1 INFO-VAX	Sun, 28 Dec 2003	Volume 2003 : Issue 719       Contents: Re: DCL user info , Re: Floating point arithmetic support in DCL, RE: Floating point arithmetic support in DCL( Re: hp OpenVMS business solutions 2003 ?( RE: hp OpenVMS business solutions 2003 ?& Re: Is VMS 7.1-2 as stable as 7.1-1H2? Re: length of SCSNODE J Re: OpenVMS Pearl - OpenVMS V8.0, first release on Itanium, ships today!!! Re: RAMDISK or the likes> [Dutch] Book "VAX/VMS Internals and data structures" available  F ----------------------------------------------------------------------  % Date: Sun, 28 Dec 2003 12:06:01 +0100 + From: "Hans Vlems" <hvlems.nieuw@zonnet.nl>  Subject: Re: DCL user info9 Message-ID: <bsmdle$e3tln$1@ID-143435.news.uni-berlin.de>   0 "NEILFREE" <neilfree@aol.com> schreef in bericht4 news:20031227193629.19782.00001174@mb-m01.aol.com...L > Can anyone tell me why the "OWNER" argument with F$GETJPI returns nothing? > ( > $WRITE SYS$OUTPUT F$GETJPI("","OWNER")  J The OWNER attribute returns the process id of the parent process. providedH that there is a parent process. A regular interactive process is its ownK parent (or owner) process and GETJPI returns an empty string. Issue a spawn F command or put this string between PIPE ... &  and you'll see that the function returns an id.    $ spawn ( %DCL-S-SPAWNED, process SYSTEM_1 spawned: %DCL-S-ATTACHED, terminal now attached to process SYSTEM_1' $ WRITE SYS$OUTPUT F$GETJPI("","OWNER")  00000226. $ pipe WRITE SYS$OUTPUT F$GETJPI("","OWNER") & 00000327 $    This was on AXP/VMS V7.3.    Hans   ------------------------------  # Date: Sun, 28 Dec 2003 06:44:52 GMT 0 From: "Guy Peleg" <guy.peleg@hp.com_remove_this>5 Subject: Re: Floating point arithmetic support in DCL 3 Message-ID: <EZuHb.11247$m66.3307@news.cpqcorp.net>   @ DCL symbols are linked to each other using a doubly linked list.@ Each symbol contains a name field, value field and a type field.  G If you have access to the VMS listings, take a look at [DCL]DCLDEF.SDL.    Guy 7 "JF Mezei" <jfmezei.spamnot@istop.com> wrote in message # news:3FEB3DDB.33AC78AC@istop.com...  > Guy Peleg wrote: > > I will look at native  > > floating point supportJ > > and the amount of efforts involved. Who knows maybe we will be able to > > provide it for V8.2......  > L > Wouldn't the average VMS engineer be able to implement this in one evening ? J > And I fully expect you to backport this to VAX VMS 7.2 as well, but I amI > willing to wait an extra day or two before that patch becomes available  :-)  > :-) :-) :-) ;-) :-)  > H > Out of curiosity, what information do you store in the definition of a symbolD > ? Is there a publicly available structure hidden in a library that	 describes 4 > it, or is that available only in VMS source code ?   ------------------------------  % Date: Sun, 28 Dec 2003 05:43:03 -0800 # From: "Tom Linden" <tom@kednos.com> 5 Subject: RE: Floating point arithmetic support in DCL 9 Message-ID: <NDEMLKKEBOIFBMJLCECIGEBKCJAA.tom@kednos.com>    /* -----Original Message----- 8 /* From: Guy Peleg [mailto:guy.peleg@hp.com_remove_this]- /* Sent: Saturday, December 27, 2003 10:45 PM  /* To: Info-VAX@Mvb.Saic.Com8 /* Subject: Re: Floating point arithmetic support in DCL /* /*C /* DCL symbols are linked to each other using a doubly linked list. C /* Each symbol contains a name field, value field and a type field.   E Could you elaborate?  Do you first do a hash lookup and then walk the  list?  Why doubly linked?    /*J /* If you have access to the VMS listings, take a look at [DCL]DCLDEF.SDL. /* /* Guy: /* "JF Mezei" <jfmezei.spamnot@istop.com> wrote in message& /* news:3FEB3DDB.33AC78AC@istop.com... /* > Guy Peleg wrote:  /* > > I will look at native /* > > floating point support B /* > > and the amount of efforts involved. Who knows maybe we will
 /* be able to   /* > > provide it for V8.2...... /* >C /* > Wouldn't the average VMS engineer be able to implement this in  /* one evening /* ?> /* > And I fully expect you to backport this to VAX VMS 7.2 as /* well, but I am L /* > willing to wait an extra day or two before that patch becomes available /* :-) /* > :-) :-) :-) ;-) :-) /* >K /* > Out of curiosity, what information do you store in the definition of a 	 /* symbol G /* > ? Is there a publicly available structure hidden in a library that  /* describes7 /* > it, or is that available only in VMS source code ?  /* /* /* ---) /* Incoming mail is certified Virus Free. = /* Checked by AVG anti-virus system (http://www.grisoft.com). D /* Version: 6.0.553 / Virus Database: 345 - Release Date: 12/18/2003 /* --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).A Version: 6.0.553 / Virus Database: 345 - Release Date: 12/18/2003    ------------------------------    Date: 28 Dec 2003 07:46:46 -08001 From: keithparris_NOSPAM@yahoo.com (Keith Parris) 1 Subject: Re: hp OpenVMS business solutions 2003 ? = Message-ID: <cf15391e.0312280746.6162a600@posting.google.com>   q peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER) wrote in message news:<KEJEb.145799$dt3.95514@news.chello.at>... N > I just stumbled over an old URL again and found it contains many dead links., > 	http://www.lpsolutions.com/ovms2003/html/9 > Is it only my MOZILLA D1.5 or is this really the case ?   ! I see broken links under IE also.   > You might just want to order a copy of the CD for yourself, at4 http://h71000.www7.hp.com/business_solutions_cd.html  D I'd be surprised if a copy of this CD didn't ship with the 7.3-2 kit also.    ------------------------------  % Date: Sun, 28 Dec 2003 08:07:58 -0800 # From: "Tom Linden" <tom@kednos.com> 1 Subject: RE: hp OpenVMS business solutions 2003 ? 9 Message-ID: <NDEMLKKEBOIFBMJLCECICEBMCJAA.tom@kednos.com>    /* -----Original Message----- ; /* From: Keith Parris [mailto:keithparris_NOSPAM@yahoo.com] * /* Sent: Sunday, December 28, 2003 7:47 AM /* To: Info-VAX@Mvb.Saic.Com4 /* Subject: Re: hp OpenVMS business solutions 2003 ? /*   /*  = /* peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER) wrote in  : /* message news:<KEJEb.145799$dt3.95514@news.chello.at>...A /* > I just stumbled over an old URL again and found it contains   /* many dead links. / /* > 	http://www.lpsolutions.com/ovms2003/html/ < /* > Is it only my MOZILLA D1.5 or is this really the case ? /*  $ /* I see broken links under IE also.   Also under Opera.  /*  A /* You might just want to order a copy of the CD for yourself, at 7 /* http://h71000.www7.hp.com/business_solutions_cd.html  /*  G /* I'd be surprised if a copy of this CD didn't ship with the 7.3-2 kit  /* also.   Wasn't in my kit.    /*   /* ---) /* Incoming mail is certified Virus Free. = /* Checked by AVG anti-virus system (http://www.grisoft.com). D /* Version: 6.0.553 / Virus Database: 345 - Release Date: 12/18/2003 /*   --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).A Version: 6.0.553 / Virus Database: 345 - Release Date: 12/18/2003    ------------------------------    Date: 28 Dec 2003 06:55:25 -0800( From: bob@instantwhip.com (Bob Ceculski)/ Subject: Re: Is VMS 7.1-2 as stable as 7.1-1H2? = Message-ID: <d7791aa1.0312280655.51d74cfd@posting.google.com>   q "konabear" <maurert@ameritech.net> wrote in message news:<mq1Hb.19100$P%1.17472731@newssvr28.news.prodigy.com>... L > My point if you're running smoothly on 7.1-1h2, then I wouldn't upgrade toJ > another unsupported (by OpenVMS engineering) version and possibly expose > yourself a new bug.  > K > If there's something specific you need in 7.1-2, then you'll want to test J > the heck out of it first then upgrade, but keep your 7.1-1h2 system diskM > image for a few weeks incase you run into something that causes you to need  > to back out. > N > Arguably 7.1-2 is more stable than 7.1-1h2, but that is theoretical based onE > 7.1-1h2 being a hardware release and 7.1-2 being mostly a rollup of L > patches.But that is a broad generalization; to me it's more important thatD > you already consider 7.1-1h2 stable enough to not need to upgrade. >  > Todd  @ but I am asking becuase you can't run DS10 or 20s on 7.1-1H2 ... so, we must upgrade ...    ------------------------------  % Date: Sun, 28 Dec 2003 08:47:50 -0500   From: John Santos <JOHN@egh.com> Subject: Re: length of SCSNODE4 Message-ID: <1031228084110.998A-100000@Ives.egh.com>  & On Sat, 27 Dec 2003, Hans Vlems wrote:   > N > "Phillip Helbig---remove CLOTHES to reply" <helbig@astro.multiCLOTHESvax.de>3 > schreef in bericht news:bsjpor$gjc$1@online.de... J > > In article <bsjm7h$daf1c$1@ID-143435.news.uni-berlin.de>, "Hans Vlems"$ > > <hvlems.nieuw@zonnet.nl> writes: > > H > > > > > Here's what happens if you do:  Nothing.  VMS pads it for you.	 > > > > > 5 > > > > > MAZDA$ sea sys$system:modparams.dat scsnode  > > > > > SCSNODE="MAZDA" , > > > > > MAZDA$ scsnode = f$gets("SCSNODE")  > > > > > MAZDA$ sho sym scsnode  > > > > >   SCSNODE = "MAZDA   " > > > > : > > > > I see no padding on VAX 7.3.  I do on ALPHA 7.3-1.   (Late followup!)   I *DO* see padding on VAX 7.3    $ show sys/nopr G OpenVMS V7.3  on node BIRD  28-DEC-2003 08:42:43.14  Uptime  5 03:44:32 
 $ show cpu   BIRD, a MicroVAX 3600 SeriesH Multiprocessing is DISABLED. Uniprocessing synchronization image loaded.   PRIMARY CPU = 00 Active CPUs:      00 Configured CPUs:  00) $ search sys$system:modparams.dat scsnode  SCSNODE="BIRD"! $ mcr sysgen SYSGEN>  SHO SCSNODE I Parameter Name            Current    Default     Min.     Max.     Unit    Dynamic I --------------            -------    -------    -------  -------   ----    ------- F SCSNODE                   "BIRD  "    "    "    "    "    "ZZZZ" Ascii SYSGEN>   Exit $ scsnode = f$gets("SCSNODE")  $ sho sym scsnode    SCSNODE = "BIRD  "  C Node name appears padded to 6 characters with spaces both in SYSGEN  and f$getsyi() output.   > > > > 6 > > > Interesting. Did you see my post on VAX/VMS 7.2? > >  > > No, must have missed it. > >  > OK, a copy :-) > M > If you run DECnet then the nodename must be the same as SCSNODE. The SYSGEN J > help (on VAX/VMS 7.2) states that the name may be up to 6 characters. It. > also warns not to set it to the NULL string.G > I never use padding in MODPARAMS but the end up in SYSGEN padded with 	 > blanks:  >  > $ mc sysgen sh scsnodeI > Parameter Name            Current    Default     Min.     Max.     Unit 	 > Dynamic N > --------------            -------    -------    -------  -------   ----  --- > ----H > SCSNODE                   "XENON "    "    "    "    "    "ZZZZ" Ascii > $ ty sys$system:modparams.dat  > WINDOW_SYSTEM   =1I > ! CLUSTER_CONFIG appending for ADD operation on 10-NOV-2001 12:46:53.54  > VOTES           = 1  > expected_votes  = votes  > DISK_QUORUM     ="" < > AGEN$INCLUDE_PARAMS SYS$MANAGER:AGEN$NEW_NODE_DEFAULTS.DAT > SCSNODE         ="XENON" > SCSSYSTEMID     =1026  > C > There is a trailing blank in the SYSGEN output, difficult to see. ! > As said : this is VAX/VMS V7.2.  >  > Merry Christmas Philip,  >  > Hans   --   John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------    Date: 28 Dec 2003 07:21:40 -08001 From: keithparris_NOSPAM@yahoo.com (Keith Parris) S Subject: Re: OpenVMS Pearl - OpenVMS V8.0, first release on Itanium, ships today!!! = Message-ID: <cf15391e.0312280721.3bac01cc@posting.google.com>   v koehler@eisner.nospam.encompasserve.org (Bob Koehler) wrote in message news:<W8vBwWOz+4zv@eisner.encompasserve.org>...2 >    can we get a hobbyist license to go with thatB >    inexpensive, used Integrity Server we just picked up on eBay?  D The Hobbyist program has been quite successful on VAX and Alpha, andD has strong support from HP, so I'm fully confident that the HobbyistA program will be extended to include Itanium systems, in due time.   0 A quick check on eBay shows no used rx2600s yet.   ------------------------------    Date: 28 Dec 2003 08:44:39 -08001 From: keithparris_NOSPAM@yahoo.com (Keith Parris) ! Subject: Re: RAMDISK or the likes = Message-ID: <cf15391e.0312280844.798822ee@posting.google.com>   a brandon@dalsemi.com (John Brandon) wrote in message news:<03121917002912@dscis6-0.dalsemi.com>... P > I have a ESA12000 with HSG80 controllers (paired) using 9.1, 18.2, and 36.4 GB > disk drives. > B > We have a lot of locking (250% increase over the last 6 months).  E Good data point (I'd guess it reflects an increase in workload of the = same magnitude over the same time period, but does that sound E reasonable to you?). In any case, it doesn't help you know if locking & itself represents a bottleneck or not.  + Locking can be a bottleneck in cases where: F o  Locking is local, but the rate is so high you run out of CPU due toB the high interrupt-state load, or too much CPU time gets wasted in
 MP-Synch time ? o  Locking is remote, and the adapter(s) or interconnect become 
 saturated, or @ o  The lock-master node for a given lock tree becomes overloaded handling the lock requests  C If you want to quantify the lock request workload, you can use SDA> C LCK SHOW ACTIVE on each node or get a cluster-wide summary from the @ LOCK_ACTV tool on the V6 Freeware CD [KP_LOCKTOOLS] directory). F MONITOR MODES can be used to identify saturation of a CPU in interrupt state.  C I/O bottlenecks often result in queues of requests building to busy E disk devices.  Various performance tools (including MONITOR DISK) can  display these queues.   C Both locking and I/O bottlenecks can result in lock queues.  An I/O E bottleneck can give the same symptoms as a locking bottleneck because F since a lock tends to be held while an I/O is done, if it takes a longF time to complete an I/O, lock queues can build up from other processes waiting for the same lock.  ? Lock queues can be found with the Lock Contention capability of C Availability Manager / DECamds, or with the LCKQUE tool from the V6  Freeware CD.   > Running FOCUS and DBMS-32. > J > I have been thinking about ideas to help increase the performance on the1 > DBMS-32 and the subject of RAMDISK has come up.  > K > I can think of two ideas here, 1) Solid State disk and 2) Disk in Memory.   A I find the term "RAMDISK" confusing, because I've seen it used to A refer both to hardware products which store data in RAM chips but D which interface to a system with a disk-style interface (like the EZF series of solid-state disks), and also to refer to disk pseudo-driversE which present a chunk of system main memory to software as if it were ! a disk (like the DECram product).   @ Either one can be a way to reduce I/O latency by eliminating theA head-positioning time and rotational latency inherent in magnetic  disks.  C Note that you get some of the same sort of benefit as a solid-state ? disk from HSG80 cache memory -- reads which hit in cache can be A satisfied with response times similar to that of a SSD, and write E which go into write-back cache also provide response times similar to F that of a SSD (and even random write I/Os benefit, whereas the benefitF for reads depends on access patterns and how many reads "hit" in cacheD memory).  With solid-state disk, it's sort of like an entire disk isD cached. But because SSDs are more expensive than magnetic disks, youF generally have to decide what pieces of the data you'll put on SSD andF what you'll leave on magnetic disk.  With cache, that decision is madeC by the controller.  Also, SSDs often come with battery backup and a : magnetic disk to preserve the data across a power failure.  A With something like DECram, you use some amount of main memory to C store data.  Since this goes away if the system fails, unless it is A just used to contain a copy of read-only data, you'd follow Rob's A advice and shadow it with a magnetic disk (or SSD).  Reads can be D directed to the DECram disk and get higher performance, while writesF have to go to both, and will be slower.  As with SSDs, you as a system? manager have to determine what and how much to put into DECram.   F Assuming you have enough available main memory, you can get a softwareE loan-of-products PAK for DECram and try that out without any up-front  cost.   F The VTDPY utility on the HSG80s can help you determine the average I/OC size (data rate divided by request rate), read/write ratio, and the C current hit rates in the HSG80 cache (to get an idea of locality of F reference of the request stream and how effectively it is being cached at present).   ------------------------------  % Date: Sun, 28 Dec 2003 16:54:00 +0100 B From: Michiel Erens <I.dont.want.spam@this.mailaddress.is.invalid>G Subject: [Dutch] Book "VAX/VMS Internals and data structures" available 7 Message-ID: <3FEEFC98.4C1C@this.mailaddress.is.invalid>    [Dutch] 5 Voor de liefhebber heb ik een exemplaar van het boek  8 "VAX/VMS internals and data structures version 5.2" van  Goldenberg/Kenah liggen.   Gratis af te halen in Delft.  9 Als je wel geinteresseerd bent, maar Delft te ver vindt,  9 wil ik m tegen voorwaarden wel opsturen. Maar een afhaler  heeft voorrang.   1 Mail naar erensm @ ilse.nl als je interesse hebt.    --   ME Posted by news://news.nb.nu    ------------------------------   End of INFO-VAX 2003.719 ************************