0 INFO-VAX	Mon, 13 Feb 2006	Volume 2006 : Issue 87      Contents:% Re: cluster-wide logicals and startup % Re: cluster-wide logicals and startup % Re: cluster-wide logicals and startup * Re: defining logicals which point to disks) Re: does OPERATOR.LOG contain everything? ) Re: does OPERATOR.LOG contain everything? ) Re: does OPERATOR.LOG contain everything? = Re: Gartner wakes up company executives to X86-64 scalability = RE: Gartner wakes up company executives to X86-64 scalability = Re: Gartner wakes up company executives to X86-64 scalability = RE: Gartner wakes up company executives to X86-64 scalability = Re: Gartner wakes up company executives to X86-64 scalability = Re: Gartner wakes up company executives to X86-64 scalability = Re: Gartner wakes up company executives to X86-64 scalability ; Re: how to check for the existence of a logical-name table?  Re: null terminated strings  Re: PDPs in the news!  Re: Perl, DBI, and DBD::RDB  Re: Perl, DBI, and DBD::RDB   F ----------------------------------------------------------------------  % Date: Sun, 12 Feb 2006 21:19:20 +0100 2 From: martin@radiogaga.harz.de (Martin Vorlaender). Subject: Re: cluster-wide logicals and startup; Message-ID: <43ef9848.524144494f47414741@radiogaga.harz.de>    Phillip Helbig wrote: 7 > martin@radiogaga.harz.de (Martin Vorlaender) writes:  K >> 1. If F$GETSYI("CLUSTER_MEMBER") is false, the node isn't in a cluster - & >>    there'll be no cluster logicals. > H > OK, that's what I was thinking, though JF says that LNM$SYSCLUSTER is K > available even on standalone nodes.  I don't have a standalone node so I   > can't check it.   = JF is right. Just looked it up in the Cluster Systems manual:    5.4 Clusterwide Logical Names  ... A Because LNM$CLUSTER and LNM$SYSCLUSTER_TABLE exist on all systems F running OpenVMS 7.2 or later, the programs and command procedures thatB use clusterwide logical names can be developed, tested, and run on nonclustered systems.    Sorry for the confusion.   cu,    Martin --  D One OS to rule them all       | Martin Vorlaender  |  OpenVMS rules!7 One OS to find them           | work: mv@pdv-systeme.de J One OS to bring them all      |   http://www.pdv-systeme.de/users/martinv/> And in the Darkness bind them.| home: martin@radiogaga.harz.de   ------------------------------  % Date: Sun, 12 Feb 2006 15:50:07 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> . Subject: Re: cluster-wide logicals and startup, Message-ID: <43EF9F7E.880A9C3E@teksavvy.com>  / Phillip Helbig---remove CLOTHES to reply wrote: J > Suppose the cluster is booting.  If I am the first node to boot, I might8 > be the only node at first, but later others will join.  G Correct. But this is a concern for more than logicals. You also need to  look at disks, queues etc.E (This is especialy true with MSCP served shadowsets since waiting for H all members to be present allows them to be mounted at once and not cost" a copy/merge/minimerge operation).  G So you could loop with F$CSID for a max of say 5 minutes to wait to see H if other nodes haver joined the cluster. If so, you know that you should  also wait for logicals later on.   ------------------------------  % Date: Sun, 12 Feb 2006 16:22:54 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> . Subject: Re: cluster-wide logicals and startup, Message-ID: <43EFA72C.E5A0F6FB@teksavvy.com>   Martin Vorlaender wrote:? > JF is right. Just looked it up in the Cluster Systems manual:   H Oh Shucks.... I consumed my annual single "JF Is right" credit  so earlyI in the year... now for the rest of the year, I won't be right anymore :-)   F Actually the credit goes to one of the VMS engineers because I asked aF similar question in the past and he answered it with confirmation that, the clusterwide table exists on all systems.   ------------------------------    Date: 12 Feb 2006 11:54:24 -0800$ From: "AEF" <spamsink2001@yahoo.com>3 Subject: Re: defining logicals which point to disks C Message-ID: <1139774064.771452.140050@g47g2000cwa.googlegroups.com>   / Phillip Helbig---remove CLOTHES to reply wrote: E > In article <1139713556.691165.210700@f14g2000cwb.googlegroups.com>, ( > "AEF" <spamsink2001@yahoo.com> writes: > E > The stuff further down looks OK, but these examples seem beside the  > point. >  > > Example: > >  > > $ DEFINE LC LOGIN.COM  > F > The logical points to a file.  I was discussing having it point to a > directory.  & You had it point to a "PHYSICAL_DISK".   > > > > %DCL-I-SUPERSEDE, previous value of LC has been superseded > > $ DIR LC:[]  > ' > I'm surprised this even works at all.   F It works as long as no file-spec component is specified twice: once inE the definition of the logical name and once in the part following the & colon. So DIR LC:BLAH.COM would fail.    AEF    ------------------------------  + Date: Sun, 12 Feb 2006 18:40:47 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)2 Subject: Re: does OPERATOR.LOG contain everything?$ Message-ID: <dsnvfe$qpk$1@online.de>  H In article <43ef709a$1@news.langstoeger.at>, peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER) writes:    y > In article <dsnfpk$hg1$1@online.de>, helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) writes: K > >In article <43ef4275$1@news.langstoeger.at>, peter@langstoeger.at (Peter   > >'EPLAN' LANGSTOEGER) writes:  > > ( > >> 3) OPERATOR.LOG is created on boot  > >  > >When, exactly?  > M > By running SYS$STARTUP:VMS$CONFIG-050_OPCOM.COM in the SYSMAN CONFIG phase.  > C > $ TYPE SYS$STARTUP:VMS$PHASES.DAT	!list all SYSMAN startup phases 5 > $ MCR SYSMAN				!list all SYSMAN startup procedures 2 > SYSMAN> STARTUP SET DATABASE STARTUP$STARTUP_VMS  > SYSMAN> STARTUP SHOW FILE/FULL6 > SYSMAN> STARTUP SET DATABASE STARTUP$STARTUP_LAYERED  > SYSMAN> STARTUP SHOW FILE/FULL > 3 > btw: Don't forget to check SYS$SYSTEM:STARTUP.COM  > O > In my case, boottime is about 39sec earlier than creation time of the OP*.LOG   C I'm defining the name of the log file via SYLOGICALS.COM.  Is this  G before or after the operator log is created?  If after, does a new one  H get opened at the defined location?  If before, can I count on it being  always before?   ------------------------------    Date: 12 Feb 2006 20:31:31 +01006 From: peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER)2 Subject: Re: does OPERATOR.LOG contain everything?, Message-ID: <43ef9b23$1@news.langstoeger.at>  w In article <dsnvfe$qpk$1@online.de>, helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) writes: D >I'm defining the name of the log file via SYLOGICALS.COM.  Is this H >before or after the operator log is created?  If after, does a new one I >get opened at the defined location?  If before, can I count on it being   >always before?    How about checking yourself ? K SYLOGICALS is run in the DEVICES phase (SYS$STARTUP:VMS$DEVICE_STARTUP.COM) * which is 2 phases before the CONFIG phase.  ! $ TYPE SYS$STARTUP:VMS$PHASES.DAT  INITIAL  DEVICES 	 PRECONFIG  CONFIG BASEENVIRON  LPBEGIN  LPMAIN LPBETA END   C Define STARTUP_P2 as "DC" and see in your SYS$SYSTEM:STARTUP.LOG...    --   Peter "EPLAN" LANGSTOEGER % Network and OpenVMS system specialist  E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------  % Date: Sun, 12 Feb 2006 15:53:03 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> 2 Subject: Re: does OPERATOR.LOG contain everything?+ Message-ID: <43EFA02E.BF7D325@teksavvy.com>   / Phillip Helbig---remove CLOTHES to reply wrote: & > > 3) OPERATOR.LOG is created on boot >  > When, exactly?  B There is a message on opa0 when opcom is created. It is done afterG sylogicals.com has been executed (since it looks at logicals defined in 2 it), but before systartup_vms.com has been called.   ------------------------------  % Date: Sun, 12 Feb 2006 15:04:04 -0500 ( From: Bill Todd <billtodd@metrocast.net>F Subject: Re: Gartner wakes up company executives to X86-64 scalability= Message-ID: <rJmdnR3EzMQoCXLeRVn-uQ@metrocastcablevision.com>    Karsten Nyblad wrote:    ...   I > I have a hunch that many of IBM's excellent results on big servers are  J > due to DB2 scaling better than Oracle.  I have never seen any benchmark G > results to confirm it, e.g. benchmarks on Oracle and DB2 on the same   > machines.   D Well, TPC-C results provide at least an approximation of that.  The I first 8-core POWER5 results did give DB2 about a 16% edge over Oracle (I  B tend to suspect because IBM knew a lot more about how to tune DB2 F optimally on the platform than it did Oracle), but later on an Oracle E 32-core system achieved very close to 50% the score of a earlier DB2  E 64-core system (and the DB2 POWER5 results are *very* linear in core  E count).  It's possible, though hardly certain, that the 64-core, and  I other, DB2 results could have been improved at that later date given the  H additional tuning time and experience had IBM made the effort to update G them - which means you might want to add 'identical tuning experience'   to your criteria above.    - bill   ------------------------------  % Date: Sun, 12 Feb 2006 19:34:24 -0500 ' From: "Main, Kerry" <Kerry.Main@hp.com> F Subject: RE: Gartner wakes up company executives to X86-64 scalabilityR Message-ID: <FD827B33AB0D9C4E92EACEEFEE2BA2FB7D3AB1@tayexc19.americas.cpqcorp.net>   > -----Original Message-----4 > From: Alan Greig [mailto:greigaln@netscape.net]=20" > Sent: February 12, 2006 11:51 AM > To: Info-VAX@Mvb.Saic.Com H > Subject: Re: Gartner wakes up company executives to X86-64 scalability >=20 >=20 >=20 > Main, Kerry wrote: >=20 > >=20G > > Yep, meant to say that (reason why I gave Exchange as example), but ? > > another good example is that many Windows (and as I have=20  > been told Linux G > > as well) applications are either single or have very few threads ..  >=20B > Which is one reason Sun realises that Solaris on scalable X64=20 > systems is=20 B > very attractive to many. But the scalability of Windows/Linux=20 > apps isn't=20  > standing still.  >=20> > Just came across a report that said Intel claimed the new=20 > quad core X64=20J > (demo'd in public a couple of days ago) is intended to scale up to at=20 > least 32 cores on a chip:  >=20@ > http://www.engadget.com/2006/02/10/intel-shows-off-first-quad- > core-chip/ >=20= > "(Intel CTO) Rattner says the Clovertown architecture is=20  > scalable to 32=20 I > or more cores on a single chip, which should satiate our speed jones=20 I > until quantum computers finally go mainstream. According to Rattner,=20 F > chips with hundreds of cores are theoretically possible with this=20> > technology, which means we'll probably be able to do some=20 > pretty serious=20 & > damage on timed games of Solitaire." >=20 >=20 > --=20  > Alan Greig     Alan,   @ Keep in mind that the average utilization today for most Windows? applications in prime time is approx 10-15%. Most CIO's are not E impressed at all today with faster hardware. They already have way to  much.   H Part of this is the one-app, one server model culture and part is due toF Windows/Linux culture of not wanting to share applications on the same( system (app stacking vs OS stacking).=20  F The average UNIX server utilization in prime time is a bit better i.e.> 15-30%, but is still a long way from the target 60-80% average; utilization goal that most companies would like to achieve.   @ So, while there will no doubt be some applications that can takeG advantage of newer hardware, telling a CIO that he can lower his server C utilization from 10-15% to something lower is not going to get much E attention. For the Windows and Linux world, virtualization techniques F like VMware will only reduce HW costs a certain amount - it still doesE not reduce the number of OS instances which in the end is tied to FTE 0 counts (the biggest cost in an IT organization).  F My point is that server hardware will not be replaced as it was in the? past (one for one refresh) as current servers are grossly under F utilized. The challenge with much faster HW is that the culture issuesG start raising their ugly heads as App stacking starts to get considered G again as a means to reduce costs and increase server utilization rates.        Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------  % Date: Sun, 12 Feb 2006 20:20:27 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> F Subject: Re: Gartner wakes up company executives to X86-64 scalability, Message-ID: <43EFDED9.36225448@teksavvy.com>   "Main, Kerry" wrote:  B > Keep in mind that the average utilization today for most Windows/ > applications in prime time is approx 10-15%.      3 There are many issues which you are not mentioning:   D The CPU may be at 15% average, but what about peak times ? You couldE have staff twiddling their thumbs for one hour,. Then they all get an C email, you want enough CPU to be able to serve these clients within  reasonable time.    D Secondly, CPU utilisation isn't the only thing. You need to considerH disk acesses/paths, as well as networking paths.  With switches, you canH have 2 CPUs serve 2 users at full network bandwidth. Moved this into one/ server and each client gets half the bandwidth.   @ And consider VMS clustering. You need 3 nodes to make clusteringH worthwile so you can do rolling upgrades and shut down each machine. ButH that may result in average CPU that is much lower than necessary, but it gives you additional uptime.  D Same with Tandem where each CPU is load balanced so that it can takeP over the tasks of another CPU shoudl the later fail. So it usually is underused.   ------------------------------  % Date: Sun, 12 Feb 2006 22:00:28 -0500 ' From: "Main, Kerry" <Kerry.Main@hp.com> F Subject: RE: Gartner wakes up company executives to X86-64 scalabilityR Message-ID: <FD827B33AB0D9C4E92EACEEFEE2BA2FB7D3AB3@tayexc19.americas.cpqcorp.net>   > -----Original Message-----9 > From: JF Mezei [mailto:jfmezei.spamnot@teksavvy.com]=20 ! > Sent: February 12, 2006 8:20 PM  > To: Info-VAX@Mvb.Saic.Com H > Subject: Re: Gartner wakes up company executives to X86-64 scalability >=20 > "Main, Kerry" wrote: >=20D > > Keep in mind that the average utilization today for most Windows3 > > applications in prime time is approx 10-15%.=20  >=20 >=205 > There are many issues which you are not mentioning:  >=20F > The CPU may be at 15% average, but what about peak times ? You couldG > have staff twiddling their thumbs for one hour,. Then they all get an E > email, you want enough CPU to be able to serve these clients within  > reasonable time.=20  >=20F > Secondly, CPU utilisation isn't the only thing. You need to consider; > disk acesses/paths, as well as networking paths.  With=20  > switches, you can ? > have 2 CPUs serve 2 users at full network bandwidth. Moved=20  > this into one 1 > server and each client gets half the bandwidth.  >=20B > And consider VMS clustering. You need 3 nodes to make clustering@ > worthwile so you can do rolling upgrades and shut down each=20 > machine. But; > that may result in average CPU that is much lower than=20  > necessary, but it  > gives you additional uptime. >=20F > Same with Tandem where each CPU is load balanced so that it can take? > over the tasks of another CPU shoudl the later fail. So it=20  > usually is underused.  >=20   JF ..   F Pick up any Customer RFP these days and they will emphasize that their@ servers are grossly under utilized. To paraphrase a recent largeD financial company external RFP - "we have approximately 8,000 WintelF servers and have determined that the average utilization in prime time is approx 10%."   H To put this in another light - ask any CIO if they would be happy if all& of their employees were only 10% busy.  H Btw, you only need 2 OpenVMS nodes to do rolling upgrades. You do need 2 system disks + quorum disk.    :-)    Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------  % Date: Sun, 12 Feb 2006 23:47:39 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> F Subject: Re: Gartner wakes up company executives to X86-64 scalability, Message-ID: <43F00F5C.4BA742BE@teksavvy.com>   "Main, Kerry" wrote:   H > Pick up any Customer RFP these days and they will emphasize that their% > servers are grossly under utilized.   C One would need to go back to the initial decisions to buy that many H machines and the justification to do so. Microsoft had guidelines on howF many MSMail users per machine, followed by how many Exchange users perE machine. It could be that initially, they could only put 50 users per E machine and upgraded all those machines even though exhcnage might be F able to handle 100 users per machine. Also, from a management point of@ view, merging different machines into one is not an easy task if& Microsoft doesn't have tools to do so.  D So keeping underitilised machines may in fact cost less than the man5 hours required to simplify the fleet of wintel boxes.       F > financial company external RFP - "we have approximately 8,000 WintelH > servers and have determined that the average utilization in prime time > is approx 10%."   B I'd bet that they used to have 8000 smaller machines whose average= utilisation was  90% back when machines were 10 times slower.   J > To put this in another light - ask any CIO if they would be happy if all( > of their employees were only 10% busy.  G Ask human resources folks if they are happy dedicating a team of highly C paid experts to combine 2 servers and save a few hundred dollars of  maintenance per year.   J > Btw, you only need 2 OpenVMS nodes to do rolling upgrades. You do need 2 > system disks + quorum disk.   H Which means you need a "neutral" fisk array that finctions independantly of either node.    ------------------------------  % Date: Mon, 13 Feb 2006 00:10:30 -0500 ( From: Bill Todd <billtodd@metrocast.net>F Subject: Re: Gartner wakes up company executives to X86-64 scalability= Message-ID: <B4ednThF-LJaiW3eRVn-rg@metrocastcablevision.com>    Main, Kerry wrote:   ...   H > Pick up any Customer RFP these days and they will emphasize that theirB > servers are grossly under utilized. To paraphrase a recent largeF > financial company external RFP - "we have approximately 8,000 WintelH > servers and have determined that the average utilization in prime time > is approx 10%."   F BFD.  The *real* question is whether that level of utilization is any @ kind of problem per se, or whether their problems lie elsewhere.  ? 8,000 servers sounds rather like separate servers for separate  I departments - or even just groups within departments.  If they need this  G separation for some reason, then there's likely no utilization 'issue'  H at all:  it may well not be *possible* to buy servers weak enough to be H more fully utilized given the loads they're placing on them, so they're C somewhere close to optimal 'utilization' given the nature of their   environment and workload.   F Of course, if they could consolidate those servers (but have not done G so) that's an entirely different matter - and the price they're paying  F in increased maintenance for so many individual systems likely dwarfs F any price they're paying for sub-optimal utilization (i.e., what they H would save if they could substitute 8,000 slightly cheaper servers with G 'better' utilization).  And even if it remains prudent to run only one  I server application per instance of Windows, VMWare-like approaches offer  E an eminently feasible path toward such consolidation even in Windows  
 environments.    > J > To put this in another light - ask any CIO if they would be happy if all( > of their employees were only 10% busy.  G Rubbish.  An employee's salary plus benefits and overhead for a single  G year runs close to two full orders of magnitude more than the purchase  E price (which covers multiple years of service, but there's of course  H maintenance to consider as well) for the kind of low-end servers you're E most likely talking about here (given that there are 8,000 of them).  I Inefficiencies that would never be tolerable at employee cost levels may  H be entirely tolerable in low-end servers if there are other reasons for $ partitioning the workload so finely.   - bill   ------------------------------  % Date: Mon, 13 Feb 2006 01:10:28 -0500 ( From: Bill Todd <billtodd@metrocast.net>F Subject: Re: Gartner wakes up company executives to X86-64 scalability= Message-ID: <TsydnUlj8uJIv23eRVn-gQ@metrocastcablevision.com>    JF Mezei wrote:  > "Main, Kerry" wrote:   ...   K >> Btw, you only need 2 OpenVMS nodes to do rolling upgrades. You do need 2  >> system disks + quorum disk. > J > Which means you need a "neutral" fisk array that finctions independantly > of either node.   @ No, it means exactly what it says:  you just need a quorum disk # accessible from both cluster nodes.    - bill   ------------------------------  # Date: Sun, 12 Feb 2006 21:19:09 GMT " From:   VAXman-  @SendSpamHere.ORGD Subject: Re: how to check for the existence of a logical-name table?0 Message-ID: <00A51378.842F29D0@SendSpamHere.ORG>  w In article <dsmv2d$5nh$3@online.de>, helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) writes:  >  > : >In article <00A512BC.3B2F15BA@SendSpamHere.ORG>, VAXman-  >@SendSpamHere.ORG writes:   > H >> Does $ SHOW LOGICAL/STRUCTURE show you what you want?  Try PIPE and a
 >> Search. > G >OK, that will work.  I prefer using a lexical function to parsing the  J >output of SEARCH, but that doesn't seem possible here.  (I also use PIPE H >and SEARCH to check to see whether a node has the TCPIP cluster alias.) > 1 >> Or, try to create a logical name in the table:  >>  , >> $ DEFINE/USER/TABLE=SYSMAN$NODE_TABLE X Y7 >> %SYSTEM-F-NOLOGTAB, no logical name table name match  >>  I >> If the error is return, create the table.  If not, remove the spurious  >> logical name X. > G >Well, that gets me a message, which I would prefer not to have.  If I  G >were happy with that, I could just create the table and if it already  E >exists I just get an informational message rather than "-F-", which  I >looks less disturbing.  (Actually, I think that this should be "-W-" or  B >even "-E-" and not "-I-".  What do other folks think about this?) >   H Use Graham Burley's suggestion anbout F$TRNLNM.  I thought that F$TRNLNMH would return this but when I tried it I user LNM$SYSTEM_TABLE, not LNM$-H SYSTEM_DIRECTORY and, of course, it failed.  Too late and too many pints to be posting, I suppose.    --  K VAXman- A Bored Certified VMS Kernel Mode Hacker   VAXman(at)TMESIS(dot)COM              5   "Well my son, life is like a beanstalk, isn't it?"     ------------------------------    Date: 12 Feb 2006 11:47:10 -0800$ From: "AEF" <spamsink2001@yahoo.com>$ Subject: Re: null terminated stringsC Message-ID: <1139773630.025700.291740@g44g2000cwa.googlegroups.com>    Dave Froble wrote: > JF Mezei wrote:  > > Dave Froble wrote: > > 1 > >>C offers too many opportunities for mistakes.  > >  > > L > > Why do they still make manual transmission cars ? That clutch allows too$ > > many opportunities for mistakes. > > L > > Yes, if you are writing a nice business applications, your all-automaticB > > Cadillac with head-up display on widnshield, IR cameras to seeG > > people/animals at night and all the gadgets you need will make your ) > > business application easier to write.  > > H > > But your Cadiallac isn't good to get off road, or on a safari in theL > > desert where sand will clog your air filter and your gadgets won't be of
 > > much use.  > > I > > C may not have some of the features that make other langages superior K > > for certain applications, but it has greater versatility. And because C L > > allows you to go off-road (dealing with pointers), you need to have more8 > > discipline because things can break if you abuse it. > I > Any argument such as you raise here can be turned around and used as an H > argument for using assembly language instead of C.  If you're going toI > hold sacred absolute versitility, then quit using C and use an assembly  > language.  > I > Now waiting for you to come back with "C is portable and implemented on ) > many different systems".  Don't bother.     D [Lame-attempt-at-humor warning!: Not for the humor-impaired, and not for the bad-humor--impaired:]   E Hell, why stop at assembly language? Buy some raw materials, fab your G own chips, and build your computer anyway you like! Versatility is then D limited only by your knowledge of physics, engineering, and computerG science; and your budget. Then you'll really not be just a user; you'll  be the builder.    ------------------------------    Date: 12 Feb 2006 21:51:51 -05003 From: Rich Alderson <news@alderson.users.panix.com>  Subject: Re: PDPs in the news!. Message-ID: <mddd5hs3r94.fsf@panix5.panix.com>  4 "Dave Weatherall" <djw-nothere@nospam.nohow> writes:  2 > On Sat, 11 Feb 2006 20:45:59 UTC, Rich Alderson ( > <news@alderson.users.panix.com> wrote:  L >> I know someone who is actively looking for an 11/60 (not me, not anyone IO >> work for).  Please contact me off-line so I can put the two of you in touch.    > It is in Munich Rich.   8 Thanks, Dave.  I checked; he thinks Munich is a bit far.  1 (He usually drives the stuff he buys himself. :-)    --  L Rich Alderson                                       | /"\ ASCII ribbon     |L news@alderson.users.panix.com                       | \ / campaign against |L "You get what anybody gets. You get a lifetime."    |  x  HTML mail and    |L                          --Death, of the Endless    | / \ postings         |   ------------------------------  % Date: Sun, 12 Feb 2006 14:26:49 -0600 6 From: "Craig A. Berry" <craigberry@mac.com.spamfooler>$ Subject: Re: Perl, DBI, and DBD::RDBD Message-ID: <craigberry-48683C.14264912022006@news.isp.giganews.com>  D In article <craigberry-DDB224.13452111022006@news.isp.giganews.com>,8  "Craig A. Berry" <craigberry@mac.com.spamfooler> wrote:  I > In article <dsl7ls02j63@enews4.newsguy.com>, healyzh@aracnet.com wrote:   P > > Or does anyone know where to find a version of DBI that will build with PerlF > > 5.8-6 under OpenVMS 7.3-2?  My attemps to build DBI 1.50 have been > > questionable :^( > > $ > > 3 tests and 45 subtests skipped.L > > Failed 45/49 test scripts, 8.16% okay. 1387/1896 subtests failed, 26.85%	 > > okay.  > F > I haven't tried this version, so I don't know if your experience is C > unique.  Quite often there is some very simple piece of the test  H > infrastructure that causes everything to fail, and once ported all is  > well.   F Hmm.  I just built DBI 1.50 and encountered no issues at all, and all C tests passed successfully.  I did this with both Perl 5.8.4 and an  D up-to-the-minute development snapshot.  Here's my build environment:  	 $ cc/vers # HP C V7.1-015 on OpenVMS Alpha V8.2  $ mmk/ident 5 %MMK-I-IDENT, this is the MadGoat Make Utility V3.9-6 F -MMK-I-COPYRIGHT, Copyright  1992-2002, MadGoat Software.  All Rights 	 Reserved.   C It's unlikely the differences in Perl version or VMS version would  G cause the problems you are seeing, though I suppose it's possible.  If  E you make the following change to the generated descrip.mms and rerun  E the tests, you should get a lot more detail about what's not working:   0 --- descrip.mms;-0      Sun Feb 12 14:03:08 2006( +++ descrip.mms Sun Feb 12 14:23:55 2006 @@ -1006,7 +1006,7 @@     # --- MakeMaker test section:   -TEST_VERBOSE = 0  +TEST_VERBOSE = 1   TEST_TYPE = test_$(LINKTYPE)   TEST_FILE = test.pl  TESTDB_SW = -d    ------------------------------   Date: 13 Feb 2006 04:16:28 GMT From: healyzh@aracnet.com $ Subject: Re: Perl, DBI, and DBD::RDB, Message-ID: <dsp16s012bf@enews1.newsguy.com>  5 Craig A. Berry <craigberry@mac.com.spamfooler> wrote: H > Hmm.  I just built DBI 1.50 and encountered no issues at all, and all E > tests passed successfully.  I did this with both Perl 5.8.4 and an  F > up-to-the-minute development snapshot.  Here's my build environment:   > $ cc/vers % > HP C V7.1-015 on OpenVMS Alpha V8.2 
 > $ mmk/ident 7 > %MMK-I-IDENT, this is the MadGoat Make Utility V3.9-6 H > -MMK-I-COPYRIGHT, Copyright  1992-2002, MadGoat Software.  All Rights  > Reserved.   1 Still fighting it.  Here is my build environment: 	 $ cc/vers + Compaq C V6.4-005 on OpenVMS Alpha V7.3-2    $ mms/ident L %MMS-I-IDENT, MMS V3.4-3 (c) Compaq Computer Corporation 2000.(c) Electronic  Data Systems Limited 1995, 2000  $   H I'd not realized my C compiler was that out of date.  The crazy thing isK I've had 6.5 sitting on the shelf for a couple years.  I don't know if that  could be part of my problem.   		Zane   ------------------------------   End of INFO-VAX 2006.087 ************************                                                                                                                                                                                                                                                                                                      %d?W*T?T@pur6ĝ[w>@yߓLn0b8&C6C13CeD^6^וaI8E+3B<{顮D{(HfVP7E.o[J8P6D|qg{ŵ;/2,IM\1<jAȶjn&Ȇu?3s{ŏb.oK29lV%۳ZOVko}5$EH&ߨfFªl@{߳&nLtS[ N):^]Zڴp?S㻹ߠn19;zxznev $Px2'`~17.Qo
Ȓhqoע3fRʇ"ѾZJ7)Kg$FsI*i/@^K~MB
_5?@ו(vRv,CHtkert*W7RYŮ'ktїTl5qBlV2rN&ЮƶN;?i`g	{vQ$Qx}~P?WYG?O.xa/~kog)
?qvSlA1&߽fY2-]<2w۶Ӽ"ՌV֩ia64QcrZD["O/ aE
\msRۖ~{Ji]\H#
L$FI0Di]Dm5 'd÷bjʾu5QHe.XFv^X+?fh@
!LtIÜU|tL\dBuQM.VX$
I%ewV;>Q/@߬Qq{kw0#JjԺ1V1%&Qdl/Nt>H$W+/vnwҫ<YZ_ hg%}e7eNK8:eG*
^6[WJҎiא/ ߦT2J̈́茡
Ź4kSPE˶!];
WQ{tGawJ=Dpȡ{!	{j%NXd< )C7`9rq%޶ZC3
6&ڐ'He+fu#:o_6U0NS9i(&ݴ.rCh"P
qARl/uZV/<wWLTt2,jmyg)ć6mZ3sl]Qύx0'%^BAg}9h;A\mu䥭:f& %{m`;zZСՁ/JO1.&Z#*t*mc}nudŏKKXv
oSPiQ,}2TZ6!'w}E:Fe=s&k
;PXc|vibD297Q.+S]DM(L|ZO6.7ߠC~\L.EbCK+(@F8
{KYTS[mմ!mHDMǞtDTK}j+XRlwٳ,VPa۸UD!Rw,ho(xi8^{ԣo]uںJ4.<=cX̔mqD?ЂNVot~Ԟc.U1-c|y(ZC-g0^vI5	 'q3H4RQHM*,m]*w fUOIIdIuހT"ؽӞlb@J605!rH vXNs F
^3b;2I:ڨB4xZ։cd5$xCz`673}}61Ѐΰߑ$~}Gp8ޣcJC"ΠSukaU-X~7iU^IkLP Xqͪ"h.!}ѲG;کUvV_F$īs>9xhr;s8L߆9Q#ym2ԶfVHġlwKw*Cex k4B+>D޲^ҸHKS1ELſ@0"rv81}1}5ijURSOGP)9m\(	E!ڀo,M59X32**-:dO0틁g4W
sQ)KD {R@&QX^g|o51	0rq^!lrXHW`8u[L"=C[dHSoa^nT*v' }\(M	t9MJsmhs1e	tP L! 4r"])ȷnz^HKro<&F858[#qM:͐5Ҥ3_J_xcZ#d=HE9Y&
29R-fգ6wz(.te5><U
=0:tϾ vҙr1~TX_MFs`zQe@j}׫$a80c4{7nJΏףI81nQΨk鲛"3\c!lUP4F
,X(;N3K!'(]RhkI/8H5<H26}W,3aSmAߘyh~/]5{ݺێjpd'0tjHEDl	.qE+\6gZ+P`\I4)Uk&*Ȗ5Bem_c8A,§