1 INFO-VAX	Tue, 04 Jun 2002	Volume 2002 : Issue 307       Contents:* Re: $WAKE() lost during high AST activity?* Re: $WAKE() lost during high AST activity?* Re: $WAKE() lost during high AST activity?  RE: (Fwd: Defect-free Software?) 3100/85 scsi base address F Re: Acrobat (Was: Re: Scott Stallard, forget your VMS to HP UX dream!)1 Re: Assembly generation with OpenVMS's C compiler 1 RE: Assembly generation with OpenVMS's C compiler 0 Re: Can one filter node visibility with DECnet ? Re: Challenge: Find The Fault  Re: Challenge: Find The Fault  Re: Challenge: Find The Fault  Re: Challenge: Find The Fault * Re: Changing the default Telnet port (UCX)* Re: Changing the default Telnet port (UCX)* Re: Changing the default Telnet port (UCX)! Console cable for a Microvax 3100 % RE: Console cable for a Microvax 3100  Copying a file via FID Re: Copying a file via FID  Re: Creating ACEs from a program  Re: Creating ACEs from a program  Re: Creating ACEs from a program Re: DEFINE/TRANS=CONC  Re: DEFINE/TRANS=CONC  Re: DEFINE/TRANS=CONC  Re: DFO and ODS-5  Dir/grand Lexical function... ! Re: Dir/grand Lexical function... ! Re: Dir/grand Lexical function...  Re: dlt tape space Re: dlt tape space Re: dlt tape space Re: Exit code from PERL script Re: Exit code from PERL script Re: Exit code from PERL script% For all you hobbyists: IDE on SCSI !! 2 Re: Freeing memory declared in C but freed in PL/I- Re: has anyone ported this opensource to VMS? - Re: has anyone ported this opensource to VMS?  HP Layoffs have begun - Re: Intel to start to advertise even more :-( - Re: Intel to start to advertise even more :-( - Re: Intel to start to advertise even more :-( D Re: Is there a supported way to determine the version no of DWMOTIF?. Linux gains acceptance at expense of Microsoft2 Re: Linux gains acceptance at expense of Microsoft2 RE: Linux gains acceptance at expense of Microsoft2 Re: Linux gains acceptance at expense of Microsoft2 Re: Linux gains acceptance at expense of Microsoft Re: Newuser of PERL5 Re: Newuser of PERL5 Re: Open Letter to HP  Re: Open Letter to HP * Re: Open-source poses security risks - Da!* Re: Open-source poses security risks - Da!* Re: Open-source poses security risks - Da!* Re: Open-source poses security risks - Da!  Re: OpenVMS FAQ due next week... Re: OpenVMS VAX V6.2 on Galaxy! % Oracle Rdb V7.0.6.4 has been released ) Preview picture of Itanium II motherboard + Re: read access required to write to a file  Re: Shadow sets efficiency Re: Shadow sets efficiency SRM scsi error messageP Re: The Press and the IA-64 Port (was Re: Another analyst says VMS port won't beP Re: The Press and the IA-64 Port (was Re: Another analyst says VMS port won't be Re: VMS to UNIX/LINUX # Re: Volume shadowing and a disaster # Re: Volume shadowing and a disaster # Re: Volume shadowing and a disaster # Re: Volume shadowing and a disaster # Re: Volume shadowing and a disaster 5 Re: Would you like to see this on the VMS freeware CD   Re: [Fwd: Defect-free Software?]  F ----------------------------------------------------------------------   Date: 3 Jun 2002 18:42 CDT' From: carl@gerg.tamu.edu (Carl Perkins) 3 Subject: Re: $WAKE() lost during high AST activity? , Message-ID: <3JUN200218421861@gerg.tamu.edu>  1 JF Mezei <jfmezei.spamnot@videotron.ca> writes...  }Jeffrey Chimene wrote: T }> Agreed. However, that's not the issue in this particular case. The current designX }> employs a $qioW inside an AST. Not only will that queue the I/O request, it will alsoS }> wait for its completion; which wait is counter to the purpose of an AST routine.  } M }Routine is called to advise you that a buffer has been filled by a READ QIO. L }Prior to issuing the next READ QIO, you need to ensure that the contents of! }that buffer have been processed.  }  }so: }check "A"' iostatus block1 }QIOW (write) of the received buffer to channel B , }QIO (read) for the next data from channel A } J }Is a safe way of doing this that doesn't require you do buffer managementN }since the WRITE QIO can work on the buffer that the READ QIO just filled, andK }by the time you issue the next READ QIO, you know that the contents of the D }buffer are no longer needed and can be the targer of the next read. } J }Now, the QIOW is risky, I agree. However, one can do a risk analysis on aK }write IO blocking. For serial ports, I have found that a write QIOW rarely  }stops a program permanently.  } N }If you choose to use QIO for the write as well, you then have to contend withM }a circular list of buffers, and ensure that, for the READ QIO , you use only H }free buffers that are not currently specicied in uncompleted write QIO.  C You don't have to worry about that. If you don't want to read again E until the write is complete, you can just not queue an IO on the read H channel until the write happens. One way you can insure that this is theK case is by issuing the write QIO (not QIOW) in the read completion AST, and D issuing the read QIO (not QIOW) in the write completion AST. No QIOWE anywhere, no blocking anywhere, and you only ever need one buffer per B channel pair. But don't forget to check for success before issuing< the QIOs - "completed" doesn't necessarily mean "succeeded".  E This sort of thing is a particularly important if you also need to do G any sort of activity other than just the read from A write to B routine F ad infinitum. If you need to do something in addition to that, perhapsK something as simple as periodic "housekeeping" like just adding a timestamp F and some statistics to a log file, then you don't really want to spend* unknown amounts of time waiting in an AST.  I Besides which, you don't need a circular list and usage checking for your F buffers. You can use a queue. When a buffer becomes free, the AST thatG went off when it became free (because the write completed, for example) H puts the buffer on the "free buffer" queue. If you need a buffer you tryK to pull one off the queue. If you get one, there was one free. If you don't F there wasn't. This would typically be done by calling a routine to getG a buffer form the queue that also checks to see if it can allocate more # when there wasn't one on the queue.   M }In this context, what do you do when the output channel is XOFFED ? You will J }continue to receive data and queue up write QIO until you are out of freeO }buffers, and then what do you do ? You are in the middle of an AST, and cannot J }issue the next READ QIO because you are out of buffers. You are then in aO }situation where you have to loop to check all outstanding io status blocks for J }your outstanding write IOS until you find one that has completed. The endG }result is the same since you are stuck inside of an AST, waiting for a  }resource to become available.  K You're not stuck in an AST in such a situation. If there is no buffer space J left to issue another read, you just don't issue the next read in the AST.J Set an "out of buffers flag" (which may be more complex than a simple flagK since you may want to know which channel it was, for example) instead. Each J time you go through the main program's loop, you check that flag. If it isI set you do some housekeeping to see if there are any free buffers now and H if there are you can then issue reads on every "readless" channel, or as( many as you now have buffers for anyway.  E This looses you nothing since in the other scheme you block (i.e. you E don't issue a read) until your write is done, so clearly this must be C a situation where not issuing an immediate read all the time is not  a problem.    C You might not want to do this multiple outstanding buffers thing if A you want to limit the number of messages that can be lost if this F intermdiate process disappears and the messaging protocol and programs- on either end are not set up to deal with it.   M }the use of QIOW is far simpler and more robust in my opinion since you don't - }need to write som much more additional code.   E Simpler, yes. More robust? I don't think so. When it is sitting there H blocked by the QIOW at AST level there is absolutely nothing the programH can do to recover - well, unless you have an Exec or Kernel mode AST getH fired off occasionally to check for such a situation (which is certainlyJ not more simple than the alternative). If the reader on your write channelI hangs for an hour, you hang for an hour and you can't even emit a warning F message or call for help. This is particularly bad if you are managingD multiple read and/or write channels with one process - if one hangs, they all hang.  M }disclaimer: this philosophy applies only to channels you know well enough to M }judge the risk of becoming stuck during a write operations. If you know that K }such a channel would often get stuck in a write operation, then the design K }might have to be changed. But if all you are doing is taking data from one O }serial port and showing it out of the second serial port, and let the hardware K }and OS take care of flow control, then using QIOW for the write works well L }because the odds of it getting permanently stuck are very low, and the QIOWG }will automatically throttle your program before you fall too far back.   K If it is something that doesn't happen often, then it will seem to suddenly J stop working for no good reason when that millionth event with the "one inM a million" probability happens. Given the speed of a computer and not knowing L the data rate or equipment involved, that could be next week, or next month,G or next year - or it could be a couple of times a day. It could even be  never, if you are lucky.  D How much of this the original poster needs to be concerened about isI debatable. It depends on the details of what he is trying to do. However, I since the program is not working as expected it is entirely possible that D an occasional long wait is causing timing related problems. RemovingG the wait could eliminate the problem (although I would suggest figuring J out why it happened anyway since there is something wrong with the program; logic, and if it can bite you now it might bite you later).   C My *guess* as to the current hibernation problem's cause is that it D may be related to the timing and order in which the timer is set andE the hibernation started as well as the conditions under which each is I done. With an AST that does a QIOW you can't be certain that the interval J between to commands in the program will be short even if they are adjacentF lines of code in the program. If there is an "if X" related to either,E or both, of them then it may be the set of conditions for setting the I timer being met on the first pass after a long wait (which doesn't result I in hibernating because the timer already went off due to an above average I duration QIOW and so the wake flag was set) but then not being met on the G second pass after the long wait. There is also the point that no matter I how many times the $WAKE is called, the flag it sets is only one bit - it H doesn't keep a count, so it will only ever ignore one $HIBER even if theH timer went off multiple times while wating for the hibernation to start.I If any attempt to match the count by doing multiple hibernations is made, + it will tend to make you hibernate forever.    --- Carl   ------------------------------  % Date: Mon, 03 Jun 2002 20:28:40 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> 3 Subject: Re: $WAKE() lost during high AST activity? + Message-ID: <3CFC09AD.BE570E6@videotron.ca>    Carl Perkins wrote: M > case is by issuing the write QIO (not QIOW) in the read completion AST, and > > issuing the read QIO (not QIOW) in the write completion AST.  L Ok, good idea. You are right in that it accomplishes the same "flow control"L as using the QIOW, while letting your program continue processing other ASTs, and mainline even if the write QIO is stuck.  K For my TCP routines (which run as mainline, not AST), I trigger a recurring M AST which runs every 5 minutes. Each read/write IO that succeeds results in a L variable being incremented. When the recurring AST runs, it checks to see ifJ the IO count has changed since it last ran and if not, it then figures outM t6hat the link is stuck and takes actions (in my case, kills the program with I an exit status that tells the DCL to restart it automatically after a few  minutes and send me email).   F I really do miss the unavailability of IO$M_TIMEOUT from the TCPIP QION interface. It is a real pain to have to add your own settimrs and then check 2W event flags to see whenether you got there due to a time expiring or the IO completing.    ------------------------------   Date: 3 Jun 2002 17:56:10 -0700 / From: chris@applied-synergy.com (Chris Scheers) 3 Subject: Re: $WAKE() lost during high AST activity? < Message-ID: <754a27c1.0206031656.1efa4c6@posting.google.com>  Y JF Mezei <jfmezei@videotron.ca> wrote in message news:<3CFA8749.8F8CEA18@videotron.ca>...  > Chris Scheers wrote:H > > If something in the AST activity consumes the $WAKE, then the $HIBER > > might not wake up. > % > Something I am not sure of anymore.  > L > I know that if I do a SET PROC/RESUME followed by SET PROC/SUSP, the /SUSP= > won't do much because there is already a "RESUME" flag set.  > / > Does the $HIBER and $WAKE work the same way ?  > O > If a $WAKE executes before a $HIBER, would the $HIBER simply skip since there   > is a wakeup flag already set ? > L > I agree with your suggestion of the $SCHDWK which can be set to repeatedly- > issue the $WAKE for you at fixed intervals.   B Yes, there is a wakeup flag, but it is only a flag, not a counter.  > ASTs can do funny things to program flow, so you can get funnyE results, especially when an AST calls an RTL and you have no idea how  the RTL is implemented.   ? The problem is that this sequence works:  Wake-Hiber-Wake-Hiber   + But this one doesn't: Wake-Wake-Hiber-Hiber   F If an AST calls an RTL that does a wake-hiber and you don't know about it, you can easily get:   C       $SETIMR_AST(Wake), some-other-AST(Wake-Hiber), non-AST(Hiber)   D The some-other-AST(Wake-Hiber) eats the pending wakeup flag from the@ $SETIMR_AST, so the non-AST(Hiber) does not have a corresponding $WAKE.  F A $SCHDWK ensures that any $HIBER is eventually woken.  It is possible@ to code to handle extraneous $WAKEs, but it is very difficult to handle missing $WAKEs.   ------------------------------  $ Date: Mon, 3 Jun 2002 16:04:20 -0400* From: WILLIAM WEBB <WWEBB1@email.usps.gov>) Subject: RE: (Fwd: Defect-free Software?) - Message-ID: <0033000066571683000002L032*@MHS>   H =0A"All real programs contain errors until proven otherwise - which is = impossible."  ( Gilb's Sixth Law of Computer Reliability   -----Original Message-----/ From: Info-VAX-Request@Mvb.Saic.Com at INTERNET # Sent: Monday, June 03, 2002 3:11 PM B To: Webb, William W Raleigh, NC; Info-VAX@Mvb.Saic.Com at INTERNET) Subject: RE: [Fwd: Defect-free Software?]     H In article <3CFB8476.5F1948F5@cisco.com>, J Ahlstrom <jahlstro@cisco.co= m> writes:   H >> The test engineers say that they can't imagine how they could design=  a worseH >> load on the system than their current test load within the192 hours,=  andH >> that they believe this load is way beyond the heaviest / worst ever = recordedC >> at one of their customers.  There also are physical upper limits H >> (bottlenecks) on the speed and volume of the demands that can be pum= ped H >> through in live operation, which the testers artificially bypass and=  exceed  >> in the test lab.   H    Their test engineers have poor imagination.  Let a couple crackers a= t 7    it.  Try feeding it some not-so-well defined input.=    ------------------------------  # Date: Mon, 03 Jun 2002 23:25:20 GMT 1 From: Richard Banks <rbanks_@_arel_com_au.nospam> " Subject: 3100/85 scsi base address; Message-ID: <Xns92235FB0FF9D6rbanksatarelcomau@61.9.128.12>    Hi,   H I'm trying to find the base address for the NCR 53C94 SCSI adapter in a 
 uVax 3100/85.   J It doesn't appear to be at 0x200C0000 (since e/p of that address crashes).  G Apparently the address is available in appendix A of the Microvax 3100  K System Maintenance guide (part #EK-M3100-SM), but I don't want to pay $130   just to find a memory address.  K If anyone knows the addresses or could tell me how to find them I would be   very appreciative.  	 - RichardW   ------------------------------  # Date: Mon, 03 Jun 2002 22:48:27 GMTt# From: "John Smith" <a@nonymous.com>AO Subject: Re: Acrobat (Was: Re: Scott Stallard, forget your VMS to HP UX dream!)cJ Message-ID: <%mSK8.199178$t8_.150646@news01.bloor.is.net.cable.rogers.com>   www.elcomsoft.come  G Isn't this the Russian company that had its programmer arrrested in Las / Vegas last year on some trumped up DMCA charge?   C -------------------------------------------------------------------g  & Advanced PDF Password Recovery: README& ======================================   Introduction   Requirements   Known bugs and limitations   Technical supportT   Registration   Copyright and licenser!   Where to get the latest version      Introduction ------------  K This program (Advanced PDF Password Recovery, or simply APDFPR) can be usedi3 to decrypt protected Adobe Acrobat PDF files, whichrG have "owner" password set, preventing the file from editing (changing), L printing, selecting text and graphics (and copying them into the Clipboard),D or adding/changing annotations and form fields (in any combination).  K Decryption is being done instantly. Decrypted file can be opened in any PDFcH viewer (e.g. Adobe Acrobat Reader) without any restrictions -- i.e. withK edit/copy/print/annotate functions enabled. Please note that APDFPR doesn'taI work with documents which have user-level passwords (preventing the filesuA from being opened), if both user and owner passwords are unknown.s   ....etc....     : "JF Mezei" <jfmezei.spamnot@videotron.ca> wrote in message& news:3CF1581A.558597E9@videotron.ca... > "antonio.carlini" wrote:0 > > AFAIK the security is just that the software1 > > refuses to perform the actions you requested,c/ > > i.e. "it won't" not "it can't". Can you not 0 > > simply use (or write) a PDF reader that will, > > print (or extract to "unprotected" PDF)? >aK > Which makes one wonder why Compaq/HP spaecifically set their documents soP thatK > we could not selct/copy text from them. If they are affraid that we wouldy beC > quoting their statements, then it means that they knew that their 
 statements% > were going to generate discussions.e >iI > Perhaps Stallard put the inflamatory statements about migration to Unixa there G > on purpose: his bosses may not have believed that VMS customers wouldh react I > that way, so he wanted to prove to his bosses that this was in fact the  case.d >sK > And he put that into document not widely distributed, at the end in a Q/A K > section. And yet, hoardes of VMS followers found it and disseminated that(2 > information within hours of its being published. > K > From now on, all of HP should be fully aware that they cannot dick aroundd withJ > VMS customers and that we do read between the lines and cannot be fooled by > "don't worry" statements.a   ------------------------------   Date: 3 Jun 2002 17:15 CDT' From: carl@gerg.tamu.edu (Carl Perkins)-: Subject: Re: Assembly generation with OpenVMS's C compiler, Message-ID: <3JUN200217155070@gerg.tamu.edu>   steffen@illumina.no writes...  }Jesper Naur wrote:)< }> Steffen Grnneberg <steffen@illumina.no> wrote in message4 }> news:LGnK8.8318$_15.245035@news4.ulv.nextra.no...9 }>> How would I make the c compiler output assembly code?vC }>> And, how would I compile the generated code afterwords with cc?  }> pD }> You can list the generated code by specifying the /MAC qualifyer, }> as follows: }>  $ }>     $ cc/lis/mac <your_c_program> }> oF }> However, the output from that is unsuitable for assembling using anH }> assembler - but why would you want to do that? CC compiles the C-code$ }> directly to linkable object code. }> t }>     Best regardsv }>     Jesper Naur }  }  }Thanks alot :)t }Why? To learn ofcourse :)  ? Generally speaking, the only thing you are likely to learn fromr9 this is just how much of a masochist you are (or aren't).    --- Carl   ------------------------------  $ Date: Mon, 3 Jun 2002 15:31:44 -0700# From: "Tom Linden" <tom@kednos.com>i: Subject: RE: Assembly generation with OpenVMS's C compiler9 Message-ID: <CIEJLCMNHNNDLLOOGNJIAENNFBAA.tom@kednos.com>o   >-----Original Message----- / >From: Carl Perkins [mailto:carl@gerg.tamu.edu]k$ >Sent: Monday, June 03, 2002 3:15 PM >To: Info-VAX@Mvb.Saic.Com; >Subject: Re: Assembly generation with OpenVMS's C compiler  >I >o >steffen@illumina.no writes... >}Jesper Naur wrote:= >}> Steffen Grnneberg <steffen@illumina.no> wrote in messaget5 >}> news:LGnK8.8318$_15.245035@news4.ulv.nextra.no...a: >}>> How would I make the c compiler output assembly code?D >}>> And, how would I compile the generated code afterwords with cc? >}> E >}> You can list the generated code by specifying the /MAC qualifyer,n >}> as follows:p >}>O% >}>     $ cc/lis/mac <your_c_program>  >}>iG >}> However, the output from that is unsuitable for assembling using ansI >}> assembler - but why would you want to do that? CC compiles the C-codee% >}> directly to linkable object code.e >}>d >}>     Best regards >}>     Jesper Nauro >} >} >}Thanks alot :) >}Why? To learn ofcourse :)o >e@ >Generally speaking, the only thing you are likely to learn from: >this is just how much of a masochist you are (or aren't).  J Actually if your business is writing compilers, you can learn quite a bit,3 but maybe that is what you meant by a masochist :-)r   > 	 >--- Carls >m >---' >Incoming mail is certified Virus Free. ; >Checked by AVG anti-virus system (http://www.grisoft.com).'A >Version: 6.0.363 / Virus Database: 201 - Release Date: 5/21/2002  >e --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.363 / Virus Database: 201 - Release Date: 5/21/2002   ------------------------------   Date: 3 Jun 2002 23:22:07 GMTt2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)9 Subject: Re: Can one filter node visibility with DECnet ?f* Message-ID: <adgtmv$bqo$2@web1.cup.hp.com>  v In article <58ba0101.0205300230.4ab6fe24@posting.google.com>, andrew.rycroft@intrinsitech.com (Andrew Rycroft) writes:  F :I have 2 separate ethernet networks connected via and OpenVMS system.F :The networks run DECnet Phase IV. The neworks are currently connectedG :via a VAX acting as a router. I want to be able to control which nodes E :on network A can see, and communicate with which nodes on network B.x :Is there a way to do this ?  G   Short of a DDCMP-capable firewall, one of the few ways I am aware of aH   this capability involves placing a level one router between two level H   two routers.  This violation of the DECnet area routing configuration G   requirements means that you must connect through the level one router F   to get to the other side of the network, using explicit routing pathH   specifications, the so-called Poor Man's Routing (PMR) specifications.J   You'll also want to look around for details of the undocumented NML$LOG J   and FAL$LOG logical names, as you can use these to audit through-access.<   Nodes on either side will not see nodes in the other area.  F   The best approach, of course, involves bringing the host security upJ   to current revision and recommendations on the various systems involved.  N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  # Date: Mon, 03 Jun 2002 18:51:18 GMTb# From: "John Smith" <a@nonymous.com>i& Subject: Re: Challenge: Find The FaultI Message-ID: <GUOK8.177976$ah_.65531@news01.bloor.is.net.cable.rogers.com>t  - Assuming that propositions 1 & 2 are related:e   #37 - AOL Time Warnerc> or at a stretch (if you consider mostly tv to be 'publishing')
 #85 Viacom   How am I doing so far?    < "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message! news:3CF58DB1.8C3BB67B@fsi.net...o- > Ok, newsgroup naysayers! I set thee a task:  >EJ > Explain, in as few words as possible, how any of the following fictionalJ > statements in any way compromises company identities or confidence or inB > any way threatens or compromises corporate or national security: >o > Number 1:eJ > Computing industry insiders today announced that a major U.S. publishingG > concern has purchased over $200 million (dollars U.S.) worth of Alpha F > computers and licenses for OpenVMS and layered products. An InfoWildD > spokesman called it a major boost for HP and its recently acquiredA > OpenVMS operating system, long thought to be dead. In a privatefI > interview, he said that this proves OpenVMS is not dead, but is in factn > alive and well.  >  > Number 2:nJ > A major midwest computer distributor announced this week that during itsH > most recent fiscal quarter its sales were shored up by a large sale ofJ > Alpha computers to a Fortune 100 customer. It said that additional salesF > dollars were received from the purchases of licenses for OpenVMS andJ > related database server software from a major vendor that it declined to > identify.i >s > Number 3:lF > HP's recent acquisition of Compaq Computer Corporation may have beenH > justified by HP's announcement this week of a large uptick in sales ofH > its enterprise server operating system known as OpenVMS. While largelyA > considered dead and on the decline in the industry at large, HP B > announced sales of OpenVMS licenses during the past three fiscalJ > quarters that all but erase alleged recent attrition in OpenVMS's marketI > share. At the close of trading today, the price of HP's stock reflectednE > the market's knowledge of OpenVMS's generous profit margins as HP'sgI > stock price increased by ... in spite of market uncertainties about theeH > future of the Alpha in light of Compaq's commitments to Intel's Itanic$ > processor prior to the merger. ... >e > - * -  >iF > There you have it - show me *ANY*thing in the above that identifies,J > beyond the shadow of any possible doubt, any individual OpenVMS customerI > company or in anyway compromises their confidence or security or in any $ > way compromises national security. >h$ > I dare you - go ahead! I dare you! >y > -- > David J. Dachterae > dba DJE Systems  > http://www.djesys.com/ >l* > Unofficial Affordable OpenVMS Home Page:! > http://www.djesys.com/vms/soho/    ------------------------------   Date: 3 Jun 2002 16:03:43 -0700 ( From: bob@instantwhip.com (Bob Ceculski)& Subject: Re: Challenge: Find The Fault= Message-ID: <d7791aa1.0206031503.6aee5e5e@posting.google.com>a  ` "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message news:<3CF58DB1.8C3BB67B@fsi.net>.... > Ok, newsgroup naysayers! I set thee a task:  > J > Explain, in as few words as possible, how any of the following fictionalJ > statements in any way compromises company identities or confidence or inB > any way threatens or compromises corporate or national security: >  > Number 1:hJ > Computing industry insiders today announced that a major U.S. publishingG > concern has purchased over $200 million (dollars U.S.) worth of Alpha F > computers and licenses for OpenVMS and layered products. An InfoWildD > spokesman called it a major boost for HP and its recently acquiredA > OpenVMS operating system, long thought to be dead. In a private7I > interview, he said that this proves OpenVMS is not dead, but is in factQ > alive and well.e >  > Number 2:AJ > A major midwest computer distributor announced this week that during itsH > most recent fiscal quarter its sales were shored up by a large sale ofJ > Alpha computers to a Fortune 100 customer. It said that additional salesF > dollars were received from the purchases of licenses for OpenVMS andJ > related database server software from a major vendor that it declined to > identify.s >  > Number 3: F > HP's recent acquisition of Compaq Computer Corporation may have beenH > justified by HP's announcement this week of a large uptick in sales ofH > its enterprise server operating system known as OpenVMS. While largelyA > considered dead and on the decline in the industry at large, HPeB > announced sales of OpenVMS licenses during the past three fiscalJ > quarters that all but erase alleged recent attrition in OpenVMS's marketI > share. At the close of trading today, the price of HP's stock reflected E > the market's knowledge of OpenVMS's generous profit margins as HP'siI > stock price increased by ... in spite of market uncertainties about theeH > future of the Alpha in light of Compaq's commitments to Intel's Itanic$ > processor prior to the merger. ... >  > - * -i > F > There you have it - show me *ANY*thing in the above that identifies,J > beyond the shadow of any possible doubt, any individual OpenVMS customerI > company or in anyway compromises their confidence or security or in anyn$ > way compromises national security. > $ > I dare you - go ahead! I dare you!  9 are you sure this is fictional?  sounds real to me ... :)S   ------------------------------  # Date: Tue, 04 Jun 2002 03:19:36 GMTs1 From: "David J. Dachtera" <djesys.nospam@fsi.net>E& Subject: Re: Challenge: Find The Fault' Message-ID: <3CFC355D.E3E204C5@fsi.net>d   John Smith wrote:i > / > Assuming that propositions 1 & 2 are related:  >  > #37 - AOL Time Warnern@ > or at a stretch (if you consider mostly tv to be 'publishing') > #85 Viacom >  > How am I doing so far?  D Depends. What are you talking about? To what do these numbers refer?C ...and to what in the original text do are they intended to relate?    --   David J. Dachterad dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  # Date: Tue, 04 Jun 2002 03:20:17 GMTa1 From: "David J. Dachtera" <djesys.nospam@fsi.net>o& Subject: Re: Challenge: Find The Fault' Message-ID: <3CFC3586.FAEE4761@fsi.net>T   Bob Ceculski wrote:h > b > "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message news:<3CF58DB1.8C3BB67B@fsi.net>.../ > > Ok, newsgroup naysayers! I set thee a task:h > >yL > > Explain, in as few words as possible, how any of the following fictionalL > > statements in any way compromises company identities or confidence or inD > > any way threatens or compromises corporate or national security: > >W
 > > Number 1:.L > > Computing industry insiders today announced that a major U.S. publishingI > > concern has purchased over $200 million (dollars U.S.) worth of AlphaeH > > computers and licenses for OpenVMS and layered products. An InfoWildF > > spokesman called it a major boost for HP and its recently acquiredC > > OpenVMS operating system, long thought to be dead. In a private K > > interview, he said that this proves OpenVMS is not dead, but is in facte > > alive and well.a > >e
 > > Number 2:4L > > A major midwest computer distributor announced this week that during itsJ > > most recent fiscal quarter its sales were shored up by a large sale ofL > > Alpha computers to a Fortune 100 customer. It said that additional salesH > > dollars were received from the purchases of licenses for OpenVMS andL > > related database server software from a major vendor that it declined to
 > > identify.o > > 
 > > Number 3:eH > > HP's recent acquisition of Compaq Computer Corporation may have beenJ > > justified by HP's announcement this week of a large uptick in sales ofJ > > its enterprise server operating system known as OpenVMS. While largelyC > > considered dead and on the decline in the industry at large, HP-D > > announced sales of OpenVMS licenses during the past three fiscalL > > quarters that all but erase alleged recent attrition in OpenVMS's marketK > > share. At the close of trading today, the price of HP's stock reflected G > > the market's knowledge of OpenVMS's generous profit margins as HP'sxK > > stock price increased by ... in spite of market uncertainties about thesJ > > future of the Alpha in light of Compaq's commitments to Intel's Itanic& > > processor prior to the merger. ... > >K	 > > - * -  > > H > > There you have it - show me *ANY*thing in the above that identifies,L > > beyond the shadow of any possible doubt, any individual OpenVMS customerK > > company or in anyway compromises their confidence or security or in any2& > > way compromises national security. > >s& > > I dare you - go ahead! I dare you! > ; > are you sure this is fictional?  sounds real to me ... :)S  . I sure hope so! I just made it up as I went...   -- > David J. Dachterao dba DJE Systemst http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------   Date: 3 Jun 2002 23:55:53 GMT-2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)3 Subject: Re: Changing the default Telnet port (UCX)c* Message-ID: <adgvm9$bqo$3@web1.cup.hp.com>  N In article <oOGJ8.3287$VP6.288314@stones>, t202 <t202@altavista.co.uk> writes:L :Can anyone tell me how to change the listener port for telnet from 80 to a ) :different one, using standard UCX stack.R  M   Please provide us with some background here.  Specifically, what problem(s)"N   are you trying to solve here?  (Yes, I do understand how you think that you M   might be able to solve the unspecified problem, but I do not yet understanda   the problem itself.)  M   AFAIK, there is no supported way to get the present TCP/IP Services telnet n3   server to listen on more than one port at a time.a  L   For details on some of the background -- OpenVMS version, platform, TCP/IPI   version, etc -- that can be useful when answering a question -- and theeH   details which can be provided in the original posting to get a faster '   answer -- please see the OpenVMS FAQ..  N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  # Date: Tue, 04 Jun 2002 01:39:49 GMT ; From: "John Gemignani, Jr." <John.Non.Gemignani@non.hp.com>r3 Subject: Re: Changing the default Telnet port (UCX)f* Message-ID: <3CFBE080.7F7026EC@non.hp.com>   Hoff Hoffman wrote:t > P > In article <oOGJ8.3287$VP6.288314@stones>, t202 <t202@altavista.co.uk> writes:M > :Can anyone tell me how to change the listener port for telnet from 80 to ai+ > :different one, using standard UCX stack.o > O >   Please provide us with some background here.  Specifically, what problem(s)dO >   are you trying to solve here?  (Yes, I do understand how you think that you O >   might be able to solve the unspecified problem, but I do not yet understando >   the problem itself.) > N >   AFAIK, there is no supported way to get the present TCP/IP Services telnet5 >   server to listen on more than one port at a time.h > N >   For details on some of the background -- OpenVMS version, platform, TCP/IPK >   version, etc -- that can be useful when answering a question -- and the-I >   details which can be provided in the original posting to get a fasterr) >   answer -- please see the OpenVMS FAQ.t > P >  ---------------------------- #include <rtfaq.h> -----------------------------L >       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.comP >  --------------------------- pure personal opinion ---------------------------N >    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com  B I thought I submitted this for release notes a few releases back.  Please forgiveD me for not having the time right now to prove it, but I believe that this should_@ still work.  This was for TCPIP V5.0 or V5.1 from what I recall.  0 Q:  How do I create an ALTERNATE TELNET service?  H A:  There is no clean way to do this, but it is possible.  Use a command     similar to the following:d  $         TCPIP> SET SERVICE TELNET2 -@                 /FILE=SYS$STARTUP:TCPIP$REMOTE_TTY_STARTUP.COM -&                 /FLAGS=(LISTEN,RTTY) -+                 /PORT=10023 /PROTOCOL=TCP -(%                 /PROCESS_NAME="N/A" --!                 /USER_NAME=SYSTEMJ  A     Note that the /FILE qualifier must specify the name of a file:A     which exists at the time of the ENABLE SERVICE command.  ThisEC     file is not read however, so its content is meaningless.  Also,->     the process and user names are not used by this particular?     service because the RTTY flag causes it to be ignored.  Thet@     USER_NAME MUST contain a valid username at the time that the     service is enabled..  C     Note that there is no such thing as an alternate RLOGIN serviceoD     as there is usually no way for a client to select any port other
     than 512.d   ------------------------------  $ Date: Tue, 4 Jun 2002 12:40:19 +1000= From: "Mark\(unMASK\)Forsyth" <forsytMhm@optAushoSme.com.aKu>t3 Subject: Re: Changing the default Telnet port (UCX)n5 Message-ID: <4618637C563142773C5111497AB102DB@plague>p  F "John Gemignani, Jr." <John.Non.Gemignani@non.hp.com> wrote in message$ news:3CFBE080.7F7026EC@non.hp.com...  	 [deletia]a  J > > In article <oOGJ8.3287$VP6.288314@stones>, t202 <t202@altavista.co.uk> writes:}J > > :Can anyone tell me how to change the listener port for telnet from 80 to a- > > :different one, using standard UCX stack.l  	 [deletia]r  B > still work.  This was for TCPIP V5.0 or V5.1 from what I recall.  K Works for UCX V4.2 - ECO 1 (yeah, I know) too. If I try to enable telnet onu two ports I get :-  % mwf on PLAGUE >> ucx ena serv telnet2-1 %UCX-E-STARTERROR, Error starting TELNET2 servicee% -SYSTEM-F-DEVACTIVE, device is activen  G When I try to start the second one. If I disable the original (port 23)a) service I can enable telnet2 no problems.    ie.i   mwf on PLAGUE >> ucx sho serv   L Service             Port  Proto    Process          Address            State  ; BIND                  53  TCP,UDP  UCX$BIND         0.0.0.0- Enabled-; FTP                   21  TCP      UCX$FTPD         0.0.0.0z Enableds; MOUNT                 10  UDP      UCX$NFS_M        0.0.0.0t EnabledC; NFS                 2049  UDP      UCX$NFS          0.0.0.0c Enabled1; POP                  110  TCP      UCX$POP          0.0.0.0t Enabled.; PORTMAPPER           111  TCP,UDP  UCX$PORTM        0.0.0.0e Enabledn; RDBSERVER            611  TCP      RDB70            0.0.0.0e Enabledn; REXEC                512  TCP      UCX$REXECD       0.0.0.0. Enabledo; RLOGIN               513  TCP      not defined      0.0.0.0c Enablede; RSH                  514  TCP      UCX$RSHD         0.0.0.0? Enabledr; SMBD                 139  TCP      SMBD             0.0.0.0  Enabledh; SMTP                  25  TCP      UCX$SMTP         0.0.0.0  Enabled ; TELNET                23  TCP      not defined      0.0.0.0g Enabled * mwf on PLAGUE >> ucx set service telnet2 -@ _mwf on PLAGUE >> /file=sys$manager:ucx$remote_tty_startup.com -( _mwf on PLAGUE >> /FLAGS=(LISTEN,RTTY) -- _mwf on PLAGUE >> /PORT=10023 /PROTOCOL=TCP -M' _mwf on PLAGUE >> /PROCESS_NAME="N/A" - # _mwf on PLAGUE >> /user_name=systemm mwf on PLAGUE >> ucx sho servo  L Service             Port  Proto    Process          Address            State  ; BIND                  53  TCP,UDP  UCX$BIND         0.0.0.0, Enablede; FTP                   21  TCP      UCX$FTPD         0.0.0.0o Enabledo; MOUNT                 10  UDP      UCX$NFS_M        0.0.0.0- Enabled-; NFS                 2049  UDP      UCX$NFS          0.0.0.0  Enabledi; POP                  110  TCP      UCX$POP          0.0.0.0  Enabled-; PORTMAPPER           111  TCP,UDP  UCX$PORTM        0.0.0.0- Enabled ; RDBSERVER            611  TCP      RDB70            0.0.0.0l Enabledc; REXEC                512  TCP      UCX$REXECD       0.0.0.00 Enabled1; RLOGIN               513  TCP      not defined      0.0.0.0a Enabledi; RSH                  514  TCP      UCX$RSHD         0.0.0.0r Enableda; SMBD                 139  TCP      SMBD             0.0.0.0t Enabled3; SMTP                  25  TCP      UCX$SMTP         0.0.0.0  Enabled'; TELNET                23  TCP      not defined      0.0.0.0d Enablede; TELNET2            10023  TCP      N/A              0.0.0.0C Disabled% mwf on PLAGUE >> ucx disa serv telnets% mwf on PLAGUE >> ucx ena serv telnet2f mwf on PLAGUE >> lo 4   FORSYTHM     logged out at  4-JUN-2002 12:34:53.74     Accounting information:nD   Buffered I/O count:                591      Peak working set size: 3680@   Direct I/O count:                  380      Peak virtual size: 175456>   Page faults:                      2355      Mounted volumes: 0rC Connection to plague closed. 00:00:04.31      Elapsed time:       0s 00:01:58.66a) You have mail in /var/spool/mail/forsythma& forsythm@really:~$ telnet plague 10023 Trying 192.168.1.250...n Connected to plague. Escape character is '^]'.       ?  Welcome to OpenVMS (TM) Alpha Operating System, Version V7.1-2b   Username: forsythm	 Password: I    Welcome to OpenVMS (TM) Alpha Operating System, Version V7.1-2 on nodee PLAGUE>     Last interactive login on Tuesday,  4-JUN-2002 12:32:55.08A     Last non-interactive login on Monday,  3-JUN-2002 10:55:28.00I         Oorooi	 Mark F...s   >o2 > Q:  How do I create an ALTERNATE TELNET service? >OJ > A:  There is no clean way to do this, but it is possible.  Use a command >     similar to the following:u >r& >         TCPIP> SET SERVICE TELNET2 -B >                 /FILE=SYS$STARTUP:TCPIP$REMOTE_TTY_STARTUP.COM -( >                 /FLAGS=(LISTEN,RTTY) -- >                 /PORT=10023 /PROTOCOL=TCP -n' >                 /PROCESS_NAME="N/A" -d# >                 /USER_NAME=SYSTEMy > C >     Note that the /FILE qualifier must specify the name of a fileeC >     which exists at the time of the ENABLE SERVICE command.  ThispE >     file is not read however, so its content is meaningless.  Also,'@ >     the process and user names are not used by this particularA >     service because the RTTY flag causes it to be ignored.  ThefB >     USER_NAME MUST contain a valid username at the time that the >     service is enabled.e > E >     Note that there is no such thing as an alternate RLOGIN service F >     as there is usually no way for a client to select any port other >     than 512.y   ------------------------------   Date: 3 Jun 2002 12:17:12 -0700w, From: srp336@getcoactive.com (Steve Pfister)* Subject: Console cable for a Microvax 3100= Message-ID: <45126e60.0206031117.12462f4f@posting.google.com>   E I just bought a Microvax 3100 from ebay, and now I'm trying to figurei out how to get it running...  A What do I need to buy to get a console cable to connect it to the-A 9-pin serial port on my PC? Since I only need this one cable, I'dBC rather buy a pre-made cable than try and make one (besides, I don't78 have the best luck with cables that I've made myself...)  C I was about to order a Dec 423 MMJ->MMJ cable and a modular adapterB? that's MMJ and DB-9 female, but I thought I'd check here first.    Thanks!i   ------------------------------  $ Date: Mon, 3 Jun 2002 15:29:08 -0400* From: WILLIAM WEBB <WWEBB1@email.usps.gov>. Subject: RE: Console cable for a Microvax 3100- Message-ID: <0033000066565566000002L062*@MHS>   8 =0AThis adapter plus a BC16E-xx should do what you want.  H http://www.partner.compaq.com:9003/public/cheat_sheets/cables/padapters= html#H85 71-J  " Have you checked the MicroVAX FAQ?6 http://anacin.nsc.vcu.edu/~jim/mvax/mvax_faq_text.html   WWWebb   -----Original Message-----/ From: Info-VAX-Request@Mvb.Saic.Com at INTERNET # Sent: Monday, June 03, 2002 3:16 PM7B To: Webb, William W Raleigh, NC; Info-VAX@Mvb.Saic.Com at INTERNET* Subject: Console cable for a Microvax 3100    E I just bought a Microvax 3100 from ebay, and now I'm trying to figurei out how to get it running...  A What do I need to buy to get a console cable to connect it to theaA 9-pin serial port on my PC? Since I only need this one cable, I'deC rather buy a pre-made cable than try and make one (besides, I don'ta8 have the best luck with cables that I've made myself...)  C I was about to order a Dec 423 MMJ->MMJ cable and a modular adaptera? that's MMJ and DB-9 female, but I thought I'd check here first.e   Thanks!=   ------------------------------   Date: 3 Jun 2002 13:41:55 -0700y. From: Jack.Trachtman@vmmc.org (Jack Trachtman) Subject: Copying a file via FID = Message-ID: <69d784c4.0206031241.754e6680@posting.google.com>u  C I want to make a copy of a file but I only have the file's FID.  IstB there some nonobvious command in BACKUP, CONVERT, or whatever thatF will let me do this (including giving the output file a new filename)?  Thankse   ------------------------------   Date: 3 JUN 2002 21:17:10 GMTt+ From: Dave Greenwood <greenwoodde@ornl.gov>e# Subject: Re: Copying a file via FIDm1 Message-ID: <3JUN02.21171078@feda01.fed.ornl.gov>e  F In a previous article, Jack.Trachtman@vmmc.org (Jack Trachtman) wrote:E > I want to make a copy of a file but I only have the file's FID.  Is,D > there some nonobvious command in BACKUP, CONVERT, or whatever thatH > will let me do this (including giving the output file a new filename)?	 >  Thanks   E Assuming that this problem can be translated to something like "givendE a file's FID, how do I get its name so I can copy it?", you should be - able to get the filename from a command like:a  7   $ dump/identifier=(3,2,0)/header/blocks=count=0 disk:e   Or you could use DFU:3  !   DFU> SEARCH disk:/FID=nnnn/FULLw   Dave --------------9 Dave Greenwood                Email: Greenwoodde@ORNL.GOVWH Oak Ridge National Lab        %STD-W-DISCLAIMER, I only speak for myself   ------------------------------   Date: 3 Jun 2002 12:19:51 -0700a" From: cstranslations@msn.com (Joe)) Subject: Re: Creating ACEs from a programx< Message-ID: <d56d1c2d.0206031119.3bcb36f@posting.google.com>  [ Arne Vajhj <arne.vajhoej@gtech.com> wrote in message news:<3CFB5AFE.6426000B@gtech.com>...w > Joe wrote:> 2 > http://www.hhs.dk/anonymous/pub/vms/misc/ACL.MAR7 > http://www.hhs.dk/anonymous/pub/vms/misc/TEST_ACL.FOR 7 > http://www.hhs.dk/anonymous/pub/vms/misc/TEST_ACL.PASe > @ > If you read Macro-32 then you should be able to figure it out. > : > If you just need to use it, then maybe you can just call > the routines in ACL.MAR. >  > Arne  8 Seem to have it figured out but will take a look anyway.   Thanks joe    ------------------------------  % Date: Mon, 03 Jun 2002 21:59:55 -0400s' From: Howard S Shubs <howard@shubs.net>s) Subject: Re: Creating ACEs from a program < Message-ID: <howard-04305C.21595503062002@enews.newsguy.com>  < In article <d56d1c2d.0206030931.ffa3e3f@posting.google.com>,$  cstranslations@msn.com (Joe) wrote:  ? > Then I have some really, really, really deep dark memory frompG > about 6+ years back involving the above 16-bit bit field on somethingkE > else I (and that it's backwards - clear bit grants "access" and setn > bit denies access).o   Yeah.t    1 > On top of that the whole bit field seems to run & > backwards so that they're (W,G,O,S).  E Yup.  Backwards and upside-down from the normal way of looking at it.      > I suppose one could make someeE > "edian" type argument on the final point but for christsake... is adH > few extra sentences (I'll forgo a diagram) on what is suppose to be in6 > the OSS$_PROTECTION buffer asking for all that much?  H They have a diagram in the docs, I forget where, but you've already got G it figured out: the fields are backwards, and the sense of the bits is  	 reversed.    -- s# "Run in circles, scream and shout!"r I hope you have good backups! ) Are there any more networked SJFs around?    ------------------------------   Date: 4 Jun 2002 01:43:46 GMTt2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)) Subject: Re: Creating ACEs from a programa* Message-ID: <adh60i$bqo$7@web1.cup.hp.com>  b In article <d56d1c2d.0205310758.2c7db7ce@posting.google.com>, cstranslations@msn.com (Joe) writes:G :Trying to create an ACE entry on a file from a program (and not pass atD :DCL SET SEC command through LIB$SPAWN). Looking at the fine manualsD :and I'm thinking some calls to $PARSE_ACL, $SET_SECURITY, and maybe :$FORMAT_ACL are in order...  8 :Must admit I'm finding the fine manuals a bit obtuse...5 :Anyone out there willing to share a simple example?    ,   OpenVMS version, platform, and language?    E   There's a Fortran example available via the http://askq.compaq.com/aC   search tool, here are three examples of calling security-related  F   OpenVMS system services from C -- use of a fairly recent C compiler B   is assumed by the comment syntax used in some of these examples:     /*> **  Copyright 2002 Compaq Information Technologies Group, L.P. */ #include        <acedef.h> #include        <acldef.h> #include        <descrip.h>t  #include        <lib$routines.h> #include        <ssdef.h>  #include        <starlet.h>. #include        <stdio.h>- #include        <stsdef.h>   #define MAXACE 256   main( int argc, char *argv[] )   {a   int RetStat;?   $DESCRIPTOR( AceBefore, "(IDENTIFIER=[SYSTEM],ACCESS=READ)");D   char AceRawBuf[MAXACE];.   char AceAfterBuf[MAXACE];7  K   struct dsc$descriptor AceRaw = {MAXACE,DSC$K_DTYPE_T,DSC$K_CLASS_S,NULL};oM   struct dsc$descriptor AceAfter = {MAXACE,DSC$K_DTYPE_T,DSC$K_CLASS_S,NULL};t     /*   //  Display the source...e   */   lib$put_output( &AceBefore );c  ,   AceRaw.dsc$a_pointer = (void *) AceRawBuf;0   AceAfter.dsc$a_pointer = (void *) AceAfterBuf;     /*(   // convert the ACE from text to binary   */&   RetStat = sys$parse_acl( &AceBefore,     &AceRaw,     NULL, NULL, NULL );p&   if (!$VMS_STATUS_SUCCESS( RetStat ))     lib$signal( RetStat );     /*@   // the first byte of the raw ACE contains the total ACE length   */%   AceRaw.dsc$w_length = AceRawBuf[0];P     /*2   // convert the ACE from binary to text format...   */$   RetStat = sys$format_acl( &AceRaw,&     &AceAfter.dsc$w_length, &AceAfter,#     NULL, NULL, NULL, NULL, NULL );e&   if (!$VMS_STATUS_SUCCESS( RetStat ))     lib$signal( RetStat );     /*   //  Display the results...   */   lib$put_output( &AceAfter );     return SS$_NORMAL;   }      	---     //> //  Copyright 2002 Compaq Information Technologies Group, L.P. // #include <acedef.h>  #include <descrip.h> #include <ssdef.h> #include <starlet.h> #include <stdio.h> #include <stsdef.h>0 #include <lib$routines.h>h   #define MAXACE 256    int main(int argc, char **argv )   {:   int RetStat;J   $DESCRIPTOR( AceText, "(AUDIT=SECURITY,ACCESS=DELETE+CONTROL+SUCCESS)");&   struct dsc$descriptor AceBufferDesc;   char AceBuffer[MAXACE];t   char *AceAuditName;u  '   AceBufferDesc.dsc$w_length  = MAXACE;a.   AceBufferDesc.dsc$b_dtype   = DSC$K_DTYPE_T;.   AceBufferDesc.dsc$b_class   = DSC$K_CLASS_S;*   AceBufferDesc.dsc$a_pointer = AceBuffer;  ?   RetStat = sys$parse_acl( &AceText, &AceBufferDesc, 0, 0, 0 ); &   if (!$VMS_STATUS_SUCCESS( RetStat ))     lib$signal( RetStat );  ?   AceAuditName = ((struct acedef *)AceBuffer)->ace$t_auditname; %   printf("%16.16s\n", AceAuditName );,     return SS$_NORMAL;   }      	--r   //> //  Copyright 2002 Compaq Information Technologies Group, L.P. // // // //++
 //  Facility:l // //	Examplesb // //  Version: V1.3r //
 //  Abstract:/ /// //      GBLSEC.C -- global section example coder //< //	Creates a backing storage file, and maps a global section= //	into process virtual address space (twice), and then showst //	basic section operations. // //  Author:- //	Stephen Hoffman // //  Creation Date:  2-Jan-1990 // //  Modification History:s //, //	Stephen Hoffman	    remove unused "stubs" //	V1.1 27-Aug-1997-/ //	Stephen Hoffman	    added sys$updsecw calls.- //	V1.2 07-Apr-1998h: //	Stephen Hoffman	    added $set_security, $dgblsc calls. //	V1.3 17-Sep-1999m // //  Build Instructions:a //# //      $ CC/DECC/PREFIX=ALL GBLSECf //      $ LINK GBLSECr //      $ RUN GBLSEC // // //   #include <descrip.h> #include <lib$routines.h>I #include <ossdef.h>h #include <psldef.h>o #include <rms.h> #include <secdef.h>c #include <ssdef.h> #include <starlet.h> #include <stdio.h> #include <string.h>  #include <stsdef.h>i #include <unistd.h>>   #define P0SPACE ((void*)0x0200)  #define GBLSECMAX 10 #define MAXIL 10   struct SecStructDef    {    int GblSec[GBLSECMAX];   };  @ ChurnData( char *Lbl, struct SecStructDef *Base, int WriteFlag )     {-
     int i;     int RetStat;  >     //  Now read or write some data in the specified window...  G     printf( "Global section <%s> base is located at %p\n", Lbl, Base );i     if ( WriteFlag )       {r=       printf( "Writing data into the global section ...\n" );s&       for ( i = 0; i < GBLSECMAX; i++) 	Base->GblSec[i] = i;h       }a     else       {g=       printf( "Reading data from the global section ...\n" );e&       for ( i = 0; i < GBLSECMAX; i++)F          printf( "  %s->GblSec[%d] = %d\n", Lbl, i, Base->GblSec[i] );       }t       return SS$_NORMAL;     }n   main()     {o     int RetStat;8     $DESCRIPTOR( SecDsc, "FACNAM_GLOBAL_SECTION_NAME" );3     $DESCRIPTOR( ClsDsc, "SYSTEM_GLOBAL_SECTION" ); 
     int i;     unsigned short int Iosb[4];:(     void *InAdr1[2] = {P0SPACE,P0SPACE};#     void *RetAdr1[2] = {NULL,NULL};7(     void *InAdr2[2] = {P0SPACE,P0SPACE};#     void *RetAdr2[2] = {NULL,NULL}; +     unsigned long int OwnerUIC = geteuid();.*     unsigned short int ProtMask = 0x0ff00;     struct ItemList3	         {@         short int ItemLength;          short int ItemCode;a         void *ItemBuffer;e         void *ItemRetLen;f         } SetSecIL[MAXIL];%     struct SecStructDef *Sec1, *Sec2;k     unsigned long int GblFlags;1    H     GblFlags = (SEC$M_EXPREG | SEC$M_GBL | SEC$M_SYSGBL | SEC$M_PAGFIL);  3     //  Create, zero, and map the global section...AB     //  Then overlay a structure (Sec1) onto the global section...  7     RetStat = sys$crmpsc( InAdr1, RetAdr1, PSL$C_USER, n7 	GblFlags | SEC$M_DZRO, &SecDsc, 0, 0, 0, 1, 0, 0, 0 ); (     if (!$VMS_STATUS_SUCCESS( RetStat )) 	lib$signal( RetStat );w     Sec1 = RetAdr1[0];    &     //	map the global section.  Again.B     //  Then overlay a structure (Sec2) onto the global section...  7     RetStat = sys$crmpsc( InAdr2, RetAdr2, PSL$C_USER,  * 	GblFlags, &SecDsc, 0, 0, 0, 1, 0, 0, 0 );(     if (!$VMS_STATUS_SUCCESS( RetStat )) 	lib$signal( RetStat );      Sec2 = RetAdr2[0];    C     //  Now write some data into one "window"...  And then read the -     //  data back in from the other "window"..  %     ChurnData( "Section1", Sec1, 1 );1%     ChurnData( "Section1", Sec1, 0 );n%     ChurnData( "Section2", Sec2, 0 );   F     //  Now flush out the contents of the section to disk...  Twice...G     //  Once for each of the two address ranges that we currently have n)     //  mapped into the global section...   8     RetStat = sys$updsecw( RetAdr1, NULL, PSL$C_USER, 0,         0, Iosb, NULL, NULL );(     if (!$VMS_STATUS_SUCCESS( RetStat )) 	lib$signal( RetStat );5(     if (!$VMS_STATUS_SUCCESS( Iosb[0] )) 	lib$signal( Iosb[0] );   8     RetStat = sys$updsecw( RetAdr2, NULL, PSL$C_USER, 0,         0, Iosb, NULL, NULL );(     if (!$VMS_STATUS_SUCCESS( RetStat )) 	lib$signal( RetStat );-(     if (!$VMS_STATUS_SUCCESS( Iosb[0] )) 	lib$signal( Iosb[0] );=    1     //  Reset the protection and the ownership...=     //?     //  The sys$set_security service works with pagefile-backedsB     //  sections, but manipulation of file-backed sections is not B     //  permitted.  (You must unmap everything, reset the security?     //  attributes on the backing file, then remap everything.)P  
     i = 0;/     SetSecIL[i].ItemLength     = sizeof( int ); ,     SetSecIL[i].ItemCode       = OSS$_OWNER;+     SetSecIL[i].ItemBuffer     = &OwnerUIC;.&     SetSecIL[i++].ItemRetLen   = NULL;5     SetSecIL[i].ItemLength     = sizeof( short int );A1     SetSecIL[i].ItemCode       = OSS$_PROTECTION;.+     SetSecIL[i].ItemBuffer     = &ProtMask; &     SetSecIL[i++].ItemRetLen   = NULL;#     SetSecIL[i].ItemLength     = 0;X#     SetSecIL[i].ItemCode       = 0;G&     SetSecIL[i].ItemBuffer     = NULL;&     SetSecIL[i++].ItemRetLen   = NULL;   8     RetStat = sys$set_security( &ClsDsc, &SecDsc, NULL, %       OSS$M_RELCTX, SetSecIL, 0, 0 ); (     if (!$VMS_STATUS_SUCCESS( RetStat )) 	lib$signal( RetStat );X         9     //  And redisplay the contents of the section, as wasn1     //  maintained by the backing storage file...   %     ChurnData( "Section2", Sec2, 0 );.  =     // Now delete the section virtual address space ranges...U  6     RetStat = sys$deltva( RetAdr1, NULL, PSL$C_USER );(     if (!$VMS_STATUS_SUCCESS( RetStat )) 	lib$signal( RetStat );a6     RetStat = sys$deltva( RetAdr2, NULL, PSL$C_USER );(     if (!$VMS_STATUS_SUCCESS( RetStat )) 	lib$signal( RetStat );0    8     //  delete the global section...  (If we need to...)        if ( GblFlags & SEC$M_PERM )       {c,       RetStat = sys$dgblsc( 0, &SecDsc, 0 );*       if (!$VMS_STATUS_SUCCESS( RetStat ))         lib$signal( RetStat );       }r       return SS$_NORMAL;     }i    N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Mon, 03 Jun 2002 22:37:27 +0200 ' From: root <root@localhost.localdomain>o Subject: Re: DEFINE/TRANS=CONC* Message-ID: <adgk27$6nm$1@news1.xs4all.nl>   > I > Sure, one can write a few lines of code to get the physical device name J > for the definition and still keep this hidden from the user.  Not having* > to do so would be more elegant, however.  5 One line suffices: (well broken in 2 for readability)2& assume XXX_ROOT    is DSA1:[APPL_XXX.]         @         $ newroot = f$parse("XXX_ROOT:[PERL5]",,,,"NO_CONCEAL")-:                          - "000000." - "][" - "].;" + ".]"  7         $ define /trans=(conc,term) perl_root 'newroot'   D And it doen't do a a complete emulation of the MFD in the sense that8 perl_root:[0,0] will never work as dsa1:[0,0] will work.2 also the 000000.dir is only visible in a REAL mfd.   just my .02  Regards, Nico Baggus    ------------------------------  % Date: Mon, 03 Jun 2002 16:30:20 -0400s- From: "Richard D. Piccard" <piccard@ohio.edu>h Subject: Re: DEFINE/TRANS=CONC( Message-ID: <3CFBD1D3.F470E215@ohio.edu>  ! What I have done for some time is    $ webdisk = f$trnlnm("user1")-= $ define www_root1 'webdisk'[http_server.]/transl=(conc,term)3   wherec   $ sh log user1/full)C    "USER1" [exec] = "DQA0:" [concealed,terminal] (LNM$SYSTEM_TABLE)2  H The single-line variation to define www_root1 is left as an exercise for the reader.w   				RDP      Phillip Helbig wrote:i >  > > In the Dcl Concepts, > $ > Which manual is this specifically? > K > > I read you can only use the CONCEALED attribute with logical names thataL > > represent physical devices. If memory serves me, you can have the string, > > .][ only once in a disk:directory syntax > F > This sounds like a good description; I just didn't know where it was
 > documented.- > I > I encountered this problem not while doing DCL for relaxation, but in a-I > real-world situation.  Is there some fundamental reason WHY this is notWI > allowed?  Allowing it (up to some maximum nesting level, perhaps) wouldgF > be in keeping with the "concealed device looks like a real device to > applications" philosophy.  > I > Sure, one can write a few lines of code to get the physical device namelJ > for the definition and still keep this hidden from the user.  Not having* > to do so would be more elegant, however.   -- iB ==================================================================B Dick Piccard                           Academic Technology ManagerB piccard@ohio.edu                                 Computer ServicesB http://oak.cats.ohiou.edu/~piccard/                Ohio University   ------------------------------   Date: 3 Jun 2002 19:02 CDT' From: carl@gerg.tamu.edu (Carl Perkins)n Subject: Re: DEFINE/TRANS=CONC, Message-ID: <3JUN200219025190@gerg.tamu.edu>  = Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> writes... I }I encountered this problem not while doing DCL for relaxation, but in a  I }real-world situation.  Is there some fundamental reason WHY this is not aI }allowed?  Allowing it (up to some maximum nesting level, perhaps) would -  E It already is. It's just that the "maximum nesting level" allows onlya one concealed logical...   --- Carl   ------------------------------  # Date: Mon, 03 Jun 2002 20:47:16 GMToL From: winston@SSRL.SLAC.STANFORD.EDU ("Alan Winston - SSRL Admin Cmptg Mgr") Subject: Re: DFO and ODS-58 Message-ID: <00A0EE8D.A313A252@SSRL04.SLAC.STANFORD.EDU>  b In article <995e39b6.0206030722.61924602@posting.google.com>, ewilts@ewilts.org (Ed Wilts) writes:@ >It was my impression that DFO 2.6 would support ODS-5 volumes. E >However, it won't let me add a new ODS-5 volume, nor will it processDG >an existing volume that was converted to ODS-5.  The error message is:r >DFO>show nfs001/free/volume8 >%DFG-W-NOTODSDEV, NFS001 is not an ODS-2 mounted device  B Unless you've been playing games with logical names, isn't NFS001 K an NFS-mounted device?  I would really, really not run a local defragmenterf3 on a remotely-mounted, possibly-not-even-VMS, disk.s   -- Alann  O ===============================================================================c0  Alan Winston --- WINSTON@SSRL.SLAC.STANFORD.EDUM  Disclaimer: I speak only for myself, not SLAC or SSRL   Phone:  650/926-3056 M  Physical mail to: SSRL -- SLAC BIN 69, PO BOX 4349, STANFORD, CA  94309-0210 O ===============================================================================i   ------------------------------   Date: 3 Jun 2002 15:32:42 -0700 - From: contracer11@uol.com.br (Shiva MahaDeva)0& Subject: Dir/grand Lexical function...= Message-ID: <ddf392ea.0206031432.1b8858b8@posting.google.com>s  9 Id like to know which lexical or DCL procedure I need tor7 get total used blocks in a directory, like the command < Dir/grand_total. Thanks in advance...   ------------------------------  % Date: Tue, 04 Jun 2002 00:27:56 +0100h From: nic <junk@127.0.0.1>* Subject: Re: Dir/grand Lexical function...' Message-ID: <3CFBFB7C.C5026E@127.0.0.1>u   Shiva MahaDeva wrote:h > ; > Id like to know which lexical or DCL procedure I need to 8 > get total used blocks in a directory, like the command > Dir/grand_total. > Thanks in advance...  D There isn't one, but there's nothing to stop you writing a bt of DCL code to total up the blocks. 2  5 However it could overflow with today's size of disks.   G OTOH you could just dump the output to a file and read it in using DCL!5   -- 0 Regards, Nic Clews (from home), nic at python dot demon dot co dot uk (play) nclews at csc dot com (work)   ------------------------------   Date: 3 Jun 2002 19:38 CDT' From: carl@gerg.tamu.edu (Carl Perkins) * Subject: Re: Dir/grand Lexical function..., Message-ID: <3JUN200219385139@gerg.tamu.edu>  o In article <ddf392ea.0206031432.1b8858b8@posting.google.com>, contracer11@uol.com.br (Shiva MahaDeva) writes...t: }Id like to know which lexical or DCL procedure I need to8 }get total used blocks in a directory, like the command  }Dir/grand_total.  }Thanks in advance...    Perhaps something like:e   $ filespec = "*.*;*" $ filecount= 0 $ alloc = 0 
 $ used = 0 $ lastfile= " "a $LOOP: $ file= F$Search(filespec)! $ If file .EQS. "" Then Goto DONE > $! hangs if no wildcards - keeps returning the same file so...' $ If file .EQS. lastfile Then Goto DONEt $ thisalloc = 0b $ thisused = 0* $ thisalloc= F$File_Attributes(file,"ALQ")* $ thisused = F$File_Attributes(file,"EOF") $ alloc= alloc + thisalloc $ used= used + thisusedk $ lastfile= file $ filecount= filecount+1 $ Goto LOOPe $DONE:; $ Write Sys$Output "Grand total of ",filecount," files, ",-      used,"/",alloc," blocks" $ Exit  < Note: this has a problem with some files that are open: bothA F$File_Attributes calls give you a warning for any open file thatlA is open but not with shared access (or something like that - someiA open files fail some don't, the ones that fail don't don't returnB> the size info so unless they are size 0 this will not give you> the same result as the dir/grand/size=all gives if you hit any such files).   --- Carl   ------------------------------   Date: 03 Jun 2002 19:13:03 GMT4 From: "Jim Strehlow" <JimStrehlowNoSpam@data911.com> Subject: Re: dlt tape space 0 Message-ID: <adgf3v$ois@dispatch.concentric.net>   There is no such command.t  ; On a new system I often perform a backup loop or other loopp& that will continue copying to the tape. past where the normal backup would have ended.  K I then see how many times the loop executes or how far into the next backupaG files copied to the tape before being prompted for a continuation tape.iB I QUIT the backup loop and $BACKUP/LIST=filename if I did not have+ $BACKUP/LOG enabled in the original backup.h  G It is not accurate; but it provides me some sort of feedback and "feel"  for how much "space is left".2   Jim Strehlow, www.Data911.com] Alameda, CA, USA  8 "Chuck Aaron" <caaron@ceris.purdue.edu> wrote in message* news:3CFB880A.24BF1F9C@ceris.purdue.edu...G > What is the command that will provide you the available space left onC > any given DLT tape?S >L	 > Thanks,u >d > Chuckp   ------------------------------  % Date: Mon, 03 Jun 2002 15:18:52 -0500i+ From: Chuck Aaron <caaron@ceris.purdue.edu>  Subject: Re: dlt tape spacec0 Message-ID: <3CFBCF2C.2095802F@ceris.purdue.edu>   Jim,  # Thank you for your input very much.c   Chuck Aaronf   Jim Strehlow wrote:/ >  > There is no such command.B > = > On a new system I often perform a backup loop or other loopS( > that will continue copying to the tape0 > past where the normal backup would have ended. > M > I then see how many times the loop executes or how far into the next backup I > files copied to the tape before being prompted for a continuation tape.oD > I QUIT the backup loop and $BACKUP/LIST=filename if I did not have- > $BACKUP/LOG enabled in the original backup.S > I > It is not accurate; but it provides me some sort of feedback and "feel"t > for how much "space is left".e >  > Jim Strehlow, www.Data911.com  > Alameda, CA, USA > : > "Chuck Aaron" <caaron@ceris.purdue.edu> wrote in message, > news:3CFB880A.24BF1F9C@ceris.purdue.edu...I > > What is the command that will provide you the available space left ont > > any given DLT tape?o > >s > > Thanks,n > >A	 > > Chuckn   ------------------------------  $ Date: Tue, 4 Jun 2002 13:39:00 +0800) From: "Steven Xie" <r33300@email.mot.com>S Subject: Re: dlt tape spacer+ Message-ID: <adhjq1$6bb$1@newshost.mot.com>]  I I use f$getdvi("mkb100","opcnt") to calculate the backup. It can help youB" estimate how much tape space left.    ? "Jim Strehlow" <JimStrehlowNoSpam@data911.com> wrote in message.* news:adgf3v$ois@dispatch.concentric.net... > There is no such command.f >D= > On a new system I often perform a backup loop or other loopi( > that will continue copying to the tape0 > past where the normal backup would have ended. >nF > I then see how many times the loop executes or how far into the next backupI > files copied to the tape before being prompted for a continuation tape. D > I QUIT the backup loop and $BACKUP/LIST=filename if I did not have- > $BACKUP/LOG enabled in the original backup.c >nI > It is not accurate; but it provides me some sort of feedback and "feel"c > for how much "space is left".r >  > Jim Strehlow, www.Data911.com/ > Alameda, CA, USA >n: > "Chuck Aaron" <caaron@ceris.purdue.edu> wrote in message, > news:3CFB880A.24BF1F9C@ceris.purdue.edu...I > > What is the command that will provide you the available space left on  > > any given DLT tape?d > >s > > Thanks,s > >/	 > > Chuckr >9 >    ------------------------------  $ Date: Mon, 3 Jun 2002 18:31:31 -0400  From: John Santos <JOHN@egh.com>' Subject: Re: Exit code from PERL scriptE5 Message-ID: <1020603181819.1319B-100000@Ives.egh.com>/  % On 3 Jun 2002, Larry Kilgallen wrote:i  h > In article <19e2ed27.0206030020.777e6f59@posting.google.com>, meetkrishnas@hotmail.com (Krish) writes: > I > > batch queue, using sys$sndjbc, for execution. The PERL script that iscH > > being executed has its own predefined exit values. Now when the perlH > > script is executed within the batch queue, after completion it exitsI > > with an exit value. We also get the same message when executed simplyB > > from the .COM file:  > > , > > %SYSTEM-F-NOMSG, Message number 000007D4 > > J > > 000007D4 translates to 2004, which is the return value within the perlG > > script for successful completion. Now how do we capture this return G > > value, given from PERL script. Because the $STATUS in the .com file & > > does not reflect this  exit value. > 7 > If you are saying you want a particular text message:a > A > 	1. Change the Perl script to use a site-specific facility codeL > 	   in it's exit values. > C > 	2. Use the SET MESSAGE command to compile an appropriate message  > 	   file.R > D > 	3. Use the SET MESSAGE command to select the resulting executableB > 	   in any process that is supposed to interpret these messages. > C > The value 000007D4 already belongs to the SYSTEM facility in VMS.d  B If PERL wishes to return a status code to VMS (generically) or DCL@ (specifically), it should honor VMS status conventions.  CallingA sys$exit(2004) where 2004 is some status code of its own internalnB meaning not related to VMS status code values is just plain wrong.  ? There are lots of ways to solve this problem.  One would be for G PERL to have its own exit routine that sets a symbol (e.g. PERL_STATUS)MC to the desired value, and then calls sys$exit with a VMS success or  failure status as appropriate.  D Any DCL script (or program looking at a spawned process's completionC status) would first use the VMS $STATUS to determine the success orc? failure of the operation, and then use PERL_STATUS to determinec@ any specific PERL status.  (For a spawned process, it would need< some method to pass PERL_STATUS back to the parent process.)  B Another method would be to define a set of VMS status values, withF approriate severity levels and a PERL-specific facility code, for all C possible PERL exit statuses, and have the PERL exit routine map the,C PERL status values into this set.  (The set of possible PERL status$B values might be too large to make this possible, for example if itB is an arbitrary 32-bit integer with a convention that 0 is failure( and all other values represent success.)  > The underlying problem here is attempting to misuse VMS status values for something else.   -- 0 John Santosr Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  % Date: Mon, 03 Jun 2002 20:15:58 -0500 C From: "Craig A. Berry" <craig.berry@nospam.SignalTreeSolutions.com>t' Subject: Re: Exit code from PERL script H Message-ID: <craig.berry-D25A7E.20155803062002@news.directvinternet.com>  = In article <19e2ed27.0206030020.777e6f59@posting.google.com>,)(  meetkrishnas@hotmail.com (Krish) wrote:  D > The request needs to be processed using PERL script. So the C fileF > creates a temporary .COM file and write to this file the appropriateH > PERL command with its corresponding arguments (Eg $PERL TEST.PL " SHOWG > DEVICE" ). It then closes this file and submits this .COM file into aa0 > batch queue, using sys$sndjbc, for execution.   F A Perl interpreter can be embedded in a C program (that's essentially F what mod_perl does) but your approach is probably good enough as long ' as you don't need extreme scaleability./   > The PERL script that isi5 > being executed has its own predefined exit values. L  E This is where the trouble begins. Exit codes made up out of thin air SC for programs written on other operating systems are likely to mean iG something very different on VMS than what they were intended to mean.  iH In particular, the severity mechanism of a VMS condition value may lead F your DCL procedure to believe an error has occurred when none has, or @ conversely, that the Perl program has succeeded when it has not.   > Now when the perliF > script is executed within the batch queue, after completion it exitsG > with an exit value. We also get the same message when executed simplyX > from the .COM file:  > * > %SYSTEM-F-NOMSG, Message number 000007D4 > H > 000007D4 translates to 2004, which is the return value within the perlE > script for successful completion. Now how do we capture this returnrE > value, given from PERL script. Because the $STATUS in the .com file $ > does not reflect this  exit value.  
 Sure it does:e   $ perl -e "exit 0x7d4;" ( %SYSTEM-F-NOMSG, Message number 000007D4 $ show symbol $statusa   $STATUS == "%X000007D4"     H > We will appreciate any help in capturing this exit value from the PERL	 > script.t  B Capturing the exit value from Perl or any other image is simply a E matter of examining the $STATUS symbol.  You are probably hitting an cA error handler, though, because the made-up exit value the script  E returns has a fatal severity.  Check any ON ERROR statements in your  F command procedure and that will probably tell you where you are going B instead of the statement immediately after the one that runs Perl.  E You really do need to change the Perl script to use exit values that n= make sense on VMS, or at the very least observe the severity MH conventions that are covered in the OpenVMS Programming Concepts Manual  and elsewhere.   ------------------------------  % Date: Mon, 03 Jun 2002 22:04:07 -0500jC From: "Craig A. Berry" <craig.berry@nospam.SignalTreeSolutions.com>>' Subject: Re: Exit code from PERL script H Message-ID: <craig.berry-8EAA10.22040703062002@news.directvinternet.com>  5 In article <1020603181819.1319B-100000@Ives.egh.com>,u"  John Santos <JOHN@egh.com> wrote:  D > If PERL wishes to return a status code to VMS (generically) or DCL; > (specifically), it should honor VMS status conventions.  X  F Yes it should, and (with caveats noted below) it does.  That's Perl.  D Any given Perl program, however, can do whatever it likes, and UNIX @ programmers are used to making up their own exit codes on a per  application basis. r  B Perl's exit() uses the C RTL's exit(), which is a relatively thin E wrapper around sys$exit.  The ANSI C standard does, however, require  H that EXIT_SUCCESS or 0 be considered a successful exit and EXIT_FAILURE ? (1 by convention but not required to be so by the standard) be DH considered a failure.  That's why the C RTL's exit() transposes exit(0) G into sys$exit(SS$_NORMAL), and of course SS$_NORMAL is 1.  They pretty oC much had to do this in order to comply with the standard.  I think uG there is some macro definition that turns off the 0 --> 1 translation; T) see the C RTL docs on exit() for details.n  G Perl on VMS adopts the C RTL's behavior for an exit code of 0 (indeed, iC it just passes a 0 on to exit()). But for exit(1) Perl works a bit  G harder than the C RTL in trying to compy with the de facto standard of tF 1 as a generic failure exit status.  So, Perl by default replaces the @ generic failure code of 1 with the VMS-specific (and relatively @ generic) failure code of SS$_ABORT (0x2c, 0d44).  Some examples  illustrate:r   $ perl -e "exit 0;"- $ sh sym $status   $STATUS == "%X00000001"d   $ perl -e "exit 1;", %SYSTEM-F-ABORT, abort $ sh sym $status   $STATUS == "%X0000002C"   F The translation of 1 to 44 was desireable because even a well-behaved @ UNIX program that has used only 0 and 1 as exit codes would not A otherwise behave as expected, i.e., would not register a failure  F condition, from an exit(1) call.  However, that's not going to please H everyone all the time, so if you want a 1 to mean SS$_NORMAL instead of > SS$_ABORT, you can disable the default behavior with a pragma:  & $ perl -e "use vmsish 'exit'; exit 1;" $ sh sym $status   $STATUS == "%X00000001"  $   @ Exit codes greater than 1 are not subject to any interpretation.  @ > The underlying problem here is attempting to misuse VMS status > values for something else.   Agreed.-   ------------------------------  % Date: Tue, 04 Jun 2002 00:27:35 +0200u From: Dirk Munk <munk@home.nl>. Subject: For all you hobbyists: IDE on SCSI !!& Message-ID: <3CFBED57.4050403@home.nl>  
 Look at this:-  O Disk MNKALP$DKA200:, device type WDC WD80 0JB-00CRA1, is online, mounted, file-iP      oriented device, shareable, available to cluster, error logging is enabled.  P      Error count                    1    Operations completed                160P      Owner process                 ""    Owner UIC                      [SYSTEM]P      Owner process ID        00000000    Dev Prot            S:RWPL,O:RWPL,G:R,WP      Reference count                1    Default buffer size                 512P      Current preferred CPU Id       0    Fastpath                              1P      Total blocks           156301488    Sectors per track                    96P      Total cylinders            16960    Tracks per cylinder                  96  P      Volume label          "DATADISK"    Relative volume number                0P      Cluster size                  16    Transaction count                     9P      Free blocks            143708320    Maximum files allowed           4597102P      Extend quantity                5    Mount count                           1P      Mount status              System    Cache name      "_MNKALP$DKA0:XQPCACHE"P      Extent cache size             64    Maximum blocks in extent cache 14370832P      File ID cache size            64    Blocks in extent cache                0P      Quota cache size               0    Maximum buffers in FCP cache        906P      Volume owner UIC           [1,1]    Vol Prot    S:RWCD,O:RWCD,G:RWCD,W:RWCD  Q    Volume Status:  ODS-5, subject to mount verification, file high-water marking,="        write-back caching enabled.    O This is a 80 GB Western Digital WD 800JB IDE disk connected to the SCSI bus of .Q my PWS. I used a Acard AEC-7720UW SCSI-IDE bridge to connect it. It works great, .< and in total this is a lot cheaper then a 80 GB SCSI disk !!  ! You can find more information at:t http://www.acard.com  & I'll be happy to answer any questions.     Regards,   Dirk   ------------------------------   Date: 4 Jun 2002 02:09:49 GMTu2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman); Subject: Re: Freeing memory declared in C but freed in PL/IF* Message-ID: <adh7hd$bqo$8@web1.cup.hp.com>  l In article <b22333b7.0205310542.6f7ea347@posting.google.com>, bubbapig@hotmail.com (Jeffrey Cameron) writes:   :tF :Another question on integrating PL/I and C/C++ code. In my C++ module4 :I declare an area of memory with a new statement... :oD :My problem arises when the PL/I code tries to deallocate the memory! :associated with that address....j : F :Is there something special that should be done either when allocatingF :my memory in C or deallocating it in PL/I to make this work properly?    E   You should only deallocate memory using the same call(s) that were 0E   used to allocate the memory.  Most languages do eventually use the rA   same underlying OpenVMS RTL VM calls, but can and often do use .C   different zones and different internal implementations -- mixing  B   VM calls across languages can often lead to corruptions, as you    have found.   F   There are numerous discussions of corruptions that have arisen from E   the (misuse of) memory management; see the Ask The Wizard area, andh&   start with topics (1661) and (3257).  N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Mon, 03 Jun 2002 12:33:57 -0700d0 From: Mark Berryman <Mark.Berryman@Mvb.Saic.Com>6 Subject: Re: has anyone ported this opensource to VMS?, Message-ID: <3CFB6235.1B5BCD6A@Mvb.Saic.Com>   Mike Foley wrote:  > 
 >     Sue, > B >     Wouldn't help that much. I don't think there's alot of solid= >     OpenGL support on an Alpha VMS workstation. :) Besides,o< >     VMS already has a flight simulator on the Freeware CD.> >     It's called FLIGHT. Written by Tom Dahl and John Buehler >     of DECwrite fame.a [remainder deleted]o  D Open3D provides pretty good OpenGL support (all the OpenGL I've everF tried to use, anyway).  Also, the MESA library runs just fine on VMS. A So, yes, there is good OpenGL support for Alpha VMS workstations.a  
 Mark Berrymanc   ------------------------------  % Date: Mon, 03 Jun 2002 15:22:20 -0700 ' From: David Mathog <mathog@caltech.edu>s6 Subject: Re: has anyone ported this opensource to VMS?+ Message-ID: <3CFBEC1C.860081BF@caltech.edu>t   Mark Berryman wrote: > F > Open3D provides pretty good OpenGL support (all the OpenGL I've everG > tried to use, anyway).  Also, the MESA library runs just fine on VMS.iC > So, yes, there is good OpenGL support for Alpha VMS workstations.   F Unless you need stereo 3D support as well.  Are there any VMS graphics> cards now that support stereo?  For that matter, are there any	 on Tru64?"   Regards,   David Mathog mathog@caltech.edu   ------------------------------  % Date: Mon, 03 Jun 2002 18:13:37 -0400s- From: JF Mezei <jfmezei.spamnot@videotron.ca>. Subject: HP Layoffs have begun, Message-ID: <3CFBEA0F.6E1F490F@videotron.ca>  - HP Starts Cutting Jobs, Readying New Outlook p  Mon Jun 3, 5:24 PM ET -    By Peter Henderson   L  SAN FRANCISCO (Reuters) - Hewlett-Packard  Co. has begun widespread layoffsI seen as key to the financial success of its merger  with Compaq  Computer G Corp. and expects to outline the progress of the controversial deal on l	 Tuesday.    K   "Those notices have begun," Jim Milton, senior vice president, EnterprisepN Systems Group and general manager, HP Americas, told Reuters on Monday. He wasN referring to notice given to employees they will be laid off, but declined  to% say how many workers have been told.    '  <stuff about HP's stock price removed>t  N  Fiorina expects to make most of the 10 percent cut in positions within six toK nine months of the launch date -- five to eight months from now  -- and the L first line of attrition will come from early retirement buyouts offered some 9,000 U.S. employees.   M  That buyout offer closes on Friday. Fiorina said earlier this month she willaJ give an update on that progress. Technology researcher Martin  Reynolds of. Gartner Inc. said that would be just in time.   L  "I feel they are almost beginning to push the envelope a little bit in term+ of not announcing these layoffs," he said. -  I  HP's Milton said that four of six to seven levels of management had beeneH named in most of his organization and that the threat of layoffs was not paralyzing workers.   L  "There is a little bit of apprehension, and I think it is important to noteJ that, but it pales in comparison to the positive euphoria that exists," he) said. "The momentum is obviously there." p    HOUSEKEEPING   K  Housekeeping details of integrating and slimming the two companies will bekK the highlight of the analyst conference, said Carl Hoagland, an  analyst ateM State Street Global Advisors. He said he is attracted by HP's value but stille" concerned about merger execution.   I  He also was interested to hear what HP would say about rumors that No. 2eK personal computer maker Dell Computer Corp. , deposed from its status as PCBM industry leader by the merger, was considering a move into printers, where HPo makes most of its profits. k  B <QUESTION: are HP printers more profitable than VMS/UNIX/TANDEM ?>    I  Chief Financial Officer Bob Wayman is also expected to come out with newu> financial forecasts, but Hoagland said they were not the key.   L  "If they come out with numbers, fine, I'll write them down with great care,N but the believability, the credibility just isn't there at this point. I think'  it is really way too early," he said. n  M  HP had forecast in September that the merger would cost the combined companyoL up to 4.9 percent in lost revenue but save it $2.5 billion through cost cutsH by the end of 2004, with $2 billion of the cuts due by the end of 2003.   ( <stuff about earnings per share snipped>  I  Gartner data for sales of powerful server computers in the first quarternN showed combined market share of HP and Compaq steady,  buttressing HP's claimsA that it has not lost customers during the turmoil of the merger. ,  N  "The new HP is doing better than the old HP. I am beginning to wonder how badM a shape their computing division was in," Reynolds said,  although he said HPaN Chief Operating Officer Michael Capellas, who runs the company day to day, had3 some tough cost cutting to do, and to  do quickly. e  N  "It is not really a honeymoon. He's got six months to make changes," he said.   ------------------------------    Date: 04 Jun 2002 01:04:53 +0800, From: Paul Repacholi <prep@prep.synonet.com>6 Subject: Re: Intel to start to advertise even more :-(0 Message-ID: <87wutgjnt6.fsf@k9.prep.synonet.com>  3 "David J. Dachtera" <djesys.nospam@fsi.net> writes:c  ) > Hhmmm... Is AMD nipping at their heels?    ) > Perhaps fallout from Hammer Vs. Itanic?   5 AMD is reported to be showing a quad 800MHz Hammer atp8 the moment in TW. Leggless would be a better description I think.   -- s< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Mon, 03 Jun 2002 17:13:43 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca>.6 Subject: Re: Intel to start to advertise even more :-(, Message-ID: <3CFBDC05.2DE7C5BF@videotron.ca>   Paul Repacholi wrote:C7 > AMD is reported to be showing a quad 800MHz Hammer ate: > the moment in TW. Leggless would be a better description
 > I think.  J When they say a quad 800mhz Hammer, this does mean a 4 CPU card, correct ?  L Isn't it correct to state that multi-CPU cards generally use CPUs with lower# mhz ratings than single CPU cards ?)  G Also, the article did specifically state that the chip was purposefullyrW blocked at that speed to prevent speculation on just how fast it will be once released.a   ------------------------------  % Date: Mon, 03 Jun 2002 18:51:04 -0400s- From: JF Mezei <jfmezei.spamnot@videotron.ca>s6 Subject: Re: Intel to start to advertise even more :-(, Message-ID: <3CFBF2D3.A9304624@videotron.ca>   Paul Repacholi wrote:e7 > AMD is reported to be showing a quad 800MHz Hammer ati: > the moment in TW. Leggless would be a better description
 > I think.  4 http://news.com.com/2100-1001-931225.html?tag=fd_top   States:- ##M The new Hammer chips will be capable of 2GHz and faster clock speeds and will0N be able to run  32-bit programs--the kind used on most desktops today--as wellN as 64-bit applications for servers. The Opteron chip will be aimed at servers,H while another family member, an Athlon-branded chip  formerly code-named8 ClawHammer, will debut first as a chip for desktop PCs.  ##   Also:h ##H Nvidia is developing a new version of its nForce chipset as well as bothI 32-bit and 64-bit software drivers that will work with Opteron and AthlonwG processors when they're released. Motherboard  makers Asustek Computer, F Elitegroup Computer Systems, First International Computer and GigabyteI Technology will all offer motherboards that support the new Hammer chips. H Microsoft has even announced support for the new chips. It will create aN version of its Windows operating system that is designed for use with Hammers. ##  J Looks like AMD is making efforts to outlinethe momentum of the chip. WhileF they may have plenty of motherboard makers, has any PC maker announcedE Hammer-based systems ? Intel has at least one committed to IA64 (HP).o   ------------------------------   Date: 4 Jun 2002 04:54:11 GMT 2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)M Subject: Re: Is there a supported way to determine the version no of DWMOTIF?h* Message-ID: <adhh5j$lel$2@web1.cup.hp.com>  U In article <JDferSE2J3$2@elias.decus.ch>, p_sture@elias.decus.ch (Paul Sture) writes:iB :Oops, "entirely harmless" enough to break a piece of code written :in 1995 :-) :-)  H   Bummer.  Sorry.  But please consider that code that is looking at the J   image header and the image identification strings is already looking at G   an undocumented layer of OpenVMS -- and thus is reasonably subject tooB   breakage.  The format and structure of the image headers is not J   documented anywhere outside of the source listings and definition files J   and (to a certain extent) the IDSM.  In particular, please realize that H   the OpenVMS port to IA-64 will have wholely new underpinnings in this F   area, and in the area of object modules.  Pretty much any code that G   believes it knows about an image header or about the object language     will have to change.  H   You could, of course, patch the image headers.  Specifically, the textK   string sitting in the image identification field of the image header. :-)a   	--R  G   None of this, of course, excuses the original (and minor) infraction  H   here, the use of a field test version string in a released version of H   the image(s).  But I simply don't expect to see new images or a patch E   kit released for V1.2-6 to fix this.  That said, I will pass a noteeE   -- I hesitate to call it a bug report -- along to one of the local sE   folks that oversees this area, and specifically concerning the V1.3,.   release that is presently under development.    N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Mon, 03 Jun 2002 19:16:25 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca>t7 Subject: Linux gains acceptance at expense of Microsoftc, Message-ID: <3CFBF8C2.D0386253@videotron.ca>  4 http://news.com.com/2100-1001-931027.html?tag=fd_top  M Essentially, German government has negotiated a deal with IBM to make it easyeM and cheap to buy IBM hardware loaded with SuSe Linux on it, and will now make C it easy for its own agencies/department to buy non-Microsoft stuff.L  K One of the arguments used for this move was that it was less crippling if arN healthy portion of your machines are non-Microsoft when a virus attack reaches your microsoft infrastructure.  N Is it possible that IBM sees in Linux the potential to overthrow Microsoft andN regain its prime position as computing hardware/software vendor ? Its adoptionH and highly visible investment in Linux may have been more strategic than originally understood.  N If IBM is seen as the Linux leader, then, when Microsoft goes down, IBM standsN to gain the most at the expense of the wintel manufacturers such as HP, Dell. I Sun will be in a strange position as a leader in Unix, but not for Linux.a  J I also read that HP sold some 8086 servers to Dreamworks to run Linux, andM that since Linux didn't have the graphics capabilities fast enough to show at I least 24 frames per second interactively, HP ported some of its own HP-UXlN routines to Linux and achieved the required fps rate for image displays. (ThisI info from a press release about how HP made the sale to Dreamworks for ane% upcoming animated film about a horse)n  I If Linux does become pervasive, it will be most intersting to look at theeI impact on the proprietary systems such as Tandem VMS and MVS (and perhapsoL Solaris). They are likely to steal a few customers from the proprietary UnixL such as HP-UX (those who need the extra reliability/features) while majority3 of proprietary Unix shops would be moving to Linux.t  N This means that Stallard's stated intentions to migrate VMS users to HP-UX areL dead wrong and the reverse should happen if HP wants to keep such customers.J This is especiually true since HP is stuck with IA64 while the rest of theK world seems poised to adopt Hammer which will make Hammer based system mucho7 cheaper than the low production proprietary IA64 chips.e   ------------------------------  % Date: Tue, 04 Jun 2002 00:26:01 +01000 From: nic <junk@127.0.0.1>; Subject: Re: Linux gains acceptance at expense of Microsoftr) Message-ID: <3CFBFB09.5D0425F6@127.0.0.1>i   JF Mezei wrote:E > 6 > http://news.com.com/2100-1001-931027.html?tag=fd_top >   G Interesting. I've read reports in the UK where various are trying linux  and StarOffice as alternatives.e  F HU-X's downfall might be the fact it is big endian because you have to8 do lots of data conversion as well as program migration.  E I'm guessing, because I don't know the endieness of most hardware and 0 supporting OSes but is most of the world little?   -- 1 Regards, Nic Clews (from home), nic at python dot demon dot co dot uk (play) nclews at csc dot com (work)   ------------------------------  $ Date: Mon, 3 Jun 2002 16:23:15 -0700# From: "Tom Linden" <tom@kednos.com>C; Subject: RE: Linux gains acceptance at expense of Microsoft 9 Message-ID: <CIEJLCMNHNNDLLOOGNJICENPFBAA.tom@kednos.com>m  F I believe that only Intel and Digital went little Endian.  Most of theG data in the world (a few years ago) was stored in VSAM data sets, which F is big endian.  If you think about things like storing bit strings you- soon realize why little endian was a mistake.a   >-----Original Message-----e" >From: nic [mailto:junk@127.0.0.1]$ >Sent: Monday, June 03, 2002 4:26 PM >To: Info-VAX@Mvb.Saic.Com< >Subject: Re: Linux gains acceptance at expense of Microsoft >  >  >JF Mezei wrote: >>  7 >> http://news.com.com/2100-1001-931027.html?tag=fd_top5 >> j > H >Interesting. I've read reports in the UK where various are trying linux  >and StarOffice as alternatives. >rG >HU-X's downfall might be the fact it is big endian because you have tod9 >do lots of data conversion as well as program migration.  >aF >I'm guessing, because I don't know the endieness of most hardware and1 >supporting OSes but is most of the world little?c >  >--  >Regards, Nic Clews (from home).- >nic at python dot demon dot co dot uk (play)- >nclews at csc dot com (work): >4 >---' >Incoming mail is certified Virus Free.f; >Checked by AVG anti-virus system (http://www.grisoft.com).eA >Version: 6.0.363 / Virus Database: 201 - Release Date: 5/21/2002e >2 ---7& Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.363 / Virus Database: 201 - Release Date: 5/21/2002   ------------------------------  # Date: Tue, 04 Jun 2002 00:40:23 GMTr* From: "Bill Todd" <billtodd@metrocast.net>; Subject: Re: Linux gains acceptance at expense of Microsoft @ Message-ID: <X%TK8.44992$4i.5162883@bin2.nnrp.aus1.giganews.com>  K "nic" <junk@127.0.0.1> wrote in message news:3CFBFB09.5D0425F6@127.0.0.1...h   ...l  G > I'm guessing, because I don't know the endieness of most hardware andt2 > supporting OSes but is most of the world little?  J Most non-toy (i.e., non-x86, though Hammer is looking less and less like aF toy) hardware is either-endian (MIPS, Alpha, Itanic, POWER, and SPARC,L though some have more noticeable endian-preference than others), but I don'tL know whether PA-RISC is.  OTOH, in some real senses 'most of the world' runs* on x86 hardware and is thus little-endian.  K IIRC most OSs on MIPS, POWER, and SPARC run big-endian (though NT didn't on-L MIPS and POWER and I don't think Ultrix did on MIPS).  Alpha OSs usually runH little-endian (I think including Alpha Linux), though NSK wouldn't have.J Itanic doesn't really have any OS share yet, but will run HP-UX big-endian? (and I think Linux as well) - though will run Windows (and VMS)c little-endian.  L If Hammer (or even, gag, Windows on Itanic) is a real winner, big-endian OSsE may start to get marginalized (mixing endianness is still not quite apL non-issue, especially with environments like Windows that, unlike those thatJ can run either-endian, have a specific endianness strongly ingrained).  IfL big-endian systems hold onto enough of the mid-range and higher market, thenK coexistence will presumably continue much as it is now (since the one thingsJ that *is* certain is that the x86 market isn't going to disappear any time soon).   - bill   ------------------------------  % Date: Tue, 04 Jun 2002 07:29:51 +0200v From: Dirk Munk <munk@home.nl>; Subject: Re: Linux gains acceptance at expense of Microsoft & Message-ID: <3CFC504F.6030803@home.nl>   JF Mezei wrote:r6 > http://news.com.com/2100-1001-931027.html?tag=fd_top > O > Essentially, German government has negotiated a deal with IBM to make it easysO > and cheap to buy IBM hardware loaded with SuSe Linux on it, and will now makeuE > it easy for its own agencies/department to buy non-Microsoft stuff.p  K If I'm not mistaken the entire police force of one of the German states is y switching to Linux.S   > M > One of the arguments used for this move was that it was less crippling if auP > healthy portion of your machines are non-Microsoft when a virus attack reaches  > your microsoft infrastructure. > P > Is it possible that IBM sees in Linux the potential to overthrow Microsoft andP > regain its prime position as computing hardware/software vendor ? Its adoptionJ > and highly visible investment in Linux may have been more strategic than > originally understood. > P > If IBM is seen as the Linux leader, then, when Microsoft goes down, IBM standsP > to gain the most at the expense of the wintel manufacturers such as HP, Dell. K > Sun will be in a strange position as a leader in Unix, but not for Linux.  > L > I also read that HP sold some 8086 servers to Dreamworks to run Linux, andO > that since Linux didn't have the graphics capabilities fast enough to show at,K > least 24 frames per second interactively, HP ported some of its own HP-UX P > routines to Linux and achieved the required fps rate for image displays. (ThisK > info from a press release about how HP made the sale to Dreamworks for ano' > upcoming animated film about a horse)  > K > If Linux does become pervasive, it will be most intersting to look at theuK > impact on the proprietary systems such as Tandem VMS and MVS (and perhapsoN > Solaris). They are likely to steal a few customers from the proprietary UnixN > such as HP-UX (those who need the extra reliability/features) while majority5 > of proprietary Unix shops would be moving to Linux.r  N You're right. And shouldn't we consider Linux as a "Industry Standard" Unix ??O  From that point of view Curly and Carly and the rest of the bunch should love cE Linux !! No more proprietary HP UX, but "industry standard" Linux !!!      > P > This means that Stallard's stated intentions to migrate VMS users to HP-UX areN > dead wrong and the reverse should happen if HP wants to keep such customers.L > This is especiually true since HP is stuck with IA64 while the rest of theM > world seems poised to adopt Hammer which will make Hammer based system muche9 > cheaper than the low production proprietary IA64 chips.i   ------------------------------  # Date: Mon, 03 Jun 2002 20:49:52 GMTcL From: winston@SSRL.SLAC.STANFORD.EDU ("Alan Winston - SSRL Admin Cmptg Mgr") Subject: Re: Newuser of PERL5 8 Message-ID: <00A0EE8E.0012A020@SSRL04.SLAC.STANFORD.EDU>  ^ In article <ufn4ko953jj624@corp.supernews.com>, "J. Scott Greig" <jsgreig@geminaq.com> writes: >Hello All:  >p9 >I have installed PERL5 from the OpenVMS Freeware 5.0 CD,mF >and have noticed that the startup file (Sys$Startup:Perl$Startup.Com)' >defines the PERL_ROOT logical name as:s >rJ >   PERL_ROOT" [exec] = "SYS$SYSROOT:[PERL$VMS.PERL5_005_03.]" [concealed] >s= >It seems to me, that any attempt to use PERL_ROOT:[whatever]H >will fail.t   Right.  / Define PERLSHR as PERL_ROOT:[000000]PERLSHR.EXEn  1 Define PERL as a symbol:  $PERL_ROOT:[000000]PERLi  E It's clunky, but once you set it up you don't have to think about it s	 any more.    -- Alan   O =============================================================================== 0  Alan Winston --- WINSTON@SSRL.SLAC.STANFORD.EDUM  Disclaimer: I speak only for myself, not SLAC or SSRL   Phone:  650/926-3056eM  Physical mail to: SSRL -- SLAC BIN 69, PO BOX 4349, STANFORD, CA  94309-0210GO ===============================================================================c   ------------------------------  % Date: Mon, 03 Jun 2002 19:42:37 -0500tC From: "Craig A. Berry" <craig.berry@nospam.SignalTreeSolutions.com>h Subject: Re: Newuser of PERL5hH Message-ID: <craig.berry-7D6F17.19423603062002@news.directvinternet.com>  / In article <ufn4ko953jj624@corp.supernews.com>,h.  "J. Scott Greig" <jsgreig@geminaq.com> wrote:  : > I have installed PERL5 from the OpenVMS Freeware 5.0 CD,G > and have noticed that the startup file (Sys$Startup:Perl$Startup.Com)h( > defines the PERL_ROOT logical name as: > K >    PERL_ROOT" [exec] = "SYS$SYSROOT:[PERL$VMS.PERL5_005_03.]" [concealed]r  H Perl 5.5.3 is quite long in the tooth now and there is no reason to use E it for a new installation unless you know your scripts have specific aG problems with later versions (and in that case you should probably fix h your scripts where possible).  m  E The current production release is 5.6.1; a PCSI kit is available fromr  I <http://www.openvms.compaq.com/openvms/products/ips/apache/csws_perl_reln-
 otes.html>  ? There are other pre-built and source kits for VMS available at -@ <http://www.sidhe.org/vmsperl>.  The canonical location for the - source-only distribution for any platform is , <http://www.cpan.org/src/>.l  F I think Phillip Helbig is right about the source of your problems and C it has to do with rooted logicals and not Perl per se.  However, I 2E believe more recent versions of Perl work harder to get the physical cD device name when generating the DCL that defines PERL_ROOT.  In the F meantime, all you really need to do is define PERL_ROOT to point to a D path with either a physical device name in it or a concealed device A name that is a mere synonym for a physical device name (i.e., no hG directories in it).  If you want to avoid specifying a physical device tF name, something like the following should work (unwrap long lines and 1 replace dir1 and dir2 with your own directories):f   $ perl_root_sym = I F$PARSE("SYS$SYSROOT:[000000]",,,,"NO_CONCEAL")-".][000000"-"]["-"].;"+".t dir1.dir2.]" $ show symbol perl_root_sym-1   PERL_ROOT_SYM = "BRIANA$DKA0:[SYS0.dir1.dir2.]"28 $ define/translation=concealed perl_root 'perl_root_sym'    H Note that it's up to you to determine whether SYS$SYSROOT is really the F best base location or whether SYS$COMMON, SYS$SPECIFIC, or some other ( location is better for your environment.  4 > One cannot specify the installation directory when/ > installing Perl from the OpenVMS Freeware CD.-  F Untrue.  See HELP PRODUCT INSTALL/DESTINATION if you are using one of B the Compaq-sponsored PCSI kits.  If you are unzipping a pre-built B binary just unzip it where you want it.  If you are building from C source, specify the configuration option -"Dprefix=MYDEV:[MYDIR.]"  H where MYDEV and MYDIR are the device and directory specifications where ' you want the installer to put things.  r  E Regardless of which installation method you use, if you have a valid tH installation somewhere other than where you want it, just use BACKUP to G move the entire directory structure to where it should go and redefine h PERL_ROOT appropriately.    G If you run into further trouble, feel free to write to vmsperl at perl y dot org for help.    ------------------------------   Date: 4 Jun 2002 01:23:39 GMTf2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) Subject: Re: Open Letter to HP* Message-ID: <adh4qr$bqo$6@web1.cup.hp.com>  m In article <9059bf6b.0205310725.73d173d4@posting.google.com>, jodonnell@hrblock.com (Jason O'Donnell) writes:     B :In order for me to support and promote continued investment in HP/ :OpenVMS products, I need to see the following:   C   You will see little or none of this *here*.  This is a newsgroup.ME   Please contact your Ambassador or HP OpenVMS Sales Rep or Reseller.      D :1. A press release from the CEO and other senior corporate officersF :highlighting the strengths of OpenVMS and stating that HP will investG :in marketing and development of OpenVMS as one of its premier productss :for the foreseeable future.     Available.  ; :2. Major network television commercials promoting OpenVMS.o     Not gonna happen.h  @ :3. Discussions with customers highlighting planned or potential :OpenVMS enhancements.  B   This has already been happening, and will continue.  Please see @   the Roadmap and other materials at the OpenVMS website, at the?   URL http://www.openvms.compaq.com/.  Also please see the HPS e>   website, at: http://www.compaq.com/hps/.  And please contact%   your Ambassador, Rep, or Reseller. A  N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  # Date: Tue, 04 Jun 2002 03:26:47 GMT 1 From: "David J. Dachtera" <djesys.nospam@fsi.net>n Subject: Re: Open Letter to HP& Message-ID: <3CFC370D.89D2C72@fsi.net>   Hoff Hoffman wrote:g > [snip] >   And please contact >   your Ambassador, Rep,   ! Been there, done that. No schmae.i   > or Reseller.  D Went out of business last year - can't sell VMS around here anymore.4 Even VACS is auctioning off their DEC stuff on eBay.   -- i David J. Dachterae dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/-   ------------------------------   Date: 3 Jun 2002 14:06:11 -0600-- From: koehler@encompasserve.org (Bob Koehler)-3 Subject: Re: Open-source poses security risks - Da!-3 Message-ID: <ob2mDdhiveLq@eisner.encompasserve.org>n  c In article <TdZ4Mo9EsIns@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes:0 > F > I don't know that any of the various Windows operating systems wouldH > qualify as "most proprietary operating system".  Certainly VMS is justD > as proprietary.  Availability of source listings does not make VMSK > any less proprietary.  Perhaps one might say one of the Windows operating A > systems is the "most widely-used proprietary operating system".w  D    We have to disagree.  To me access to the source listings defines%    "open", that is I can look inside.h  E    More to the point:  MS unwillingness to let their own partners seehA    inside, maintain documented interfaces in an upward compatableiC    manner, or honor contracts requiring access to the source seems  E    to make Windows more proprietary than others even though a purely w=    logical definition of the word would make that impossible.n   ------------------------------   Date: 3 Jun 2002 16:58:18 -0600 - From: Kilgallen@SpamCop.net (Larry Kilgallen)n3 Subject: Re: Open-source poses security risks - Da! 3 Message-ID: <RAM9KRoUcjbN@eisner.encompasserve.org>e  c In article <ob2mDdhiveLq@eisner.encompasserve.org>, koehler@encompasserve.org (Bob Koehler) writes: e > In article <TdZ4Mo9EsIns@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes:i >> sG >> I don't know that any of the various Windows operating systems wouldrI >> qualify as "most proprietary operating system".  Certainly VMS is just E >> as proprietary.  Availability of source listings does not make VMSmL >> any less proprietary.  Perhaps one might say one of the Windows operatingB >> systems is the "most widely-used proprietary operating system". > F >    We have to disagree.  To me access to the source listings defines' >    "open", that is I can look inside.l  F You can make the word "open" mean anything you want, but "proprietary"F means that only the owner is allowed to sell it.  That is the case forH both VMS and Windows.  The discussion was about "proprietary", not aboutG the availability of source.  For a long time (I don't know about today)a MVS came with sources.   ------------------------------  % Date: Tue, 04 Jun 2002 00:04:29 +0100e From: nic <junk@127.0.0.1>3 Subject: Re: Open-source poses security risks - Da! ) Message-ID: <3CFBF5FD.4FCEFB95@127.0.0.1>o   Larry Kilgallen wrote:H > You can make the word "open" mean anything you want, but "proprietary"H > means that only the owner is allowed to sell it.  That is the case forJ > both VMS and Windows.  The discussion was about "proprietary", not aboutI > the availability of source.  For a long time (I don't know about today)n > MVS came with sources.  H VMS used to come with sources, if microfiche qualified. you couldn't useB SEARCH on them, but a fiche reader was OK. I think the practice of: automatically shipping these stopped somewhere around V4.7  H Actually if memory serves correctly it was compilation listings, not the actual sources.O  F Same as the CD source service today. They don't cost too much and they, make digging through a dump more worthwhile. -- c Regards, Nic Clews (from home), nic at python dot demon dot co dot uk (play) nclews at csc dot com (work)   ------------------------------  % Date: Mon, 03 Jun 2002 23:58:39 -0400n( From: David Froble <davef@tsoft-inc.com>3 Subject: Re: Open-source poses security risks - Da!o, Message-ID: <3CFC3AEF.2030403@tsoft-inc.com>   Larry Kilgallen wrote:  e > In article <$7umQ9kVNruA@eisner.encompasserve.org>, koehler@encompasserve.org (Bob Koehler) writes:s > j >>In article <d7791aa1.0205311009.1d26bda8@posting.google.com>, bob@instantwhip.com (Bob Ceculski) writes: >>E >>>A conservative U.S. think-tank suggests in an upcoming report that-B >>>open-source software is inherently less secure than proprietaryI >>>software, and will warn governments against relying on open-source for1 >>>national security.5 >>>jD >>   So they're Microsoft parrots.  This is exactly what Bill's beenE >>   saying, without managing to note that the most proprietary OS inb" >>   use is also the least secure. >> > F > I don't know that any of the various Windows operating systems wouldH > qualify as "most proprietary operating system".  Certainly VMS is justD > as proprietary.  Availability of source listings does not make VMSK > any less proprietary.  Perhaps one might say one of the Windows operating A > systems is the "most widely-used proprietary operating system".  >     P Ok, lets just say that windoz is both proprietary AND source code isn't readily M available.  In contrast, VMS is proprietary, but source listings are readily i3 available.  There is a difference beteween the two.t   Dave   ------------------------------   Date: 3 Jun 2002 22:40:05 GMTt2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)) Subject: Re: OpenVMS FAQ due next week...B* Message-ID: <adgr85$bqo$1@web1.cup.hp.com>  L   To various questions which arose in this thread, in no particular order...  B   I would expect the FAQ to be indexed by various search engines, @   there are search engines on OpenVMS (htdig) and else-platform '   being actively prototyped here at HP.e     Yes, I am using Document.p  ;   Text, Postscript, HTML, and Bookreader will be available..  D   I will be posting the text format, in multiple chunks.  Since this>   is roughly eight big messages shipped out quarterly, well...  E   There will be a zip kit available for download.  This zip kit will  )   contain all available FAQ file formats.l  B   Mozilla -- as has been discussed elsewhere prefers to have both E   sufficient physical memory (256MB or better prefered) and pagefile hC   quota (and WSMAX, of course), and it definitely performs best on  C   Alpha systems (EV56 and later) of relatively recent vintage (last,$   six years; 1996 and later, or so).  F   I am not aware of anything in the FAQ that would crash a webbrowser,G   though I would expect this could be an older Mozilla?, a system that  H   is underconfigured for Mozilla?, or an as-yet unidentified bug in the @   (unspecified) browser.  That said, the updated FAQ also has a +   completely new HTML format and structure.n  G   The FAQ already contains information on host-based volume shadowing, pD   the new FAQ pulls most of this detail together into one section of&   the document -- and also indexes it.    N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------   Date: 4 Jun 2002 02:54:03 GMTn2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)( Subject: Re: OpenVMS VAX V6.2 on Galaxy!* Message-ID: <adha4b$bqo$9@web1.cup.hp.com>   In article <D2CCEA33E5CE16C9.04EC7F8D0BC4F7EE.22B6D1D88A3B48D8@lp.airnews.net>, Steve Sparrow <sdsparrow1n0spam@netscape.net> writes:iC :  What's the chance that a multi-processor Galaxy on a SC,GS,ES,DS E :Alpha could eventually support an instance running OpenVMS VAX V6.2?x     An instance?  Zero.  m  K   OpenVMS Galaxy is not relevent here.  OpenVMS Alpha is not relevent here. H   The VAX emulator -- if it's any good -- isn't relevent here.  Host CPUI   scheduling is not relevent here.  The emulator provides the appearance eG   of VAX hardware, meaning that OpenVMS VAX or any other VAX operating gI   system is expected to bootstrap on any VAX hardware configuration that nJ   the operating system contains support for.  So long as the VAX emulator J   provides the appearance of and the function of a VAX configuration that H   is supported by the particular OpenVMS VAX version, the configuration    is expected to work.  L   Now does OpenVMS VAX know anything about OpenVMS Galaxy?  No.  Can OpenVMSJ   VAX operate as an instance in an OpenVMS Galaxy environnent?  No.  CouldM   an OpenVMS VAX emulation communicate -- through shared memory or otherwise, M   via a phyiscal or emulated network -- with an OpenVMS Alpha instance?  Yes.sL   Does OpenVMS VAX have any idea how to read and operate from the system andJ   device and memory configuration tables that are central to the operationI   of OpenVMS Galaxy?  No, OpenVMS VAX assumes control over all resources.   K   Now what does the host operating system and the host scheduling see here? J   It sees an application that is running, with one or more processes.  TheI   host system schedules these according to its rules.  In this case, the eH   application simply happens to be a VAX emulator -- but the host systemG   does not provide any particular special support for this application.   K   OpenVMS Alpha started out getting its configuration information from the yH   SRM console -- the console is a key piece to the operation of OpenVMS I   Galaxy.  (OpenVMS on IA-64 gets its information from ACPI, a layer thatsH   can be treated as similar to SRM for the purposes of this discussion._H   OpenVMS VAX gets its configuration by probing the physical hardware.  F   The difference?  The console provides OpenVMS Alpha (and effectivelyK   OpenVMS on IA-64) with the system configuration that the operating systemaM   is to use, OpenVMS VAX determines this information through far more direct -H   means, and through in-built system configuration routines.  This meansK   that OpenVMS Alpha can load and run on a subset of the processors presentEM   based on the configuration provided to the operating system by the console,fL   while OpenVMS VAX and the VAX console interface would have to be modified I   to provide a similar capability.  Now could the VAX emulator operate on K   an OpenVMS Alpha instance in an OpenVMS Galaxy configuration?  Certainly.    That's how it works now.    4   But I don't think this is what you are asking for.     	--R  5 : Is the Galaxy idea being considered on the Itanium?t  D   We plan to provide Galaxy capabilites on Itanium.  We plan to haulB   over pretty much everything.  Here are the expected *exceptions*6   to this expectation, paraphased from a presentation:6     o Products not previously ported from VAX to Alpha3     o Products retired or in the retirement process=>     o Products where end of service life has been established ?       and will occur prior to the Itanium-based systems release @     o Products in (or moving to) "Mature Product" support. When B       products move to this category, customers are notified that A       no further development will occur. (Exceptions are FMS and c=       Datatrieve, which will be ported to the future OpenVMS :       Itanium-based systems.)e  D   Also see the Itanium Presentation, Slide 7.  An eye test, I admit.  0     http://www.openvms.compaq.com/tud/index.html  E   AFAIK, this information is also available elsewhere on the website.n  N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Mon, 03 Jun 2002 17:59:13 -0400 2 From: norm lastovica <norman.lastovica@oracle.com>. Subject: Oracle Rdb V7.0.6.4 has been released) Message-ID: <3CFBE6B1.8070806@oracle.com>h  6 Oracle Rdb engineering is proud to annouce that Oracle= Rdb Release 7.0.6.4 is now ready for widespread distribution.w/ The Rdb 7.0.6.4 kits are available on MetaLink.c    =          Software Errors Fixed in Oracle Rdb7 Release 7.0.6.4k=          ----------------------------------------------------e  3 	Software Errors Fixed That Apply to All InterfacesfC 	 - Bugchecks Truncating Table in Mixed-Format Area With Row Caches = 	 - Page Locks Not Always Demoted When Blocking AST Deliveredt+ 	 - LRS Looping When Global Buffers Enabled.; 	 - Unresolved 2PC Transactions Rolled Back by RMU /RECOVERd& 	 - Bugchecks at PSII2SCANSTARTBBCSCAN4 	 - Cursor on Ranked Index Returned too Many Records1 	 - Bugcheck at RDMS$$ALPHA$CONVERT_SORT+00000778L% 	 - Privileged User Bugcheck (ACCVIO)a? 	 - RDMS$CREATE_LAREA_NOLOGGING Partly Ignored for Objects witha 	   Row Caches) 	 - Exception in RDMS$$KOD_ISCAN_GET_NEXTtG 	 - Records Incorrectly Applied to a Key Entry with Sorted Ranked Index < 	 - Query With OR and Repeated AND Predicates Looped Forever? 	 - Query With Same Column in Two Clauses Returns Wrong ResultsrD 	 - GROUP BY Query Followed by CASE With EXISTS Clause Returns Wrong 	   Results"? 	 - Query With Join Predicates on Leading Segments and Equality ! 	   Filters Returns Wrong Results A 	 - Query With Transitive Join Predicates and Non-equality Filtera
 	   BugchecksuB 	 - Query With OR Predicates Including Two Similar IS NULL Clauses 	   Returns Wrong Results$: 	 - UNION Query With Constant Column Returns Wrong ResultsG 	 - Query With CAST Function Using Ranked Index Signals Exception Errori; 	 - Incorrect Handling of Range List Retrieval in OptimizerR1 	 - Bugchecks at DIOCCH$FETCH_SNAP_SEG + 00000594  	 - IVP Tests of DECC and VAXC< 	 - Left OJ Query Using Hash Retrieval Returns Wrong Results0 	 - Getting Null Values Instead of Actual Values          SQL Errors Fixedp7 	 - ALTER DOMAIN...DROP DEFAULT Reports DEFVALUNS Errorr: 	 - Exception in Nested Routines Did Not Close Index Scans6 	 - DDL Statements Generated Unexpected Runtime Errors3 	 - INSERT Cursor on a Derived Table Would Bugcheckt. 	 - SQL Query Bugchecks at SQL$$GET_QUEUE_WALK. 	 - SQL Query Bugchecks at SQL$$GET_QUEUE_WALK. 	 - Privileges Not Honored For SET TRANSACTION? 	 - Multistatement Procedures Used with Connections Resulted ine( 	   %RDB-E-OBSOLETE_METADA Error Message6 	 - Incorrect Handling of FOR Loop Select List Columns; 	 - Unexpected Error on FOR Loop With Dialect ORACLE LEVEL1a= 	 - Unexpected Truncation of Data Assigned in Precompiled SQLa           Oracle RMU Errors Fixed? 	 - RMU /VERIFY /ROOT Incorrectly Reports RMU-E-BADAIJPN and/orn 	   RMU-E-AIJNOTFNDn> 	 - RMU /VERIFY Index Warnings Eliminated on Small CardinalityB 	 - RMU /VERIFY /CONSTRAINT Now Uses Warning for CONSTFAIL MessageB 	 - RMU Incremental Backup and Restore Could Cause Truncated Table 	   Rows to Reappear+ 	 - Deleted Rows Reappear After RMU /REPAIRsD 	 - RMU /COLLECT OPTIMIZER_STATISTICS Fails When Temporary Tables in 	   Database6 	 - RMU /BACKUP /LOADER_SYNCH and Usage of Tape Labels? 	 - RMU /BACKUP to Tape can Hang on a Quit Response to a PromptVA 	 - RMU /BACKUP to Tape can Hang When Terminating on Fatal Errors 4 	 - /LOG Qualifier Default for RMU /SET LOGMINER and 	   RMU /SET ROW_CACHE9 	 - RMU /UNLOAD /AFTER_JOURNAL Exception in AIJEXT$FINISHg3 	 - RMU /UNLOAD /AFTER_JOURNAL Wildcard Table Names.A 	 - RMU /UNLOAD /AFTER_JOURNAL AIJ Backup and Restart Informatione0 	 - Unexpected COSI-F-TRU Error from RMU Extract! 	RMU Show Statistics Errors FixedwB 	 - RMU /SHOW STATISTICS Triggers Invoked From User Defined Events- 	   at Times Other Than the Refresh Intervals B 	 - RMU /SHOW STATISTICS Row Cache Information May Not Display the% 	   Information of the Cache Selected-C 	 - Inconsistency in the Hot Standby Statistics Screen of RMU /SHOWp 	   STATISTICS    =          Enhancements Provided in Oracle Rdb7 Release 7.0.6.4-=          ----------------------------------------------------n3 	 - New SET PAGE LENGTH Command for Interactive SQL-3 	 - RMU /UNLOAD /AFTER_JOURNAL Wildcard Table Namesr' 	 - New Options for SET FLAGS Statementi 	 - Enhancements to RMU Extractt   -- e> norman lastovica / oracle rdb engineering / usa / 610.696.4685   ------------------------------  % Date: Tue, 04 Jun 2002 01:07:17 -0400o- From: JF Mezei <jfmezei.spamnot@videotron.ca>o2 Subject: Preview picture of Itanium II motherboard, Message-ID: <3CFC4AEB.4799418E@videotron.ca>  L Found this exclusive preview picture of an Itanium II motherboard, showing a "glowing" Itanium processor. l  S > http://hardocp.com/image.html?image=MTAyMzE0MTM4MmFRVDh6SHlEY0xfMV8xM19sLmpwZw===-   ------------------------------   Date: 4 Jun 2002 01:10:23 GMTr2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)4 Subject: Re: read access required to write to a file* Message-ID: <adh41v$bqo$5@web1.cup.hp.com>  [ In article <611952a3.0205310358.923421c@posting.google.com>, cbdeja@my-deja.com (-) writes:n  ?   When posting, please include the version information.  Why?  f>   Because the C RTL ships with OpenVMS, and the specifics can >   and do vary over OpenVMS releases.  And there are C ECO kits>   available.  And with specific C compiler versions, you have ?   access to the backport library, which provides newer routinesa   on older releases.  G :I am writing a replacement for the C runtime library function access() A :because as it says in the documentation "access() does not checkn :ACLs".s  F   Why are you doing this?  When posting, please provide the background>   on the problem, in addition to providing a discussion of theC   proposed solution(s).  With the background, we can potententially C   provide you with alternatives that you might not have considered.   C   As for access(), the UNIX file protection semantics do not match  B   the OpenVMS semantics.  The access call is a compromise between C   the two schemes.  Classic UNIX, for instance, does not have ACLs.-  G :However, when I investigated the current behaviour of access() I foundsD :that the values it returns do not always agree with what you are IN# :ACTUALITY allowed to do to a file..  C   Please examine the complete OpenVMS security model, as listed in >C   the OpenVMS security manual -- the model is shown as an extensive D   flow-chart.  There are a number of factors involved, including theE   ACLs and the privileges and the subsystem identifiers and the UICs.n  G :For example, if you have a file which has only "W" protection code andnG :you ask access() if you have write access to it (W_OK) it will say you F :do - but in reality you must have both "R" and "W" protection code to! :write to the file using fopen().I  E   Write-only files are rather more difficult than you might initially,
   suspect.  G :Also if your file has only "R" protection code and you ask access() iffA :you can execute the file it will say no - but in reality you CANnA :execute the file because (as the documantation says) "R" impliesw :execute rights.  ?   In OpenVMS protections, read access is a superset of execute.,  F :I now have to decide whether my replacement access() function followsD :the existing access() which in my opinion seems to be defective, orC :instead attempts to report what you can in reality do to the file.m  <   Please don't even THINK of calling your function "access".  C :To do this I would need to ensure that a file had "RWD" protectionoF :code before my access() would say you had write access to it. Also ifE :a file had "R" or "X" protection code then it would say that you hadw :execute access to it. : F :I'm fairly new to VMS programming, so can anyone explain why access()> :seems defective, or any other strange rules which govern whatC :operations you can in reality perform on a file? Or just any othere :comments .....?  C   The safest approach is to simply attempt the file operation -- a hC   pre-emptive protection check is wasteful for several reasons, notg(   the least of which are the following:   A     o The application-based protection check duplicates the checkuD       that the operating system will perform -- if the local checks D       are implemented correctly, a state which is not always a safe        assumption,u  D     o the protections can change between when the check is made and C       the access is attempted, meaning you still need to have code 7.       that can recover from an access failure.  C   If you are inexorably compelled to implement this, use sys$chkpro3C   or sys$check_access -- do NOT attempt your own protection checks.)        But again, what are you up to?    N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  $ Date: Mon, 3 Jun 2002 19:01:20 -0400  From: John Santos <JOHN@egh.com># Subject: Re: Shadow sets efficiencys5 Message-ID: <1020603185101.1319C-100000@Ives.egh.com>s  , On 3 Jun 2002 mckinneyj@cpva.saic.com wrote:  7 > In article <1020602171644.1319A-100000@Ives.egh.com>,e% >  John Santos <JOHN@egh.com> writes:o0 > > On 1 Jun 2002 mckinneyj@cpva.saic.com wrote: > > 4 > >> In article <3CF79958.8040406@xs4all.nospam.nl>,1 > >>  Bart Zorn <B.Zorn@xs4all.nospam.nl> writes:o > >> > Alan E. Feldman wrote:as > >> >> "Marc Van Dyck" <marc.vandyck@skynet.be> wrote in message news:<3cf5df5e$0$6963$ba620e4c@news.skynet.be>...t > >> >>  > >> >>>Hello, > >> >>>Q > >> >>>    We are replacing the old storageworks disks of one of our clusters bya > >> >>>new universalcM > >> >>>disks. This means we'll have to copy about 1 TB net of data. The data> > >> >>>storage is entirelyiT > >> >>>on shadow sets. What is the most efficient way (elapsed time optimisation) : > >> >>>L > >> >>>1) initialize the new shadow sets and copy the data directly to them > >> >>>
 > >> >>>or > >> >>>Q > >> >>>2) copy the data on normal disks and re-mount them as shadow sets (with a  > >> >>>shadow copy) after ? > >> >>>Q > >> >>>Also, when initializing a new shadow set, is there a way to do it withoutg+ > >> >>>having a shadow copy taking place ?. > >> h > >> No. > >>   > >> >>> > >> >>>Thanks in advance, > >> >>> > >> >>>Marc Van Dyck. > >> >>  > >> >> N > >> >> In my previous post I said BACKUP/PHYSICAL does the same thing as doesJ > >> >> a shadow copy operation. I stand corrected. It would be faster, inJ > >> >> general. However, I still haven't tried it to see if the shadowingM > >> >> software would consider them both to be current members. I'll have to,# > >> >> try it when I get a chance.i > >> >>  > >> >> Disclaimer: JMHO > >> >> Alan E. Feldmand* > >> >> afeldman atski gfigroup dotski com > >> > tP > >> > I can't imagine that the shadowing software would accept the result of a P > >> > BACKUP/PHYSICAL as a valid shadowset member just because it has the same  > >> tP > >> It won't... the generation number in the shadow control block matches, but,P > >> the mount flag is also set. Since the SCB's mount flag is set the shadowingQ > >> software will decide that a merge is required. I don't know if the shadowing<L > >> software will make the correct decision as to which device contains theP > >> authoritative data. This would be a problem if the desired source has been,L > >> or, is active other than the backup and merge operations. Any potentialM > >> problem here could be avoided by simply initializing the volume that waseR > >> created using a physical backup. An unqualified INITIALIZE will just recreateQ > >> the volume label and reserved files. The remainder of the disk is unchanged. M > >> If this is done then a shadow copy will be performed rather than a merge Q > >> since the SCB is now absent. The copy will require very few write operations Q > >> since, as others have pointed out, both the copy and merge functions performcL > >> a compare prior to any write and only write when a difference is found. > > H > > I don't understand the benefit of this.  Why is it faster to read A,L > > compare with B (with the write A to B if different rarely if ever done) " > > than it is to read A, write B? > > I > > In other words don't two reads from different disks take just as longa. > > as a read from one and a write to another? > >  > A > In many cases the read operations could be overlapped and occuroG > simultaneously. The compare would be quicker than a write. Obviously,nC > if a write is required, then an extra read has already been done.O  G Okay, I think I see now...  You queue up both reads to separate bufferssE at the same time.  When both have completed, you compare the buffers.AC If the CPU is much faster than the disk, the total time is slightlydF larger than 1 read-time.  (1 read-time+1/2 rotational latency+(compareA time<<<read-time), if the disks are completely unsynchronized.)  l  A Blind read followed by write is 1 read-time + 1 write-time + (1/2d4 rotation time)*2 which is essentially twice as long.  C I think I was thinking of the good old days when DMA disk transfers?= were about the same speed as CPU memory-to-memory compares or @ copies.  I don't have numbers, but I bet with aligned big-bufferB compares on an Alpha, (no writing, so there is no cache flushing),@ wide datapaths, the CPU can compare the two buffers much quicker1 then the disks and SCSI interfaces can load them.s  % This makes much more sense to me now.t   Thanks for the explanation.i   -- p John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  # Date: Tue, 04 Jun 2002 00:17:43 GMTc* From: "Bill Todd" <billtodd@metrocast.net># Subject: Re: Shadow sets efficiencyg@ Message-ID: <HGTK8.44846$4i.5144317@bin2.nnrp.aus1.giganews.com>  - "John Santos" <JOHN@egh.com> wrote in messagen/ news:1020603185101.1319C-100000@Ives.egh.com...n   ...h  I > Okay, I think I see now...  You queue up both reads to separate buffersdG > at the same time.  When both have completed, you compare the buffers.oE > If the CPU is much faster than the disk, the total time is slightlytH > larger than 1 read-time.  (1 read-time+1/2 rotational latency+(compareA > time<<<read-time), if the disks are completely unsynchronized.)a  J Unless the compare time is truly negligible compared with the access timesK (which in the case of sequential access to the entire volume may not be the L case), doing the comparison in one pair of buffers while reading in the next pair is useful.l   >oC > Blind read followed by write is 1 read-time + 1 write-time + (1/2a6 > rotation time)*2 which is essentially twice as long.  I Except that any half-competent implementation will be doing the next read K (into a second buffer) while the write is executing, so wall-clock time foriI the entire scan is essentially the same as the read/read/compare approachtD (and faster whenever the comparison fails and a write must be done).  L I suspect the other explanation offered may be the correct one (that as longK as *most* data is identical the read/read/compare verifies the integrity of I both disks while comparing them, whereas the read/write approach does not J unless it's implemented as read/write/reread-what-you-wrote, which is muchE slower than read/read/compare unless the comparison *usually* fails).a   - bill   ------------------------------   Date: 3 Jun 2002 15:44:46 -0700a@ From: lgemedia@200-171-183-182.dsl.telesp.net.br (Luiz Emediato) Subject: SRM scsi error messagee= Message-ID: <21125970.0206031444.13736907@posting.google.com>    Hi,/  < I have a DEC Alpha XP 150 (Jensen) running firmware SRM 1.7.F I was running it perfectly well under Linux Debian 2.2r2 during a yearF or so when, after a reboot, it showed the following SCSI error message at the bootup:      video okS    mem   okp    nic   okc    ...    scsi  ?? 06 0050-  E There is no boot anymore even though trying to issue the SRM command:n	 >>> boot  ( or other aboot> command level variation.= I could not find, in the manual, a meaning for this error and09 could not fix the problem even after exchange active scsio> terminators or scsi devices. I tried to contact OpenVMS Wizard. at openvms.compaq ASK web page, unsucessfully.< Does anyone please help with this problem with directions or related documentation ?w Thanks in advance,  
 Luiz Emediato.   ------------------------------   Date: 4 Jun 2002 00:52:31 GMTF2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)Y Subject: Re: The Press and the IA-64 Port (was Re: Another analyst says VMS port won't bel* Message-ID: <adh30f$bqo$4@web1.cup.hp.com>   Alan Greig skrev:   H : I trust someone from Compaq will contact both Computer Weekly and Meta : Group to put them right...  @   Analysts have their opinions, you have yours, and I have mine.  B   Right now, I need to get my part of the OpenVMS IA-64 bootstrap @   working and I need to get the source code control system into A   shape for certain of the new OpenVMS IA-64 source code modules.   @   I noticed you missed the London Technical Update -- you missedE   out on learning the details of what OpenVMS Engineering is working oD   on, and directly from the engineers working on the code.  I could B   cite various of the available presentations from that and other A   events, but the best proof will simply involve getting OpenVMS F"   on IA-64 available and shipping.    N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Mon, 03 Jun 2002 22:00:19 -0400n- From: JF Mezei <jfmezei.spamnot@videotron.ca> Y Subject: Re: The Press and the IA-64 Port (was Re: Another analyst says VMS port won't beN, Message-ID: <3CFC1F22.69739EF9@videotron.ca>   Hoff Hoffman wrote: B >   Analysts have their opinions, you have yours, and I have mine.  K So Mr Hoffman, how tantalizing are the early retirement offers from HP thatEN employees have until this Friday to accept or reject ? Tempted to finally make it to Florida ?D  " Glad to see you're not gone yet...  M Is there a way to "ping" an employee to find out if he is/she is still at theuM employ of HP ? I take it that thsoe who are let go after the initial round of @ voluntary severance won't have much of a chance to say anything.   ------------------------------   Date: 3 Jun 2002 12:22:07 -0700E4 From: bbaxter@denvernewspaperagency.com (Bob Baxter) Subject: Re: VMS to UNIX/LINUX= Message-ID: <477fb6c5.0206031122.34beb45f@posting.google.com>t  K > I'm writing(in C++, on VMS) a application that needs to talk to a remote FI > UNIX box and be able to browse the file system on the remote box , and  ) > place files in some specific directory?nH > What's the best way of doing that?Sockets?Or invoking FTP from within  > the application?Any ideas?    C Have you considered SAMBA?  It runs on almost anything and it free.   
 Bob Baxter   ------------------------------   Date: 3 Jun 2002 16:22:14 -0700T1 From: keithparris_NOSPAM@yahoo.com (Keith Parris) , Subject: Re: Volume shadowing and a disaster= Message-ID: <cf15391e.0206031522.3806f563@posting.google.com>t  a JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message news:<3CE8D08D.8600F54A@videotron.ca>...QN > Should one be concerned about the node in building 1 having the time to sendK > shadowing commands to echo erroneous data and data corruption to the diskL< > drives in building 2 ?, or is that a very unlikely event ?  F The way Shadowing keeps the disks in both sites in synch is by sendingD MSCP Write commands (or SCSI-3 Write commands, if you're using Fibre? Channel) to each shadowset member disk concurrently.  So to gethA corruption on disk, you'd have to have a node at the failing sitenD successfully complete a remote write operation to a shadowset memberD at the surviving site, after somehow corrupting data, and before the9 node and/or inter-site link and/or site fails completely.e  E A data transfer will consist of a Command (Write), followed by a data-A transfer operation, followed by a status return.  If the bits get B corrupted in-flight, it's likely the checksums on the packets willA cause the packets to be rejected.  So for corruption to occur, itnC would almost certainly have to be an in-memory corruption somewhere-C along the line, corrupting data which is just about to be sent to a D disk.  The corruption would have to be such that it slipped past theE single-bit correction and double-bit detection in the ECC in memory.  C And this corruption would have to be just right to avoid causing anoF RMS bucket checksum mismatch if it got all the way to the remote disk.  C The odds of a node's memory being damaged severely enough to damage-F the contents of a block you are going to write, without damaging otherF areas of memory badly enough to cause VMS itself to crash, and be ableC to slip through RMS' protection on the bucket afterward, are prettyeF low, although conceivably possible in theory.  This would constitute aB case of undetected data corruption, which VMS and all the hardware. designs aim at making as unlikely as possible.  F But to protect against such a possibility, you'd probably be using RMSB Journalling (or similar techniques with a database product).  ThatC should allow you to roll back an erroneous transaction afterward ifr
 necessary.  B There are a lot of things that are much more likely, and much more6 worth spending your time worrying about, as you plan a disaster-tolerant cluster.: ----------------------------------------------------------: Keith Parris | parris <at> DECUServe <dot> decus <dot> org   ------------------------------   Date: 3 Jun 2002 17:03:41 -0700h1 From: keithparris_NOSPAM@yahoo.com (Keith Parris) , Subject: Re: Volume shadowing and a disaster< Message-ID: <cf15391e.0206031603.d432c9b@posting.google.com>  a JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message news:<3CE9502B.D42357E6@videotron.ca>...eP > In the case of volume shadowing, is this not done almost at the physical levelF > (replace this block with this contents) or is ODS-2 still involved ?  > VMS (whether HBVS is involved or not) deals in 512-byte units.  D The writer of the blocks may be XQP or RMS code, and the contents ofB the blocks will conform to ODS-2 (for file system metadata) or RMSD (for file-internal structure and records) formats, of course.  (NoteF that RMS tends to read and write in multi-block buckets; the XQP often deals with single blocks.)  J > If I change 5 bytes in the middle of a block and write it back, will theP > remote node get a 512 byte buffer and be just told to write it, or will it getO > the changed 5 bytes and told to read the block, replace the 5 bytes and writef > it back ?   ? The data transfer will contain the full 512 bytes, not just the,F changed bytes.  The disk hardware guarantees that data will be written@ in full sector units, so if power fails during a write, the data@ doesn't get cut off in mid-sector.  An RMS bucket thus might getE partially updated, but then the bucket checksum will flag that error.d  I > If the volume shadowing is done at the block/physical level, then is iteI > correct to state that the failing node can theoretically send shadowings1 > commands that would corrupt the remote drives ?r  = In theory, yes, but protection mechanisms like ECC codes, and ? checksums in things like file headers and RMS buckets, make theeD probability of undetected data corruption extremely low in practice.  . > Perhaps I should have rephrased my question:P > What mechanisms exist in the software volume shadowing to ensure the sanity of > all participating nodes ?  > N > If the failing node is the one holding quorum, does the remote node have theP > ability to either freeze itself if it thinks the quorum node is going nuts, orP > must it wait until it has lost total contact before declaring the failing node > to be gone ?  B If nodes lose contact with each other, VMS enters the ReconnectionC Interval.  Basically, during this time, I/Os may be completed under F the protection of the current set of locks held, but if an unreachable? node is the lock master for a given lock tree, or if it holds a F conflicting lock, an I/O may well get stalled because it needs a givenA lock.  If the Reconnection Interval passes without communicationsc@ being restored, a Cluster State Transition takes place to removeD unreachable nodes and to free the locks they held.  At this point, aE subset of nodes may lose quorum and have to suspend all I/O activity, 0 while the majority of nodes continue processing.  F For more detail on what happens during cluster node failures and stateC transitions, see Roy Davis' VAXcluster Principles book or the State B Transitions session notes at http://www.geocities.com/keithparris/: ----------------------------------------------------------: Keith Parris | parris <at> DECUServe <dot> decus <dot> org   ------------------------------   Date: 3 Jun 2002 17:20:09 -0700-1 From: keithparris_NOSPAM@yahoo.com (Keith Parris)r, Subject: Re: Volume shadowing and a disaster= Message-ID: <cf15391e.0206031620.4783d7d8@posting.google.com>@  a JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message news:<3CE9DCA6.9FCF1DE4@videotron.ca>...0 > Rob Young wrote:B > > http://www.openvms.compaq.com/openvms/fibre/fc_hbvs_dtc_wp.pdf ...aN > One of the recommendation in that document is that the backup node not mountO > the shadow set so that the main node has full control over the shadow set and L > a single FC path to it without worrying about backup node MSCP serving it.I > However in this scenario, wouldn't there be even less "sanity" checks ?   C One of the major challenges covered by that document is how to deallF with the aftermath of a failure of the FC inter-site link when the SCSB link survives, and a lot of that difficulty was due to the lack ofC ability to fail over between direct FC and MSCP-served paths.  (All E that difficulty goes away in 7.3-1 with the addition of that failovert capability.)  K > In a FC scenario, is there some specific protocol used to send disk driveaP > commands ? Does the master node send MSCP commands and there is a MSCP-capableP > controller in the disk array ? Or does the communication inside the FC contain > direct SCSI commands ?  B If there is no inter-site FC link, then nodes at a given site sendF MSCP I/O commands through the VMS MSCP server nodes at the other site.A  If there is a FC inter-site link, nodes at each site can send FCdE command directly to the disk controllers at either site.  FC commandsrC are SCSI-3 commands, very similar to the SCSI-2 (Read, Write, etc.)e commands we're used to.e  P > Either way, will the backup node be advised if the master nodes starts to sendO > funky commands over the FC ? Or will its only notification of the master node J > going nuts happen when the master node is take out of the cluster due to! > failure at the SCS link level ?   = There's no way for the backup node to know if undetected dataHF corruption were somehow occurring between a given node and the disks. E When a node fails, the cluster state transition will deal with that. tE And if the corruption causes RMS bucket checksums to be in error, the ' surviving node will soon discover that.d: ----------------------------------------------------------: Keith Parris | parris <at> DECUServe <dot> decus <dot> org   ------------------------------  # Date: Tue, 04 Jun 2002 01:03:31 GMT * From: "Bill Todd" <billtodd@metrocast.net>, Subject: Re: Volume shadowing and a disasterB Message-ID: <DlUK8.111726$jm.11040056@bin6.nnrp.aus1.giganews.com>  > "Keith Parris" <keithparris_NOSPAM@yahoo.com> wrote in message6 news:cf15391e.0206031603.d432c9b@posting.google.com...: > JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message( news:<3CE9502B.D42357E6@videotron.ca>...   ...w  A > The data transfer will contain the full 512 bytes, not just theoH > changed bytes.  The disk hardware guarantees that data will be writtenB > in full sector units, so if power fails during a write, the data$ > doesn't get cut off in mid-sector.  D There is a condition that has been observed (in PCs) with sufficientJ frequency to be alarming:  on power loss first memory or the bus clamps toK nulls and only some time later does the disk stop accepting (now null) data J from it to complete a sector write.  As a result, the data written towardsL the end of the disk sector has been corrupted, but as far as the disk ECC is! concerned it is completely valid.l  J ODS-x's checksums in critical sectors would catch such corruption (if diskI sector transfers were truly atomic and disk ECCs bullet-proof, there'd berD little need for those checksums, though they'd still catch otherwiseL undetected bus or interface errors).  For many years I've though it would beH useful for disks to support a host-managed out-of-band checksum for eachL sector that would provide end-to-end protection all the way from (presumablyK ECC-protected) main memory out to disk and back again, at some cost in host L CPU cycles for sectors the host chose to protect that way - and in fact someJ high-end storage boxes do things like this internally, though can't extend$ them transparently into host memory.     An RMS bucket thus might getG > partially updated, but then the bucket checksum will flag that error.w  I Unless RMS bucket structure changed to include actual checksums somewherehI along the line, you are giving it too much credit here.  The last I knew,kH RMS used an incrementing byte at the start of a bucket and correspondingK byte at the end to catch *common* cases of corruption, but this won't catch D things like a write that the disk reorders (to take advantage of itsL rotational position) such that it starts writing in the middle of the bucketL and then fails either before reaching the bucket's end or after writing bothI start and end check-bytes but before wrapping completely back to where ito began.  F Furthermore, even check-bytes are present only in indexed-file bucketsH (unless they got added to Relative file buckets, but they don't get usedB that much anyway).  There's no protection whatsoever (checksums orG check-bytes) in Sequential files, though RMS may cursorially verify thehL structural integrity of what it reads in by checking for things like illegal$ record lengths in a VAR record file.   - bill   ------------------------------  # Date: Tue, 04 Jun 2002 01:17:49 GMTr* From: "Bill Todd" <billtodd@metrocast.net>, Subject: Re: Volume shadowing and a disasterB Message-ID: <1zUK8.110478$Kp.11205452@bin7.nnrp.aus1.giganews.com>  > "Keith Parris" <keithparris_NOSPAM@yahoo.com> wrote in message7 news:cf15391e.0206031522.3806f563@posting.google.com...a   ...m  G > A data transfer will consist of a Command (Write), followed by a dataiC > transfer operation, followed by a status return.  If the bits getiD > corrupted in-flight, it's likely the checksums on the packets willC > cause the packets to be rejected.  So for corruption to occur, it E > would almost certainly have to be an in-memory corruption somewherehE > along the line, corrupting data which is just about to be sent to aoF > disk.  The corruption would have to be such that it slipped past theF > single-bit correction and double-bit detection in the ECC in memory.  L See the description of transient behavior in PCs during power failure that IK just wrote in another response.  It's effectively a kind of interface erroriJ (one side of the interface has died, but the other hasn't yet) and results2 in bogus data that from that point on looks valid.  E > And this corruption would have to be just right to avoid causing an3H > RMS bucket checksum mismatch if it got all the way to the remote disk.  H Again, I fear you may be overestimating RMS's protective mechanisms (see details in other post).e   ...s  H > But to protect against such a possibility, you'd probably be using RMSD > Journalling (or similar techniques with a database product).  ThatE > should allow you to roll back an erroneous transaction afterward ifS > necessary.  J While that's likely with a database (since it's probably not a user optionH there), my guess would be that perhaps most environments that strive forJ disaster-tolerance don't recognize the need for it with RMS.  And *almost*) all the time they'll get away without it.e  G On the other hand, most logging implementations I've seen or read about H assume that log sectors will either be written correctly, not written atK all, or (rarely) partially written and caught by the disk's ECC mechanisms.dC I.e., the database (or possibly RMS) log itself may be undetectibly-G corrupted in such cases by the kind of failure I described, because thehJ implementors did not realize that it needed to be protected against (e.g., by a log-record checksum).   - bill   ------------------------------  % Date: Mon, 03 Jun 2002 22:02:40 -0400 ' From: Howard S Shubs <howard@shubs.net>h> Subject: Re: Would you like to see this on the VMS freeware CD< Message-ID: <howard-D7167C.22024003062002@enews.newsguy.com>  9 In article <3CFB54AF.1113693D@NOSPAM.byron.ext.telia.se>,d0  ">>> ^P" <plj@NOSPAM.byron.ext.telia.se> wrote:  * > Yes, I would like a copy to look at now.  A Me too, as the concept of USB devices in VMS is something I find y: amusing.  Seems like it might be a kind of security issue.   -- o# "Run in circles, scream and shout!"a I hope you have good backups! ) Are there any more networked SJFs around?V   ------------------------------   Date: 3 Jun 2002 14:11:31 -0600-- From: koehler@encompasserve.org (Bob Koehler)-) Subject: Re: [Fwd: Defect-free Software?]e3 Message-ID: <SJ$GjKdVr$mK@eisner.encompasserve.org>-  Q In article <3CFB8476.5F1948F5@cisco.com>, J Ahlstrom <jahlstro@cisco.com> writes:n  O >> The test engineers say that they can't imagine how they could design a worse-K >> load on the system than their current test load within the192 hours, and.O >> that they believe this load is way beyond the heaviest / worst ever recordedrC >> at one of their customers.  There also are physical upper limitsoJ >> (bottlenecks) on the speed and volume of the demands that can be pumpedN >> through in live operation, which the testers artificially bypass and exceed >> in the test lab.a  H    Their test engineers have poor imagination.  Let a couple crackers at6    it.  Try feeding it some not-so-well defined input.   ------------------------------   End of INFO-VAX 2002.307 ************************