.edu> schreef in bericht / news:bb846v$6nqhk$1@ID-135708.news.dfncis.de... A > In article <OFCADE3386.F3E1315B-ON85256D36.00585AB1@metso.com>,   > norm.raphael@metso.com writes: > > + > > Is this a(nother)  homework assignment?  > % > In VAX basic, I seriously doubt it.  >  > bill >  > --  L > Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesF > bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. > University of Scranton   |@ > Scranton, Pennsylvania   |         #include <std.disclaimer.h>   ------------------------------  # Date: Fri, 30 May 2003 18:21:21 GMT # From: "John Smith" <a@nonymous.com>  Subject: HP gets squeezed J Message-ID: <BiNBa.292475$M81.177045@news02.bloor.is.net.cable.rogers.com>  7 http://www.interex.org/hpwnews/content/hpwn5.29.03.html   ? "A recent survey of 100 CIOs by Merrill Lynch & Co. showed that E opinions about HP as a vendor are mixed, while those about rivals IBM 0 and Sun are very low. " goes the lead paragraph.  	 [opinion] : But that's pretty much the positive side of the story.....
 [/opinion]  ? "When asked whether "HP will be gaining or losing share in your " account," listed by product areas,A only printers and PC servers registered a gain: 38 and 11 percent 
 respectively.   0 HP would lose share, according to the survey, in Unix servers (-10 percent),  services (-10 percent),  software (-12 percent),  PCs (-14 percent) and even storage (-17 percent). "  4 Let's keep advertising and marketing those PeeCee's.   ------------------------------  # Date: Fri, 30 May 2003 22:46:43 GMT 2 From: "frank brown" <frank.brown@ci.seattle.wa.us># Subject: missing sys$errorlog file? 2 Message-ID: <nbRBa.34$mc6.25943@news-west.eli.net>  L VAX/VMS 5.5-2 here...I thought when you renamed the system errorlog file, itI was supposed to create a new one.  So you can manage logfiles.  Well this K VAX hasn't created a new SYS$SYSROOT:[SYSERR]ERRLOG.SYS since I renamed the L last one a couple of weeks ago.  Doesn't seem right.  What can I do short of- a reboot to kick out a new SYS$ERRORLOG file?    Frank Brown ) Seattle Fire Dept - home of ancient VAXen  http://www.inwa.net/~frog/   ------------------------------  % Date: Fri, 30 May 2003 19:41:06 -0400 ' From: Stuart Fuller <stufuller@usa.net> ' Subject: Re: missing sys$errorlog file? 0 Message-ID: <i6q8bb.jmp.ln@dadsys2.fuller.local>   frank brown wrote:  K > VAX/VMS 5.5-2 here...I thought when you renamed the system errorlog file,  > itK > was supposed to create a new one.  So you can manage logfiles.  Well this I > VAX hasn't created a new SYS$SYSROOT:[SYSERR]ERRLOG.SYS since I renamed  > the K > last one a couple of weeks ago.  Doesn't seem right.  What can I do short 2 > of a reboot to kick out a new SYS$ERRORLOG file? > 
 > Frank Brown + > Seattle Fire Dept - home of ancient VAXen  > http://www.inwa.net/~frog/  L Is the ERRFMT process (the process that writes the error log file) running?   If not, then:  "         @sys$system:startup errfmt   should start it up.    --             Stu    ------------------------------    Date: 30 May 2003 21:02:53 -0700. From: spamsink2001@yahoo.com (Alan E. Feldman)' Subject: Re: missing sys$errorlog file? = Message-ID: <b096a4ee.0305302002.75e8dba5@posting.google.com>   l "frank brown" <frank.brown@ci.seattle.wa.us> wrote in message news:<nbRBa.34$mc6.25943@news-west.eli.net>...N > VAX/VMS 5.5-2 here...I thought when you renamed the system errorlog file, itK > was supposed to create a new one.  So you can manage logfiles.  Well this M > VAX hasn't created a new SYS$SYSROOT:[SYSERR]ERRLOG.SYS since I renamed the N > last one a couple of weeks ago.  Doesn't seem right.  What can I do short of/ > a reboot to kick out a new SYS$ERRORLOG file?  > 
 > Frank Brown + > Seattle Fire Dept - home of ancient VAXen  > http://www.inwa.net/~frog/  F Run SHOW SYSTEM and see if the process ERRFMT is running. Sometimes it, crashes. If it is not running, start it with        $ @SYS$SYSTEM:STARTUP ERRFMT  @ Check the SYSTEM account's mail. When my ERRFMT crashes (rarely,B usually shortly after a boot), it attempts to send mail indicatingD such to the SYSTEM account. (I am running systems with VMS v6.1,v6.2 VAX.)    Disclaimer: JMHO Alan E. Feldman    ------------------------------  % Date: Fri, 30 May 2003 19:24:31 +0100 * From: "Matt Simis" <mattsimis@hotmail.com>5 Subject: Re: Mystery Alpha CPUS, what do they fit in? 5 Message-ID: <bb87l0$6ns4q$1@ID-131939.news.dfncis.de>   4 "David Mathog" <mathog@caltech.edu> wrote in message2 news:20030530102459.577a9cf9.mathog@caltech.edu...$ > On Thu, 29 May 2003 11:08:48 +0100- > "Matt Simis" <mattsimis@hotmail.com> wrote:  >  > > I just cant get a straightJ > > answer on why there seems to be an abundance of socket based cpus with no > > motherboards?  > E > According to news over the last week or so the Opteron is currently H > in a similar state.  However, there will be motherboards for that soonG > but there's never going to be another one made for these Alpha chips.  >  >  > > Does any one have a clue?  > F > Not the management of the companies that made these CPUs - otherwiseH > they would have sold them in systems at retail rather than dumping theA > essentially unusable CPUs (no mobo's) through channels one step ' > shy of a dumpster.  It's particularly E > odd that they aren't being sold with mobo's, since if those existed ? > they'd be equally valueless without their CPU.  That tells me A > that the mobo's were never made and all of these Alpha CPUs are  > just fancy paperweights. > 
 > Regards, >  > David Mathog > mathog@caltech.edu@ > Manager, Sequence Analysis Facility, Biology Division, Caltech    J Hmm, you could be correct, but I thought Samsung made the motherboards andJ these CPUs, seems unlikely that they would make produce the more costly toH manufacture CPUs before the motherboards. How come socketed 21264s never appeared on any roadmaps?   J  However, surely someone who worked with API\Samsung\Whoever still browses3 this NG and could give us a more definitive answer?      Matt   ------------------------------    Date: 30 May 2003 14:08:34 -0500+ From: young_r@encompasserve.org (Rob Young) 5 Subject: Re: Mystery Alpha CPUS, what do they fit in? 3 Message-ID: <bU3hErHF1tOn@eisner.encompasserve.org>   b In article <20030530102459.577a9cf9.mathog@caltech.edu>, David Mathog <mathog@caltech.edu> writes:$ > On Thu, 29 May 2003 11:08:48 +0100- > "Matt Simis" <mattsimis@hotmail.com> wrote:  >  >> I just cant get a straight L >> answer on why there seems to be an abundance of socket based cpus with no >> motherboards? > E > According to news over the last week or so the Opteron is currently  > in a similar state.   7 	There is a good summary of why no Opteron boards here:   ( http://www.theinquirer.net/?article=9763  L The Opteron version of AMD64 is targetted at servers. The chipsets availableM now don't have AGP. Consequently the only thing that can be built is a server . board with little utility for other purposes.   L The Taiwan MB guys aren't crazy. They generally build multipurpose boards toM maximize their sales potential in return for the effort needed to design them . and perhaps more importantly build inventory.    ---   ) 	The Tawain MB guys aren't crazy.  Later:   M Mike, As one of the dozen Opteron launch partners, we have been going through M the same issues. We luckily got a few of the MSI boards and are waiting for a  proper working Tyan board.  M From what I have heard, the motherboard manufacturers had late shipments from - Broadcom and even AMD for the main chipsets.     ---   9 	So Opteron is/was a good paper launch for the most part.   4 >  However, there will be motherboards for that soon  ; 	Yeah, but every day that goes by is lost opportunity.  Sun 9 	would *love* to use Opterons but of course are forced to  	used Xeons.   				Rob    ------------------------------  # Date: Sat, 31 May 2003 01:36:43 GMT # From: "John Smith" <a@nonymous.com> 5 Subject: Re: Mystery Alpha CPUS, what do they fit in? J Message-ID: <LGTBa.313697$w7k.165402@news04.bloor.is.net.cable.rogers.com>  5 "Matt Simis" <mattsimis@hotmail.com> wrote in messageO/ news:bb87l0$6ns4q$1@ID-131939.news.dfncis.de...i6 > "David Mathog" <mathog@caltech.edu> wrote in message4 > news:20030530102459.577a9cf9.mathog@caltech.edu...& > > On Thu, 29 May 2003 11:08:48 +0100/ > > "Matt Simis" <mattsimis@hotmail.com> wrote:e > >L  > > > I just cant get a straightB > > > answer on why there seems to be an abundance of socket based	 cpus withh > no > > > motherboards?  > > = > > According to news over the last week or so the Opteron is 	 currentlyPE > > in a similar state.  However, there will be motherboards for that  soonB > > but there's never going to be another one made for these Alpha chips. > >t > >e > > > Does any one have a clue?  > >i> > > Not the management of the companies that made these CPUs -	 otherwise F > > they would have sold them in systems at retail rather than dumping the C > > essentially unusable CPUs (no mobo's) through channels one stept) > > shy of a dumpster.  It's particularlyo? > > odd that they aren't being sold with mobo's, since if those  existed A > > they'd be equally valueless without their CPU.  That tells meeC > > that the mobo's were never made and all of these Alpha CPUs areD > > just fancy paperweights. > >W > > Regards, > >F > > David Mathog > > mathog@caltech.eduB > > Manager, Sequence Analysis Facility, Biology Division, Caltech >8 > ; > Hmm, you could be correct, but I thought Samsung made thet motherboards andB > these CPUs, seems unlikely that they would make produce the more	 costly toeD > manufacture CPUs before the motherboards. How come socketed 21264s never_ > appeared on any roadmaps?e >oD >  However, surely someone who worked with API\Samsung\Whoever still browsest5 > this NG and could give us a more definitive answer?C    . Maybe they simply made more chips than mobo's.  E Each wafer yields a certain number of chips, varying with the size ofaF the chip area and wafer size. Of the chips 'cut' from the wafers, someA are defective, some are certifiable at lower clock rates, and the1D reminder are the useful yield at the design clock rate. Perhaps theyC simply had better yields that expected and wound up with more cpu'st than they thought they'd get.g  E And if the chip fab was in Korea and the mobo's were assembled in thetD US, then perhaps they were simply chips in-transit when the plug was$ pulled on the manufacture of mobo's.   Lots of possible explanations.   ------------------------------   Date: 30 May 03 20:02:29 +0200) From: p_sture@elias.decus.ch (Paul Sture);- Subject: Re: Needed, history of startup filesw) Message-ID: <FZkNcLuji8yj@elias.decus.ch>   ^ In article <3ED67FE2.511FFAB@pressenter.com>, Lyndon Bartels <lbartels@pressenter.com> writes:E > I've got a machine where all the startup files are all bulloxed up.t > . > The system is currently runing Alpha VMS 7.2 >  > 2 > It's been upgraded at least once, probably more. >  > H > Anyway the startup files are layers of one .com file doing nothing but? > calling a second file, which does nothing but call a third...t >  > < > If I remember correcly, sylogicals.com, systartup_vms.com,; > lat$systartup.com, etc. have been a relatively new thing.t > I > But I can't remember the genieology of them. When did Lat$systartup.come' > become the prefered file over ltload.  > A > If I remember correctly sylogicals.com superseded setuplog.com.0F > Setuplog.com used to be called from within the systartup.com. But atF > some point, sylogicals, get created, and was called independently of > systartup.com. 6 > 4 > systartup.com was superseded by systartup_vms.com. >  > * > Tell me the story of these files please? >   ; I can't promise that this is 100% accurate, but I'll try...i  A Going back to at least V4.0 (maybe as far back as V2.n) there wasoF systartup.com. I vaguely remember that upgrades overwrote the existingG one, so it was necessary to reapply your customisations to the new one.   K ISTR that this was a reason that some managers chose to keep customisationsv of these files to a minimum.  F With V5, systartup_v5.com was introduced, and at some later date, this: was superceded by systartup_vms.com. Since that timeframe,D files such as systartup_vms.com, sylogicals.com etc were distributedB as *.template for upgrades. I have found it is worth comparing theI .template files after each upgrade for changes (which is how I discovered 0 the then new way of starting DECnet V with V7.1.  9 As for the change from ltload.com to Lat$systartup.com, I D first came across this with an upgrade to V5.4 (it _might_ have beenC V6.2, I just remember which customer it was at this point in time).t   -- 5
 Paul Sture   ------------------------------  # Date: Fri, 30 May 2003 20:34:50 GMT & From: jlsue <jefflsxxxz@sbcglobal.net>" Subject: Re: Portents of VMS death8 Message-ID: <jvffdvgav6k8rb4o6jtgehcq699q5c70fv@4ax.com>  H On Thu, 29 May 2003 16:36:16 -0400, JF Mezei <jfmezei.spamnot@istop.com> wrote:  
 >jlsue wrote: J >> How do you get from a discussion about a specific cluster configuration. >> into a generic, VMS marketplace discussion? >oO >Because not everyone has an application that justifies buying a dozen nodes tomK >make a full fledged cluster where any node can go down without any impact.eJ >There are many who would like to choose VMS for a small scale system thatO >provides production on one node, test on another with production mirrored ontoiK >the test system so that the test system can be used when production fails.w >dL >If you tell them that this isn't what VMS is for, then they will go to UnixO >and Windows. Is the goal to only keep the "crme de la crme" of customers, or0G >to expand the VMS marketplace and capture the HUGE potential for smalls! >clusters in smaller businesses ?F  G If I stated anywhere that only configurations with a dozen nodes is our E target, please show me where.  We are still talking about a specific,b mis-configured imho, cluster..  @ And it's not that is "isn't what VMS is for", it's that it is anK unsupported configuration (potentially).  I still say that putting Dev/test>I in a cluster (security domain) with production would get you hit big-timerB with the auditors I've known.  Especially in the financial sector.   ------------------------------  # Date: Fri, 30 May 2003 20:37:39 GMTd& From: jlsue <jefflsxxxz@sbcglobal.net>" Subject: Re: Portents of VMS death8 Message-ID: <l4gfdv8k72pjemtb4m3dvhrk2nndag95ok@4ax.com>  H On Thu, 29 May 2003 16:40:59 -0400, JF Mezei <jfmezei.spamnot@istop.com> wrote:  
 >jlsue wrote:eG >> In this specific instance?  Not hardly.  They made a decision not tomK >> upgrade to a release that resolves those problems, so they suffered withe	 >> them. . >dN >No. Itis of my opinion that many shops with older versions of VMS have alredyG >made the decision to abandon VMS and are simly continuing to run theirwM >remaining apps on the old VMS box until they are ported to the new platform.F  E Fine, then they can't complain about the performance of the platform.-J In this case, they were doing performance tests of VMS vs Unix, and it wasJ an invalid test because one was old, and the other was newer.  And there'sI no good reason that I can see why they couldn't have tested a more recentDB version of VMS (and TCP/IP) where the performance issues have been
 addressed.   > M >And then there are some for whom upgrade is not possible becayse the ISV haspI >avbandonned VMS and the application is only certified to run at whateverd: >version was available at the time the ISV abandonned VMS.  C Once again, you wander off topic.  My statements are addressed to auK specific test run on specific configurations, and the conclusions that weregI drawn from those tests.  Please stop beat us constantly with you rhetoricn about the VMS marketplace.   ------------------------------  # Date: Fri, 30 May 2003 20:43:21 GMT0& From: jlsue <jefflsxxxz@sbcglobal.net>" Subject: Re: Portents of VMS death8 Message-ID: <bbgfdvcl7hjop41tthms2a66uo1906er8t@4ax.com>  G On 30 May 2003 07:25:56 -0700, mistdragon@zdnetonebox.com (mist dragon)i wrote:     >?D >The case is closed and vms lost it. Reitration of what, how and whyC >does not bear any fruit. There is no way customer is going back to  >vms.,D >However, this case showed very well of how it performs and claiming >thatl< >vms is best performer in the world is just marketing talk.   K Wrong again.  It was a test of a very old version of vms against a new UNIX J offering.  If you're going to continue browbeating VMS performance on thisH test, regardless of the fact that "vms lost it", then you have to expectI that you'll be called on the carpet for a stupid, invalid test.  What yout, ran does not prove that VMS couldn't cut it.  G What testing was run appears to have only been done to proves someone'suG desired outcome.  If they really wanted to test VMS performance on thistH application they would have used later version.  That certainly couldn'tI have been more work than setting up the UNIX version to perform that parte of the test.     >yG >People dont perform major tests for weeks for just fun and everyone onsG >the site loved vms and they had to proof their case against management F >so I doubt very much anyone did nail it down for incompetence or willD >of trying. VMS dvl people did their best afaik but still the system$ >did not reach the figures unix did. >,  I BUT THEY STILL USED OLD VERSION OF VMS AND TCP/IP!!!!!  Can't you see howaI that test was completely invalid as a way of proving that VMS performancehK couldn't match UNIX in this case?  I'm not saying that it could match UNIX,sH but from the test as you've described it, it didn't prove anything otherI than their unwillingness to put forth the same effort for the VMS side ofl) the test as they did to do the UNIX side.t   ------------------------------  , Date: Fri, 30 May 2003 23:20:02 +0200 (CEST)% From: Nomen Nescio <nobody@dizum.com>a" Subject: Re: Portents of VMS death8 Message-ID: <7a05a60a4c854091d3479721419168a2@dizum.com>   > F >And then there are some for whom upgrade is not possible becayse the  ISV hasfA >avbandonned VMS and the application is only certified to run at w	 whatever  > >version was available at the time the ISV abandonned VMS.      C Once again, you wander off topic.  My statements are addressed to ahG specific test run on specific configurations, and the conclusions that t	 were     nA drawn from those tests.  Please stop beat us constantly with you m	 rhetoric n about the VMS marketplace. a    = JF is a miserable french canadian.  Many french canadians aree< miserable.  Take a frenchie and move them to siberia and you@ have a recipe for misery.  JF would not be nearly as cynical andB biting if moved to Miami or New Orleans and warmed up and got some	 sunshine.z   ------------------------------  % Date: Fri, 30 May 2003 23:52:12 +05301 From: "Vivek" <visoni@bmc.com>) Subject: Re: SYS$QIOW - Performance Issue2/ Message-ID: <vdf84ctaoe3mfb@corp.supernews.com>i  H Can we signal to the read channel to start reading once or write is over from the write channel.L            7 "JF Mezei" <jfmezei.spamnot@istop.com> wrote in messageE# news:3ED25328.83065A93@istop.com...C > Vivek wrote:) > > $WRITE TEMP "Deadly chocolate mousse"t > > $READ TEMP testm# > > TEST = "Deadly chocolate mouse"o >eL > Something really fishy here... the MBA driver does not remove the second S > from Mousse :-)C > J > What this means is that at the time you issued the WRITE statement, yourJ > program had not queued an IO (eg: called QIOW) yet, and when you did the READ,e0 > you were the first in line to get that packet. >.K > Now, your program may have been busy processing the previous entry, or itu@ > could have issued some wait statement BEFORE issueing the QIO. > C > It would seem to me that the next step would be to add some printb
 statementsL > (in whatever language you are programming in, eg: printf for c) before andL > after the QIOW statement which will give you a better idea of whether your6 > program is hanging inside the QIOW or outside of it.   ------------------------------   Date: 30 May 2003 19:00 CDTa' From: carl@gerg.tamu.edu (Carl Perkins)-) Subject: Re: SYS$QIOW - Performance Issue0- Message-ID: <30MAY200319004732@gerg.tamu.edu>n  " "Vivek" <visoni@bmc.com> writes...I }Can we signal to the read channel to start reading once or write is over. }from the write channel.  @ You could, I suppose. But it would be a strange way of doing it.@ There are ways of doing this, but to some extent thay ammount toB waiting for a message from one mailbox to indicate that you should@ read from the other mailbox, which would be a silly thing to do.  A A better way would be to hang an asynchronous read on the channel E ($QIO rather than $QIOW). When it completes it will set an event flagdB and/or run an I/O completion AST routine. This is faster, and lessD complicated, than waiting for a signal of some sort and then issuingB a synchronous read - especially if the system has significant loadF from other processes (which can cause a potentially long delay betweenD noticing your signal and the issuing of the $QIOW that doesn't exist in the $QIO method).   --- Carl   ------------------------------  % Date: Sat, 31 May 2003 00:23:36 +0530- From: "Vivek" <visoni@bmc.com>1 Subject: SYS$QIOW - Wtih more Details - Final Tryn/ Message-ID: <vdf9v8pli1d036@corp.supernews.com>w  : I will try to be as explanatory as possible  in this mail.  D Parent process creates mailbox both with read and write permissions.  / And then it forks a child process throgh CREPRCd  @ The child ASSIGNS to the parent mailbox with both read and write permissions.  L Through  the parents channel the parent writes to the mailbox with excellent' speed ( around a millisecond and less).@< The data written through this is arounf 1024 bytes and less.  C The Child reads from the mailbox from its channel but has problems.t@ There is fixed gap b/w each read request from the child process.  H The times taken by the Read request spans from 2 minutes to 10+ minutes.5 During this time the mailbox get more than 30 writes.h- THe process is in LEF state during this time.   1 All the write and reads are using SYS$QIOW calls.u  0 Will be more than happy to provide more details.   ------------------------------  % Date: Sat, 31 May 2003 00:31:28 +0530c From: "Vivek" <visoni@bmc.com>5 Subject: Re: SYS$QIOW - Wtih more Details - Final Tryi/ Message-ID: <vdfae0hfi8pkbe@corp.supernews.com>n  H I forgot to add...on thing...Before the first write call there is a read call from the child process.$ And it takes 16 seconds to complete.        ) "Vivek" <visoni@bmc.com> wrote in message,) news:vdf9v8pli1d036@corp.supernews.com...o< > I will try to be as explanatory as possible  in this mail. >eF > Parent process creates mailbox both with read and write permissions. >01 > And then it forks a child process throgh CREPRCr >:B > The child ASSIGNS to the parent mailbox with both read and write > permissions. >nD > Through  the parents channel the parent writes to the mailbox with	 excellenta) > speed ( around a millisecond and less).t> > The data written through this is arounf 1024 bytes and less. >nE > The Child reads from the mailbox from its channel but has problems.cB > There is fixed gap b/w each read request from the child process. >aJ > The times taken by the Read request spans from 2 minutes to 10+ minutes.7 > During this time the mailbox get more than 30 writes.a/ > THe process is in LEF state during this time.  >v3 > All the write and reads are using SYS$QIOW calls.e >g2 > Will be more than happy to provide more details. >m >n >" >o >e >s >m >r >  >m >a >  >s >E   ------------------------------  # Date: Fri, 30 May 2003 21:30:52 GMTs+ From: Ryan Moore <rmoore@rmoore.dyndns.org>m5 Subject: Re: SYS$QIOW - Wtih more Details - Final Try35 Message-ID: <Pine.LNX.4.53.0305301417340.4502@jaipur>e  G Wait a minute... is this one or two mailboxes?  You need two mailboxes.sJ One that the parent reads from and one that the child reads from.  You canB have them both read from the same mailbox, but those reads will beF serviced in whatever order the reads are queued between the parent andE child.  Probably not what you want.  It would be very easy to lead tol timing deadlocks.a  F If you don't use the IO$M_NOW qualifier, the write QIOW won't complete until that message is read.s  D It sounds like you're using one mailbox.  Let's assume the child andF parent are both writing to the same mailbox with QIOW and no one has aI read queued to the mailbox.  Then that is a deadlock.  Neither write will F complete until a read is issued.  And no read will be issued until the write completes.  I We do communication with mailboxes.  We use async QIO reads that complete F with an event flag.  You may issue the write with QIO and complete theD write asynchronously or you may use QIOW with NOW.  After your write. completes, go back and read from your mailbox.  @ This doesn't sound like a performance problem.  It sounds like a design/synchronization problem.-    ! On Sat, 31 May 2003, Vivek wrote: < > I will try to be as explanatory as possible  in this mail. >uF > Parent process creates mailbox both with read and write permissions. > 1 > And then it forks a child process throgh CREPRCd > B > The child ASSIGNS to the parent mailbox with both read and write > permissions. >hN > Through  the parents channel the parent writes to the mailbox with excellent) > speed ( around a millisecond and less). > > The data written through this is arounf 1024 bytes and less. > E > The Child reads from the mailbox from its channel but has problems.eB > There is fixed gap b/w each read request from the child process. >cJ > The times taken by the Read request spans from 2 minutes to 10+ minutes.7 > During this time the mailbox get more than 30 writes.t/ > THe process is in LEF state during this time.o > 3 > All the write and reads are using SYS$QIOW calls.e >n2 > Will be more than happy to provide more details.   ------------------------------   End of INFO-VAX 2003.299 ************************