1 INFO-VAX	Sat, 17 Jun 2000	Volume 2000 : Issue 336       Contents:P Re: (DECNET-PLUS strange poblem) Re: Proxy problem: Why does node:: workbut 0:: ) ANNOUNCEMENT: The DEC-rescue mailing list  Bind failing with IVADDR% Re: Compaq C v6.2 Single User License % RE: Compaq C v6.2 Single User License % Re: Compaq C v6.2 Single User License 7 Re: Compatible between OpenVMS V6.2 & OpenVMS AXP V7.1.  Re: CONV$RECLAIM question  Re: CONV$RECLAIM question  Re: CONV$RECLAIM question  Re: Fun VMS Facts? Re: Fun VMS Facts? Re: GNU-CC on VAX ) How do I calculate CONNECT TIME from DCL? - Re: How do I calculate CONNECT TIME from DCL? - Re: How do I calculate CONNECT TIME from DCL? - RE: How do I calculate CONNECT TIME from DCL?  Re: Intermec Barcode IP printer * Re: lib/obj question, symbols not going in* Re: lib/obj question, symbols not going in Re: Locked out... Please help  Re: Mail batch log file  Re: MOZILLA M16 ? # Netscape v3.03 and Time Calculation . RE: OpenVMS clusters vs other systems clusters. Re: OpenVMS clusters vs other systems clusters. Re: OpenVMS clusters vs other systems clusters. Re: OpenVMS clusters vs other systems clusters. Re: OpenVMS clusters vs other systems clustersA Re: OpenVMS commentaries (was Re: Gartner commentary on Wildfire) A Re: OpenVMS commentaries (was Re: Gartner commentary on Wildfire) . Oracle 8.1.6 Availability on OpenVMS 7.1 Alpha2 RE: Oracle 8.1.6 Availability on OpenVMS 7.1 Alpha Re: patches, graphic cards3 Re: Request information about GNU C for VMS (Alpha) ( Re: Searching for  " in files using dcl.; Storage Works / Snapshots / Maybe it's time to skip OpenVMS ? Re: Storage Works / Snapshots / Maybe it's time to skip OpenVMS $ Unix to OpenVMS portability concerns Re: VAX on Intel?  Re: VAX on Intel?  Re: VAX on Intel?  Re: Vest problem& Re: VMS File Caching Futures and so on& Re: VMS File Caching Futures and so on Re: VMS Security features   F ----------------------------------------------------------------------   Date: 16 Jun 2000 18:17:35 GMT* From: bleau@umtof.umd.edu (Lawrence Bleau)Y Subject: Re: (DECNET-PLUS strange poblem) Re: Proxy problem: Why does node:: workbut 0::  ) Message-ID: <8idqvv$hkm$1@hecate.umd.edu>   c In article <043501bfd729$8601d040$020a0a0a@xile.realm>, "John E. Malmberg" <wb8tyw@qsl.net> writes: / [John is reproducing the problem on his system] ' >OpenVMS 7.2 (MULTIA) with DECNET-PLUS.   ( I'm running 7.1-2, in case that matters.   >The existing files: >  >$ show log net*proxy + >  "NET$PROXY" = "SYS$SYSTEM:NET$PROXY.DAT" ) >  "NETPROXY" = "SYS$SYSTEM:NETPROXY.DAT"   + I got rid of the logical names shown above.     >$Directory SYS$SYSROOT:[SYSEXE] > 6 >NET$PROXY.DAT;1           30  21-JUN-1999 21:13:16.066 >NETPROXY.DAT;1            12  21-JUN-1999 21:13:16.80   I deleted these files, too.     >Now to start over (temporarily) >  >$set server security/exit& >$rename sys$system:net$proxy.dat .old% >$rename sys$system:netproxy.dat .old  >$set server/start security    Did this, too.  D >Recreate the empty proxy files.  Note that two files are created as
 >expected. > J >Since you are only getting one file, I suspect that it is related to your	 >problem.   O A note on this.  I just checked elsewhere in SYS$SYSROOT:, and lo and behold, I G have a NETPROXT.DAT in [SYSMGR] !  Its creation date is the same as the N NET$PROXY.DAT in SYS$SYSTEM:.  So, it would appear that, in the absence of theK logical names, a AUTH CREATE/PROXY creates NET$PROXY.DAT in SYS$SYSTEM: and J NETPROXY.DAT in the default directory, which was SYS$MANAGER: at the time.O At least this is the V7.1-2 behavior.  Since you have the logicals defined, bot . were created in the same directory ([SYSEXE]).  J I just copied the NETPROXY.DAT into SYS$SYSROOT:[SYSEXE], then I shut downF and restarted the security server.  I still get the same failure mode.   [snip]  K >Add a proxy to the http account for a web server.  This immediately breaks I >access to the SYSTEM account though DECNET.  Consistent with what I have 
 >asserted.  B I'll grant you this; fine.  It doesn't address my problem, though.   [snip]  K >Tests by removing the PROXY files one at a time show that it does not seem 1 >to care about the NETPROXY file being missing.     H In V7.2 of VMS, you mean.  I'd accept the conclusion if you were running V7.1-2.   K Btw, another respondent suggested that I install the UPDATE patch for V7.2, M saying that it fixed a similar problem.  I pointed out to him that I can't do G that, since I'm running V7.1-2.  I *do* have the UPDATE patch for 7.1-2  installed, though.  M >Conclusions:  The PROXY database is not your problem, and the PROXY database " >does work the way I said it does.  M I never doubted the latter part; it's just not germane to my narrowly phrased L problem statement.  I have made note of it, though, for when this problem is3 resolved and I want to start up the OSU server s/w.   M Note to late comers: What I have is definitely a configuration problem, not a L VMS design flaw, since another system of mine that runs the same VMS versionM works in both modes.  What John has done is eliminate the proxy database as a $ source of the configuration problem.  K >I am not sure what is causing the behavior of what you are seeing.  I have ; >three DECNET-PLUS hobby systems, and can not reproduce it.   I The other respondent I mentioned suggested I also look at what I get from > REPLY/ENABLE .  I did this, and fwiw here's the opcom message:  / Auditable event:          Network login failure 1 Event time:               16-JUN-2000 14:01:55.94 " PID:                      00000090! Process name:             NET$ACP & Username:                 DNA$SessCtrl. Remote node id:           490006AA000400F71A21 Remote node fullname:     6903 Remote username:          BLEAU G Status:                   %LOGIN-F-NOTVALID, user authorization failure   L Side question: The full node name is still a number, 6903.  It's the currentL number, btw.  Why isn't this translated to LOCAL:.ULEIS:: ?  Could this be aL network database problem after all?  In case readers missed an earlier post,L here's when I get from the DECNET_REGISTER utility (invoked with /C option):   DECNET_REGISTER> SHO NODE ULEIS   " Directory Service: Local name file   Node name:         LOCAL:.uleis  Phase IV synonym:  ULEIS9 Node address:      49::00-06:AA-00-04-00-F7-1A:20 (6.759)    Number of nodes reported on:  1   L I could believe there's a strange network database problem if using the nameN ULEIS:: failed but 0:: worked.  I'm ready to try anything now, though.  Ideas?   Lawrence Bleau University of Maryland" Physics Dept., Space Physics Group 301-405-6223 bleau@umtof.umd.edu    ------------------------------  % Date: Sat, 17 Jun 2000 00:10:38 -0500 ' From: Bill Bradford <mrbill@mrbill.net> 2 Subject: ANNOUNCEMENT: The DEC-rescue mailing list. Message-ID: <20000617001038.M15984@mrbill.net>  B This is to announce the existence of the DEC-rescue mailing list, B dedicated to keeping DEC hardware (of any type, but mainly PDP andA VAX stuff) out of the hands of scrappers, trashers, and the DUMP!   B I know we all hate to see good hardware go to the scrap yard or be sold off by the pound.  A I run another list of this type (rescue@sunhelp.org) for fans of  / Sun hardware, and it has been quite successful.   + For more info about the DECrescue list, see ' http://www.decvax.org/mailman/listinfo/    Thanks.    Bill   --  * +--------------------+-------------------+* |   Bill Bradford    |   Austin, Texas   |* +--------------------+-------------------+* | mrbill@sunhelp.org | mrbill@mrbill.net |* +--------------------+-------------------+   ------------------------------  % Date: Fri, 16 Jun 2000 12:05:13 -0700 @ From: "Russell E. Owen" <owen@astroNOJNK.washington.edu.invalid>! Subject: Bind failing with IVADDR 2 Message-ID: <8idtp9$a7gu$1@nntp6.u.washington.edu>  F I am attempting to bind a socket using SYS$QIO and getting an IVADDR.  The manual says of this error:  H Programming error. The INET address you specified using the IO$_SETMODE E function was not placed into the system. This resulted in an invalid  F port number or INET address combination. The INET address was invalid " for one of the following reasons: H 1.  Port zero and INET address zero are not allowed or port zero is not < allowed when using an IO$_ACCESS or IO$_WRITEVBLK function. H 2.  You attempted to exceed the limit of allowable permanent entries in  the ARP table.  G 3.  You attempted to bind a raw IP socket when there are no interfaces   defined in the system.  B 4.  You attempted to bind a raw IP socket to a null INET address.        H I'm sure it *is* a programming error, but I'm stumped. I can't see that I any of the cases listed apply. I have appended a short test program. I'm  G attempting to create a UDP socket for broadcast. Any insights would be   appreciated.   Notes:H - Changing the bind to using INET$C_INADDR_ANY as the address works (no ? errors are returned). But of course I'm trying to broadcast.... D - Broadcasting requires OPER privilege. But the IVADDR error occurs # whether I have OPER enabled or not.   
 -- Russell       PROGRAM TestUDP * ! Opens a UDP socket for broadcast output.C ! Minimal test program. Create and bind are separate to show error.      IMPLICIT NONE        ! includes     INCLUDE 'tinc:NetDef' '     INCLUDE 'SYS$LIBRARY:TCPIP$INETDEF'      INCLUDE '($IODef)'       ! functions       INTEGER sys$Assign, sys$QIOW!     EXTERNAL sys$Assign, sys$QIOW        ! variables (     BYTE portNetByte(2), portHostByte(2)'     INTEGER portNetshort, portHostShort +     EQUIVALENCE (portNetByte, portNetshort) -     EQUIVALENCE (portHostByte, portHostShort)      INTEGER*2 chanNum      INTEGER vmsCond, ii      STRUCTURE /ioStat_type/ +        INTEGER*2 vmsCond, nChar, dum1, dum2      ENDSTRUCTURE     RECORD /ioStat_type/ ioStat      STRUCTURE /socketChar_type/         INTEGER*2 protocol         BYTE type, domain     ENDSTRUCTURE'     RECORD /socketChar_type/ socketChar       RECORD /SOCKADDRIN/ sockAddr     STRUCTURE /itemList2_type/        INTEGER*2 len, type        INTEGER loc     END STRUCTURE &     RECORD /itemList2_type/ sockAddrIL ! begin /     ! set port and byte-swap into network order      portHostShort = 1541$     portNetByte(1) = portHostByte(2)$     portNetByte(2) = portHostByte(1)       ! Obtain an I/O channel 7     vmsCond = sys$Assign ('TCPIP$DEVICE:', chanNum, , ) 4     IF (.NOT. vmsCond) CALL lib$stop (%val(vmsCond))       ! Create the UDP socket %     socketChar.protocol = TCPIP$C_UDP #     socketChar.type = TCPIP$C_DGRAM '     socketChar.domain = TCPIP$C_AF_INET   ;     vmsCond = sys$QIOW (, %val(chanNum), %val(io$_SetMode), #     2  ioStat, , , socketChar,,,,,) )     IF (vmsCond) vmsCond = ioStat.vmsCond 4     IF (.NOT. vmsCond) CALL lib$stop (%val(vmsCond))       ! Bind the UDP socket +     sockAddr.sin$w_family = TCPIP$C_AF_INET &     sockAddr.sin$w_port = portNetshort1     sockAddr.sin$l_addr = INET$C_INADDR_BROADCAST '     DO ii = 1, LEN(sockAddr.sin$t_zero) +        sockAddr.sin$t_zero(ii:ii) = CHAR(0) 	     ENDDO   '     sockAddrIL.type = TCPIP$C_SOCK_NAME &     sockAddrIL.len = SizeOf (sockAddr)$     sockAddrIL.loc = %LOC (sockAddr)   ! this call fails ;     vmsCond = sys$QIOW (, %val(chanNum), %val(io$_SetMode), %     2  ioStat, , , , , sockAddrIL,,,) )     IF (vmsCond) vmsCond = ioStat.vmsCond 4     IF (.NOT. vmsCond) CALL lib$stop (%val(vmsCond))     END  ! end net_BrdUDPOpen   ------------------------------  + Date: Fri, 16 Jun 2000 17:10:29 +0000 (GMT) , From: Bill McLaughlin <mcbill20@hotmail.com>. Subject: Re: Compaq C v6.2 Single User License4 Message-ID: <20000616171029.31521.qmail@hotmail.com>   Todd Nelson wrote:M >I am in need of a single user license (part number QL-015AA-2B) for >Compaq   >C >version 6.2 OpenVms Alpha.  > D >I was quoted a price of around $1000.00 from compaq services... wasK >wondering if anyone had this license floating around that wanted to >sell   >it  >cheap.  >  >Thanks  >  >Todd.   Todd, K I am assuming you need a license for commercial use rather than a personal  = one. (For a personal one you could use the hobbyist license.)   F Digital's policy was that licenses could not be transferred, with the K exception of VMS and clustering. And the VMS and cluster licenses are tied  G to the hardware, so whenever you sell the hardware, you are implicitly  F selling the VMS and cluster licenses. If you bought a layered product I license (C, BASIC, MMS, etc.), you cannot transfer those licenses to the   buyer of the machine.   M I found all this out the hard way-- I purchased an Alpha from a company that  L was going out of business. They were selling all their computers for 35% of H book value. I paid for base VMS, 10 additional user VMS, clustering, C, L DECset, TCPIP, CMS and DFG. When I called Decpaq, they told me that I could L pay $300.00 to transfer the VMS and clustering, but the other licenses were  no good.  L So, theoretically, you own any VMS licenses that were purchased with or for H that particular machine but you'll have to get the single-user one from  Decpaq if you need more.  L I am a little fuzzy on policy differences between "base VMS" and additional 0 users. Maybe someone else here can correct me...   Bill McLaughlin H ________________________________________________________________________H Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com   ------------------------------  % Date: Fri, 16 Jun 2000 15:32:44 -0300 1 From: "Boyle, Darren" <boyledj@bankofbermuda.com> . Subject: RE: Compaq C v6.2 Single User LicenseK Message-ID: <9F664D538536D411BD3200508B6FF01A024780@bdant027.bda.bobda.com>    Todd,   D 	If you just want one person to use this system (and it's you) loginL from the console, there is no license required for a single user to login toJ the console, obviously if this is for somebody else to use your system you+ would not want to give them console access.  - Darren   > ----------6 > From: 	Todd Nelson[SMTP:toddnelson@lehighcounty.org]' > Sent: 	Friday, June 16, 2000 12:54 PM  > To: 	Info-VAX@Mvb.Saic.Com- > Subject: 	Compaq C v6.2 Single User License  > L > I am in need of a single user license (part number QL-015AA-2B) for Compaq > C  > version 6.2 OpenVms Alpha. > E > I was quoted a price of around $1000.00 from compaq services... was J > wondering if anyone had this license floating around that wanted to sell > it > cheap. >  > Thanks >  > Todd.  >  >  >     F **********************************************************************C This message and any files transmitted with it are confidential and J may be privileged and/or subject to the provisions of privacy legislation.M They are intended solely for the use of the individual or entity to whom they L are addressed. If the reader of this message is not the intended recipient, B please notify the sender immediately and then delete this message.I You are notified that reliance on, disclosure of, distribution or copying  of this message is prohibited.   Bank of Bermuda F **********************************************************************   ------------------------------  % Date: Fri, 16 Jun 2000 21:07:44 +0200 = From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com> . Subject: Re: Compaq C v6.2 Single User License) Message-ID: <394A7AFF.A6D2222A@gtech.com>    Todd Nelson wrote:N > I am in need of a single user license (part number QL-015AA-2B) for Compaq C > version 6.2 OpenVms Alpha. > E > I was quoted a price of around $1000.00 from compaq services... was M > wondering if anyone had this license floating around that wanted to sell it  > cheap.  " Check the license rules carefully.  0 If I remember correct, then you can not just buy a license from somebody else.    Stupid, but ...    Arne   ------------------------------  % Date: Fri, 16 Jun 2000 16:55:15 +0200 = From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com> @ Subject: Re: Compatible between OpenVMS V6.2 & OpenVMS AXP V7.1.) Message-ID: <394A3FD3.4882D94D@gtech.com>    Denis Shadrin wrote:d > Does anybody know, will my C libraries be worked on OS : OpenVMS AXP V7.1, Model :AlphaServer 4100 > 5/600, CPU type : 4100, > > if it was built on DEC Alpha/OpenVMS V6.2 by DEC C V5.0-003?   They should.  5 (unless you are really doing some very special stuff)e   Arne   ------------------------------  # Date: Fri, 16 Jun 2000 19:43:23 GMTe% From: jlsue <jlsuexxxz@dialupnet.com>V" Subject: Re: CONV$RECLAIM question8 Message-ID: <mi0lksotnnomtq5kgk9f798a78il2mti26@4ax.com>  7 I'm not sure I'd actually ever delete records for this.VB I'd opt for just using a unique 'record id' for each one, and justA update with nonsense/invalid information when it's not needed anye more.M  D Of course then you have to develop some mechanism for allocating the" invalid records to new requests.    0 Could a direct-access file work better for this?1 Not speaking for anyone, certainly not DEC/CompaqD- (get rid of the xxxx in my address to e-mail)/   ------------------------------  % Date: Fri, 16 Jun 2000 16:29:44 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca>t" Subject: Re: CONV$RECLAIM question, Message-ID: <394A8E37.A34CCAAC@videotron.ca>   jlsue wrote: > 9 > I'm not sure I'd actually ever delete records for this.sD > I'd opt for just using a unique 'record id' for each one, and justC > update with nonsense/invalid information when it's not needed any  > more.   M But because of my need to access records grouped by secondary key (and insideoM that key, in the order they were inserted), I would need to re-invent indexed K files if I were to implement it with just relative files or raw data. And I H woudl rather trust RMS to do this effectively and safely than home grownL sofwtare which I would need to test and retest. No point in re-inventing the wheel for this.i    J I don't see why people want to avoid indexed files like the plague. Once IS know what the best way of building one and cleaning up, it is not such a bad issue.n   ------------------------------  % Date: Fri, 16 Jun 2000 20:25:31 -0400 ' From: "Bill Todd" <billtodd@foo.mv.com>e" Subject: Re: CONV$RECLAIM question( Message-ID: <8iegal$l6n$1@pyrite.mv.net>  8 JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message& news:394A8E37.A34CCAAC@videotron.ca...   ...E  L > I don't see why people want to avoid indexed files like the plague. Once IJ > know what the best way of building one and cleaning up, it is not such a
 bad issue.  L It's not that people want to avoid indexed files per se, it's that they wantJ to avoid using them for purposes for which they weren't designed and don'tL work particularly well.  I was around for the original RMS design, and I canF assure you that this is one such case (around 1978 Chris Saether and IK batted around the idea of a 'no RFA' option for single-key files that would H have avoided the need for periodic 'cleaning up', but that wouldn't have, provided the multi-key access you're using).  I You have a rather specific need.  If you can figure out how to get RMS to E handle it well enough for your needs, and don't mind the necessity of:E periodic maintenance, then by all means go that route.  But it's alsorK reasonable to consider building something special-purpose, which can be fartG less general than RMS and hence need not entail huge amounts of work or 
 verification.p   - bill   ------------------------------  % Date: Fri, 16 Jun 2000 16:05:13 +0200r2 From: martin@radiogaga.harz.de (Martin Vorlaender) Subject: Re: Fun VMS Facts?c; Message-ID: <394a3419.524144494f47414741@radiogaga.harz.de>:  . John Nixon (jorlnixon@worldnet.att.net) wrote:9 : "Martin Vorlaender" <martin@radiogaga.harz.de> wrote...iL : > Rich Marcello just pointed out today in the OpenVMS Directions TeleforumK : > that Kevin Mitnick stated the only OS he couldn't break into easily waso : > VMS. :dE : Can you point me to any kind of link that I can use to support that  : statement.L : It may help tip the scales in a very large decision regarding OVMS that we : have coming up real soon.i  ; OK, I pulled out the slides again, and the exact wording isw   <QUOTE> G OpenVMS was the only environment that the world's most notorious hackero4 couldn't break into as testified to in U.S. Congress </QUOTE>  1 (Rich mentioned Mitnick's name along this slide.)]  H I guess there was a Congess (commitee?) hearing where Mitnick said that.  C Apart from that, I have no idea where to point you to - this was anV- invitational (does one say so?) presentation.    cu,n   Martin --D                        |  Martin Vorlaender  |  VMS & WNT programmer1   OpenVMS: When you    |  work: mv@pdv-systeme.de H   KNOW where you want  |        http://www.pdv-systeme.de/users/martinv/8   to go today.         |  home: martin@radiogaga.harz.de   ------------------------------  # Date: Fri, 16 Jun 2000 23:27:15 GMTd( From: Terry Kennedy <terry@gate.tmk.com> Subject: Re: Fun VMS Facts?t' Message-ID: <Fw9t5F.41t@spcuna.spc.edu>o  4 David Mathog <mathog@seqaxp.bio.caltech.edu> writes:L > I'm not saying that there's not a plus side to OpenVMS, but IO performanceM > is so much worse than the competition that nobody in their right mind wouldoK > buy one now for PC and small systems applications that were in any way IOe2 > intensive.   Ie, file serving, web serving, etc.  H   That's certainly true with the C RTL (which has made huge improvementsG since the VAX C 3.2 days, but is still quite inefficient at many common  tasks).   J   If you're not writing in C, or if you use direct calls to RMS instead of) the C RTL you can get much higher speeds.   - 	Terry Kennedy             http://www.tmk.com 5         terry@tmk.com             Jersey City, NJ USA    ------------------------------  % Date: Fri, 16 Jun 2000 16:51:20 +02009= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>  Subject: Re: GNU-CC on VAX) Message-ID: <394A3EE7.691D206F@gtech.com>    Mark wrote:06 > Does anyone know where I will find GNU-CC for a VAX? > W > I think I currently have all the files but don't know what logicals need to be setup. 4 > And I would like to get the latest version aswell.   There are some links at:  ) http://www.levitte.org/~ava/vms_gnu.htmlxC   Arne   ------------------------------    Date: 16 Jun 2000 16:16:33 -05001 From: kaplow_r@eisner.decus.org.mars (Bob Kaplow)r2 Subject: How do I calculate CONNECT TIME from DCL?+ Message-ID: <$kZLwvFdgWe6@eisner.decus.org>i  J I'm looking for a way in DCL to get connect time, similar to what is shown by:n   $ sho proc/acc  G 16-JUN-2000 14:11:08.33   User: KAPLOWRO         Process ID:   2023768Eo>                           Node: CERES            Process name: "RocketScientist"    Accounting information:tA  Buffered I/O count:     51729  Peak working set size:       7152rA  Direct I/O count:       20324  Peak virtual size:         182736oA  Page faults:            16067  Mounted volumes:                0-  Images activated:         184)  Elapsed CPU time:          0 00:01:19.31@C  Connect time:              0 06:48:35.18  <<<< this is what I wants   I can get login time from:  ( write sys$output f$getjpi("","logintim") 16-JUN-2000 07:22:33.15a  ; but what I need is the elapsed time from login time to now.uF DAYS_BETWEEN_DATES does the day calculation, but doesn't do the hours,+ minutes, or seconds. Any clever suggestionsd+                                            f 	Bob Kaplow	  E SPAM:	spamrecycle@ChooseYourmail.com	uce@ftc.gov	postmaster@127.0.0.1u   ------------------------------  # Date: Fri, 16 Jun 2000 19:53:14 GMT = From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) 6 Subject: Re: How do I calculate CONNECT TIME from DCL?0 Message-ID: <009EBB33.71B9E020@SendSpamHere.ORG>  _ In article <$kZLwvFdgWe6@eisner.decus.org>, kaplow_r@eisner.decus.org.mars (Bob Kaplow) writes:-K >I'm looking for a way in DCL to get connect time, similar to what is shown  >by: >, >$ sho proc/acc5 > H >16-JUN-2000 14:11:08.33   User: KAPLOWRO         Process ID:   2023768E? >                          Node: CERES            Process name:t >"RocketScientist" >  >Accounting information:B > Buffered I/O count:     51729  Peak working set size:       7152B > Direct I/O count:       20324  Peak virtual size:         182736B > Page faults:            16067  Mounted volumes:                0 > Images activated:         184 * > Elapsed CPU time:          0 00:01:19.31D > Connect time:              0 06:48:35.18  <<<< this is what I want >T >I can get login time from:  >p) >write sys$output f$getjpi("","logintim")_ >16-JUN-2000 07:22:33.15 >i< >but what I need is the elapsed time from login time to now.G >DAYS_BETWEEN_DATES does the day calculation, but doesn't do the hours,c, >minutes, or seconds. Any clever suggestions, >                                           
 >	Bob Kaplow	a >lF >SPAM:	spamrecycle@ChooseYourmail.com	uce@ftc.gov	postmaster@127.0.0.1  # http://www.tmesis.com/symbol/using/c  G Two procedures there.  One using SYMBOL and which actually computes thetH value you seek; another which uses PIPE to extract it from the output of a SHOW PROCESS command.e   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMf   ------------------------------  # Date: Fri, 16 Jun 2000 19:55:22 GMTe= From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-)r6 Subject: Re: How do I calculate CONNECT TIME from DCL?0 Message-ID: <009EBB33.BDE5794C@SendSpamHere.ORG>  p In article <009EBB33.71B9E020@SendSpamHere.ORG>, system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) writes:` >In article <$kZLwvFdgWe6@eisner.decus.org>, kaplow_r@eisner.decus.org.mars (Bob Kaplow) writes:L >>I'm looking for a way in DCL to get connect time, similar to what is shown >>by:  >> >>$ sho proc/acc >>I >>16-JUN-2000 14:11:08.33   User: KAPLOWRO         Process ID:   2023768EN@ >>                          Node: CERES            Process name: >>"RocketScientist"h >> >>Accounting information: C >> Buffered I/O count:     51729  Peak working set size:       7152,C >> Direct I/O count:       20324  Peak virtual size:         182736TC >> Page faults:            16067  Mounted volumes:                0   >> Images activated:         184+ >> Elapsed CPU time:          0 00:01:19.31/E >> Connect time:              0 06:48:35.18  <<<< this is what I wantl >> >>I can get login time from: >>* >>write sys$output f$getjpi("","logintim") >>16-JUN-2000 07:22:33.15  >>= >>but what I need is the elapsed time from login time to now. H >>DAYS_BETWEEN_DATES does the day calculation, but doesn't do the hours,- >>minutes, or seconds. Any clever suggestionsy- >>                                           F >>	Bob Kaplow	 >>G >>SPAM:	spamrecycle@ChooseYourmail.com	uce@ftc.gov	postmaster@127.0.0.1r >o$ >http://www.tmesis.com/symbol/using/ > H >Two procedures there.  One using SYMBOL and which actually computes theI >value you seek; another which uses PIPE to extract it from the output ofC >a SHOW PROCESS command.    H BTW, now that SYMBOL has been installed on eisner, you can check out the) SYMBOL specific procedures there.  Enjoy!k   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMd   ------------------------------  % Date: Fri, 16 Jun 2000 13:26:29 -0700_/ From: Terry Marosites <TMarosites@unitedad.com>06 Subject: RE: How do I calculate CONNECT TIME from DCL?M Message-ID: <1137A4A23A51D311B2D600105A1D5213019AEEA8@seantexch.unitedad.com>)  F Here are two methods that I was given by this group. I prefer method 1 Terry & PS thanks to all who gave there inputs    
 $ method_1$ $ login_dt = F$GetJPI("","LOGINTIM")3 $ login_time = F$CVTime(login_dt,"ABSOLUTE","TIME")N1 $ login_day = F$CVTime(login_dt,"ABSOLUTE","DAY")-* $ now = F$CVTime("","ABSOLUTE","DATETIME")/ $ current_time= F$CVTime(now,"ABSOLUTE","TIME")i. $ current_day = f$CVTime(now,"ABSOLUTE","DAY")  $ If current_day .EQS. login_day $ ThenG $   connect_time = F$CVTIME("''now'-0-''login_time'","ABSOLUTE","TIME")l $ ElseD $   If (F$Integer(current_day) - F$Integer(login_day) .GT. 1) .OR. -L        (F$CVtime("''now'-0-''login_time'","ABSOLUTE","DAY") .NES. login_day) $   Then) $     connect_time = "more than 24 hours"  $   ElseA $     tmp1 = F$CVTIME("''now'-0-''login_time'","ABSOLUTE","TIME") F $     tmp2 = F$CVTIME("''login_dt'-0-''login_time'","ABSOLUTE","TIME")D $     connect_time = F$CVTIME("''tmp1'-0-''tmp2'","ABSOLUTE","TIME")	 $   EndIfo $ EndIft $ write sys$output connect_timer      	 $Method_2x= $ PIPE ( SHOW PROCESS /ACCOUNTING /ID='F$GETJPI("","PID") | - &          SEARCH SYS$PIPE "connect" | -H          ( READ SYS$PIPE $TMP$ ; DEFINE /JOB /NOLOG $TMP$ &$TMP$ ) ) ; -        CONNECT_TIME == -D     F$edit(F$trnlnm("$TMP$"),"TRIM,COMPRESS") - "Connect time: " ; -        DEASSIGN /JOB $TMP$ $ CONNECT_TIME == CONNECT_TIME   -----Original Message-----L From: kaplow_r@eisner.decus.org.mars [mailto:kaplow_r@eisner.decus.org.mars]# Sent: Friday, June 16, 2000 2:17 PM  To: Info-VAX@Mvb.Saic.Coms2 Subject: How do I calculate CONNECT TIME from DCL?  J I'm looking for a way in DCL to get connect time, similar to what is shown by:0   $ sho proc/acc  G 16-JUN-2000 14:11:08.33   User: KAPLOWRO         Process ID:   2023768Ea>                           Node: CERES            Process name: "RocketScientist"h   Accounting information:tA  Buffered I/O count:     51729  Peak working set size:       7152 A  Direct I/O count:       20324  Peak virtual size:         182736rA  Page faults:            16067  Mounted volumes:                0_  Images activated:         184)  Elapsed CPU time:          0 00:01:19.31 C  Connect time:              0 06:48:35.18  <<<< this is what I want    I can get login time from:  ( write sys$output f$getjpi("","logintim") 16-JUN-2000 07:22:33.15b  ; but what I need is the elapsed time from login time to now.sF DAYS_BETWEEN_DATES does the day calculation, but doesn't do the hours,+ minutes, or seconds. Any clever suggestionsn*                                                    Bob Kaplow     e  L SPAM:   spamrecycle@ChooseYourmail.com  uce@ftc.gov     postmaster@127.0.0.1    5 *****************************************************o    5 *****************************************************-4 Any views or opinions are solely those of the author) and do not necessarily represent those of1 United News& Media.-5 ***************************************************** 4 The information transmitted is intended only for the1 person or entity to which it is addressed and mayo3 contain confidential and/or privileged material. Ifr3 you are not the intended recipient of this message,n. please do not read, copy, use or disclose this3 communication and notify the sender immediately. It 0 should be noted that any review, retransmission,2 dissemination or other use of, or taking action in- reliance upon, this information by persons orn- entities other than the intended recipient ist prohibited.s5 *****************************************************f **   ------------------------------  # Date: Sat, 17 Jun 2000 04:34:58 GMTe- From: "Dave Pampreen" <davepampreen@home.com>e( Subject: Re: Intermec Barcode IP printer9 Message-ID: <SdD25.2773$fR2.34402@news1.rdc1.mi.home.com>n  L We have some Intermec 3400 models, and create the queues with /nolibrary.  II don't think this makes a difference for your issue, but it's just a note.* We do, however, use port 9100-   Dave    8 "Peter Weaver" <peter.weaver@stelco.ca> wrote in message) news:skkkqkpbis4150@corp.supernews.com...eH > It has been a few years since I setup an Intermec Barcode printer, but2 > see my notes inline, they may or may not help :) >m > 4 > Swaffield, Harold [CRK:1342:EXCH] wrote in message+ > <394A2831.FBB7B8FE@americasm01.nt.com>...  > >...H > >working, as it was shipped to us by Intermec. Hardware does work SelfC > >test, and the serial port are okay. A sniffer indicates that thei > Alphaa@ > >is sending information to the print queue, the ip we setup is >e >oG > Check you sniffer to compare what the IP printer is receiving against F > what the LAT printer sees. IIRC the first few characters the BarcodeF > printer sees must be a certain string to tell the printer that it isG > valid data, if the IP data stream has some other data inserted beforegA > your start of data then everything the printer sees is ignored.s >yE > >broadcasting, but the printer does not print. I have been in touchl > withB > >Intermec, and have passed through the first level support. ThisF > >gentleman had me setup the printer so that it would print raw data, > butiF > >that still would not work.  The reply from the second level support >aF > IIRC we had problems trying the raw data mode, it did not seem to do > what it was supposed to do.u >i > >...H > >init/queue/start/process=ucx$telnetsym/on="47.180.0.105:9100"/library	 > =quedevr > >tH > >ctl/default=(flag=one,feed,noburst,form=letter)/separate=(noflag,nobu	 > rst,notu > >p > >railer) paltest_node1 > >r >i >wE > Do you have the FLAG=ONE on the other LAT Barcode printers that aresF > working? My guess is that flag page is not needed and may be gettingC > in the way. I also don't remember (it was a few years ago though)y > having to use any library. >g   ------------------------------   Date: 16 Jun 2000 12:02:26 PDTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515)3 Subject: Re: lib/obj question, symbols not going ine3 Message-ID: <W4kbAlWUA12G@mccdev.slac.stanford.edu>r  - In article <8idj1k$1cb@gap.cco.caltech.edu>,  9     	mathog@seqaxp.bio.caltech.edu (David Mathog) writes:rA > I'm seeing something funny with the librarian in OpenVMS 7.2-1.a >  > $ cc nw.ct' > $ cc/define=full/obj=full_nw.obj nw.cm > $ symbols full_nw.objg     ^^^^^^^o  *         This is _which_ program/command???  O > Type    Symbol Name                         Flags  Value  Psindx    P3     P4eR > --------------------------------------------------------------------------------O > SYM  FULL_NW                                 prgs      0      3      0      1  > $ symbols nw.obj > O > Type    Symbol Name                         Flags  Value  Psindx    P3     P4fR > --------------------------------------------------------------------------------O > SYM  NW                                      prgs      0      3      0      1  >  > but watch: > % > $ library/obj common.olb full_nw,nwc > $ lib/list common.olbE > NW( > $ lib/object/create common.olb full_nw > $ lib/list common.olbe > NW  H         Could you please provide  the  output from LIB/LIST/FULL/NAME onH     each  of  the  two libraries?  I'd like to see  some  more  complete0     information that's missing from the above...           -Ken -- pM  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edul:  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------  % Date: Fri, 16 Jun 2000 21:12:43 +0200o= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>e3 Subject: Re: lib/obj question, symbols not going in ) Message-ID: <394A7C2B.1309C28A@gtech.com>g   David Mathog wrote:tA > I'm seeing something funny with the librarian in OpenVMS 7.2-1.i >  > $ cc nw.cm' > $ cc/define=full/obj=full_nw.obj nw.ct > $ symbols full_nw.objn > O > Type    Symbol Name                         Flags  Value  Psindx    P3     P4tR > --------------------------------------------------------------------------------O > SYM  FULL_NW                                 prgs      0      3      0      1m > $ symbols nw.obj > O > Type    Symbol Name                         Flags  Value  Psindx    P3     P44R > --------------------------------------------------------------------------------O > SYM  NW                                      prgs      0      3      0      1t >  > but watch: > % > $ library/obj common.olb full_nw,nwh > $ lib/list common.olbh > NW( > $ lib/object/create common.olb full_nw > $ lib/list common.olbn > NW > L > That is, the name in the object library ends up being "NW" even though theI > name of the function inside the object file, modified by the define, is  > really FULL_NW.K   ????  . LIBR/LIST does not show symbols only modules !   Use LIBR/LIST/NAME instead.n  I > Other than by making a copy of the input file with a matching name, andsF > compiling that, is there some way to get the FULL_NW symbol into theF > library?  When the linker tries to resolve it it cannot find it now.  C I think the problem is that you have two modules with the same namep9 and you can not store them into the same object library !    Arne   ------------------------------  % Date: Fri, 16 Jun 2000 16:50:36 +0200|= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>s& Subject: Re: Locked out... Please help) Message-ID: <394A3EBC.E831091D@gtech.com>/   Nivlesh Chandra wrote:L > Hi.. I have installed VMS 7.0 on a alpha system and was playing with it. IF > was locked out since I forgot my system password but then after someJ > research was able to find out how to get in by booting in conversationalA > mode and then setting UAFALTERNATE to 1. And since there was nogN > sysuafalt.dat file present I was able to get in. Now the problem is that theK > same thing has happenend again and this time I cannot use the above sinceoL > there is a sysuafalt.dat file in the system. I tried setting the parameterK > startup when I rebooted using conversational mode to OPAO: but the system(0 > gets stuck after giving me the following lines > , > Error opening primary input file SYS$INPUTA > Error in device name or inappropriate device type for operatione > 4 > Please help ... I need to get back into the system   See the VMS FAQ:  < #MGMT5.  I've forgotten the SYSTEM password - what can I do?   Arne  E PS: There are links to the VMS FAQ and many other relevant ressources,2     at http://www.levitte.org/~ava/vms_faq.htmlx !   ------------------------------   Date: 16 Jun 00 15:18:44 PST From: mckinneyj@cpva.saic.com0  Subject: Re: Mail batch log file( Message-ID: <RLON4MN7Rox0@cpva.saic.com>  1 In article <c%625.4096$227.94893@nnrp1.uunet.ca>,l@  "Stephane Paquin" <stephane.paquin@nospam.ispatnet.com> writes:M > Does anyone can suggest a way to mail the log file of a batch at the end ofwK > the batch ? Is it possible to obtain the batch log file name while in the  > batch procedure ?1  C Maybe this will do it for you? It was intended for invocation after:A trapping an error but I'm sure you can tailor it to your needs...e   $! mail_batchlog.com $ set :=
 $ set noon% $ set symbol/scope=(nolocal,noglobal)w! $ if p1 .eqs. "" .or. p2 .eqs. ""c $ then% $   status = 276        ! SS$_INSFARG  $   goto usage $ endif8 $ if f$mode() .nes. "BATCH"  $ then$ $   status = 852        ! SS$_IVMODE $   goto usage $ endifi $ notify := 'p1' $ error_status := 'p2' $ set output_ratet$ $ ctx = f$getqui("cancel_operation")@ $ this_job = f$getqui("display_job", "entry_number",,"this_job")< $ job_name = f$getqui("display_job", "job_name",,"this_job")> $ if .not. f$getqui("display_job", "job_log_null",,"this_job") $ then' $   location = f$environment("default")  $   set default sys$login: $   pid = f$getjpi("","pid")( $   temp_file = "sys$scratch:''pid'.tmp"F $   log_spec = f$getqui("display_job","log_specification",,"this_job")8 $   log_spec = f$parse(log_spec,"''job_name'.log") - ";". $   file_device = f$parse(log_spec,,,"device")* $   file_name = f$parse(log_spec,,,"name")* $   file_type = f$parse(log_spec,,,"type")% $   file_spec = file_name + file_typee+ $   pipe show device/file 'file_device' | --D     search/match=and/output='temp_file' sys$input: 'file_spec','pid' $   open/read input 'temp_file'e $   read input recordj $   close input  $   delete 'temp_file';0* $   file_version = f$element(1,";",record), $   log_spec = log_spec + ";" + file_version& $   type/output='temp_file' 'log_spec' $   mail/noself-&         /personal_name="''job_name'" -F         /subject="Status = ''$status' => ''f$message(error_status)'" -         'temp_file' 'notify' $   delete 'temp_file';0 $   set default 'location' $ endif  $ exit 1        ! SS$_NORMAL $  $usage:s $ type sys$input:c  ' Usage: @mail_batchlog <notify> <status>u  L   in batch mode only, where <notify> is a valid VMSmail address and <status>I   is the VMS status value that is to be delivered, along with the current /   batch logfile, by mail to the notify address.g   $ exit 'status'    ------------------------------  # Date: Fri, 16 Jun 2000 20:05:01 GMT ' From: Colin Blake <colin@theblakes.com>  Subject: Re: MOZILLA M16 ?- Message-ID: <394A8868.1F5E6CCD@theblakes.com>l  / Mozilla M16 contains the GTK+ 1.2.8 shareables.    ------------------------------  # Date: Fri, 16 Jun 2000 21:11:01 GMTo/ From: "Richard L. Dyson" <rick-dyson@uiowa.edu>c, Subject: Netscape v3.03 and Time Calculation) Message-ID: <394A5195.6FB844CA@uiowa.edu>   G I have noticed a problem with my OpenVMS Netscape (v3.03-111897) that IoC use on my OpenVMS Alpha systems.  If I open a URL and then view thecD page's HTML info (VIEW -> DOCUMENT INFO) it reports a "Last ModifiedG Time" that is not correct.  That is, the OpenVMS time for the file thatHG is being served is one value and the value reported by OpenVMS/Netscape-A is equal to the time zone offset for my site (+5 hours).  The GMT-/ converted time is also five (5) hours too high.y  A NOTE:  All other Netscape clients on PCs and *ix boxes report theH correctaE time!  I don't believe it is a problem with my HTTP server (OSU HTTP,H v3.8a5).  D Has anyone else noticed this or can they confirm it on other OpenVMSH systems?  Is there some Time Zone logical I have messed up that Netscape uses?s  H This is not just a trivial problem since it means that pages that shouldG expire from the browser's cache in "a few minutes" now don't expire for  5 hours + "a few minutes". :(2   Regards, Rick --  H Richard L. Dyson                                    rick-dyson@uiowa.eduH  _   _      _____                http://www-pi.physics.uiowa.edu/~dyson/H | | | |    |_   _|   Systems Analyst                     O: 319/335-1879H | | | | of   | |     The University of Iowa            FAX: 319/335-17536 | \_/ |     _| |_    Department of Physics & Astronomy-  \___/     |_____|   Iowa City, IA 52242-1479    ------------------------------   Date: 16 Jun 2000 18:17:45 GMT0 From: sander@vmsbiz.enet.dec.com (Warren Sander)7 Subject: RE: OpenVMS clusters vs other systems clustersu* Message-ID: <8idr09$mv2@usenet.pa.dec.com>  0 We have a new white paper from Techwise Research  & 	Quantifying the Value of Availability; A detailed Comparision of Four Different RISC-Based Cluster 3     Solutions Designed to Provide High Availability   L http://www.openvms.compaq.com/openvms/whitepapers/techwise-availability.html    	In word doc, pdf and ps formats   -warren.     -- oB ------------------------------------------------------------------6 Warren Sander                        OpenVMS MarketingD Compaq Computer Corporation          Work:  warren.sander@compaq.comB 200 Forest Street MR01-3/J1          Personal: sander@ultranet.com3 Marlboro, MA 01752                   (508) 467-4875r6    My opinions are my own and I only speak for myself /           Read http://www.openvms.digital.com/ oB ------------------------------------------------------------------   ------------------------------  % Date: Fri, 16 Jun 2000 21:30:33 -0400R' From: "Bill Todd" <billtodd@foo.mv.com>C7 Subject: Re: OpenVMS clusters vs other systems clusters@( Message-ID: <8iek4j$71v$1@pyrite.mv.net>  I Just a 'me, too' comment.  I loved the book because it was simultaneouslyoK readable, detailed, and highly accurate (and helped me catch up on a lot ofMI hardware detail that I've just kind of informally assimilated over time -VF more or less).  The second edition was significantly expanded from theL first, and I hope there'll be a third edition by 2002 (the 1998 contents areL still very valuable, but 4 years sees a great deal of advance in this area).   - bill  6 Richard D. Piccard <piccard@ohio.edu> wrote in message" news:394A5609.75408BCB@ohio.edu... >I > I was very impressed byO >-F > "In Search of Clusters, Second Edition:  The Ongoing Battle in LowlyI > Parallel Computing" by Gregory F. Pfister, (C) 1998, Prentice Hall PTR, 3 > Upper Saddle River, NJ 07458, ISBN 0-13-899709-8., >MJ > The writing style is very readable; he appears to have done his homeworkF > well on VMS clustering -- there were a couple of things I would have6 > shaded differently, but nothing significantly wrong. >  > RDPo >x > Francesco Gennai wrote:C > >N< > > Where I could find information and possibly a comparison= > > between OpenVMS clustering and other systems clustering ?T? > > (Eg.: sharing nothing clusters vs OpenVMS clusters, etc...)(I > > I'm interested in URLs, books, white papers, personal opinions and so$ on.... > >I
 > > FrancescoI >               -----------F > > Francesco Gennai                     Internet : francesco.gennai @
 iat.cnr.it > >r http://mail.iat.cnr.it/gennaioC > > IAT - CNR                            Phone    : +39-050-3152592oC > > Area di Ricerca di Pisa              Fax      : +39-050-3152593o > > Via Alfieri, 1 > > 56010 Ghezzano PISA   ITALY  >1              ----------  >  > --D > ==================================================================D > Dick Piccard                           Academic Technology ManagerD > piccard@ohio.edu                                 Computer ServicesD > http://oak.cats.ohiou.edu/~piccard/                Ohio University   ------------------------------  % Date: Fri, 16 Jun 2000 23:14:32 -0400:- From: JF Mezei <jfmezei.spamnot@videotron.ca>07 Subject: Re: OpenVMS clusters vs other systems clusters , Message-ID: <394AED18.83133990@videotron.ca>  G Can NT clusters have disk shadowing across buildings ? If so, how is it 
 implemented ?e   ------------------------------  % Date: Fri, 16 Jun 2000 23:56:50 -0400 ' From: "Bill Todd" <billtodd@foo.mv.com>,7 Subject: Re: OpenVMS clusters vs other systems clusters ( Message-ID: <8iesmt$g08$1@pyrite.mv.net>  8 JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message& news:394AED18.83133990@videotron.ca...I > Can NT clusters have disk shadowing across buildings ? If so, how is it* > implemented ?*   ------------------------------  % Date: Sat, 17 Jun 2000 00:24:16 -0400r' From: "Bill Todd" <billtodd@foo.mv.com>o7 Subject: Re: OpenVMS clusters vs other systems clusters*( Message-ID: <8ieuaa$i1r$1@pyrite.mv.net>  8 JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message& news:394AED18.83133990@videotron.ca...I > Can NT clusters have disk shadowing across buildings ? If so, how is itt > implemented ?n  7 (Slip of the finger on the previous response.  Anyway:)t  G While it would seem silly not to have done so, I do not know for a fact,E that, if any enhancements were required to allow NT's SCSI drivers toeH support fail-over, these same enhancements have been added to NT's FibreG Channel drivers (I don't even know that FC drivers are supported on NT,** though I'm pretty sure they are on Win2K).  L But assuming NT (or Win2K) FC drivers equivalent to the SCSI drivers used toG support disk fail-over, the disks can nominally be up to 10 km from theSD host.  So you place a host (Win2K, if NT doesn't support FC) in eachI building (or in neighboring suburbs, up to 10 km apart) each connected tokK two pairs of dual-ported FC disks, with one disk in each pair local and theo@ other disk at the other site (and on separate FC links so that aI catastrophic connection failure won't take out both disks of a pair).  IneJ other words, exactly as you'd hook up a SCSI NT/Win2K cluster, except at a greater distance.n  L (If you don't need an active/active pair of hosts, you can just use a singleJ mirrored pair of disks and leave one host on stand-by to take them over ifH the other fails.  In either case both hosts may need an additional local system disk to boot from.)  H During normal operation, one host maintains the mirroring on one pair ofK disks (one local, one remote) and the other host maintains the mirroring oneJ the other pair (one local, one remote).  If one host fails, the other hostK acquires its mirrored pair, recovers the file system on that mirrored pair,tL and takes over whatever duties were configured to fail over (just as with anI NT SCSI-based cluster).  If a disk fails, its owner just continues, using K the remaining disk of the mirrored pair.  If a site disaster takes out bothhI a host and the two disks at that site, normal host fail-over to the othernJ host occurs, and the survivor continues processing its own workload (usingL the one remaining disk of its mirrored pair, connected by the local FC link)C and performs normal fail-over, file system recovery, and restart of F specified applications from the demolished host, using the one (local)K remaining disk of the failed host's mirrored pair (again, it's connected by( the local FC link).,   - bill   ------------------------------  % Date: Fri, 16 Jun 2000 16:23:50 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> J Subject: Re: OpenVMS commentaries (was Re: Gartner commentary on Wildfire), Message-ID: <394A8CD5.94292480@videotron.ca>   Bill Todd wrote: > J > Since a shared-disk (or even virtual-shared-disk) architecture is wholly- > unnecessary to achieve the clustering goals6    I Are there implementations of volume shadowing which allow drives to be in:B separate buildings, which do not do the "shared disk" philosophy ?  J Are there such implementations which allow a system to be shadowed accrossH buildings without the application having to be written to enable such an architecture ?  N If it is the application which does the job of keeping duplicate copies of itsK data across different file servers, can you call the OPERATING SYSTEM to beoJ disaster tolerant ? The application would be disaster tolerant despite theI operating system not being so, and you'd have to choose only applications-F capable of doing this. Any other software on the machine would not be.  N With VMS, it makes it very easy to have disaster tolerance for any applicationM on your system, even applications which can only run on one node of a clustern
 at a time.   ------------------------------  % Date: Fri, 16 Jun 2000 20:51:17 -0400-' From: "Bill Todd" <billtodd@foo.mv.com> J Subject: Re: OpenVMS commentaries (was Re: Gartner commentary on Wildfire)( Message-ID: <8iehqu$1il$1@pyrite.mv.net>  8 JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message& news:394A8CD5.94292480@videotron.ca... > Bill Todd wrote: > >lL > > Since a shared-disk (or even virtual-shared-disk) architecture is wholly/ > > unnecessary to achieve the clustering goals  >  >oK > Are there implementations of volume shadowing which allow drives to be inaD > separate buildings, which do not do the "shared disk" philosophy ?  K Absolutely (except it's 'mirroring', controlled by a single host at a time, A rather than 'volume shadowing' as practiced by VMS using multiple-A cooperating hosts - I'm not sure there's any direct equivalent to J cooperating HS'x's, though:  you may be limited to host-based mechanisms).4 Build your own using FC disks/arrays (or the similarL Ethernet-connected-storage mechanisms that are starting to appear), or buy aL storage box that handles its own remote mirroring (which might be consideredH similar to cooperating HS'x's - ISTR that ArcResearch and EMC offer suchK boxes, and likely there are others).  Where storage is concerned, 'private'  does not mean 'local'.   > L > Are there such implementations which allow a system to be shadowed accrossJ > buildings without the application having to be written to enable such an > architecture ?  J Not sure how you believe that this question differs from the previous one.L In a Tru64 or Sun cluster, multiple instances of an application on differentJ nodes can access common data in the same manners they can in a VMS clusterK (though in the Tru64 or Sun clusters the data is accessed - transparently - F through some file server rather than directly to the disk or through aC served disk), so there's no difference in the degree of applicationb% 'awareness' between the environments./   >cL > If it is the application which does the job of keeping duplicate copies of itse% > data across different file servers,i  I It's not:  the OS (or storage hardware) handles duplication, just as with?> VMS.  Perhaps that explains what you were trying to ask above.  (  can you call the OPERATING SYSTEM to beL > disaster tolerant ? The application would be disaster tolerant despite theK > operating system not being so, and you'd have to choose only applicationsrH > capable of doing this. Any other software on the machine would not be. >aD > With VMS, it makes it very easy to have disaster tolerance for any application G > on your system, even applications which can only run on one node of a  cluster0 > at a time.  F And it's equally easy in the non-shared-disk environments I described,L though my guess would be that since this is a recent occurrence, the vendors; might have less experience putting such a package together.a   - bill   ------------------------------  % Date: Fri, 16 Jun 2000 15:18:16 -0500d5 From: "Hipenbecker, Doug" <Hipenbecker.Doug@mbco.com>o7 Subject: Oracle 8.1.6 Availability on OpenVMS 7.1 Alpha E Message-ID: <DD11CB6FEB21D41184510004ACA37153163948@mbsus228.mbc.com>e  
 Greetings!  G Can anyone tell me the status of Oracle 8.1.6 on OpenVMS 7.1 Alpha?????a   Thanks   Douglas J Hipenbeckerb Information Management Servicest Miller Brewing Company
 Milwaukee, WIg dhipenbecker@mbco.comu 414-931-2971 Pager 1-888-579-6903  <<...OLE_Obj...>> t   ------------------------------  % Date: Fri, 16 Jun 2000 16:43:52 -0400p- From: "Mitchell, David R." <mitchell@WPI.EDU>d; Subject: RE: Oracle 8.1.6 Availability on OpenVMS 7.1 Alpha B Message-ID: <FE1835D68492D311BF7900508B5BEB0D0DE80B@petra.WPI.EDU>  J Our DBA has been waiting for this for a long time now.  It was supposed toH be released in April/May when we asked around the beginning of the year.K Then it was supposed to be out in June when we asked a few months ago.  I'm-J not sure what they're saying now that it's June (hmm... July anyone?), but! it should be out real soon now...-   David4   > -----Original Message-----< > From: Hipenbecker, Doug [mailto:Hipenbecker.Doug@MBCO.COM]% > Sent: Friday, June 16, 2000 4:18 PM> > To: Info-VAX@Mvb.Saic.Comu9 > Subject: Oracle 8.1.6 Availability on OpenVMS 7.1 Alpha  >  >  > Greetings! > ? > Can anyone tell me the status of Oracle 8.1.6 on OpenVMS 7.1 d > Alpha????? >  > Thanks >  > Douglas J Hipenbeckerf! > Information Management Servicesr > Miller Brewing Company > Milwaukee, WIn > dhipenbecker@mbco.coml > 414-931-2971 > Pager 1-888-579-6903 >  <<...OLE_Obj...>> b >  >  >    ------------------------------  # Date: Fri, 16 Jun 2000 18:51:34 GMTR1 From: "Mark D. Jilson" <jilly@clarityconnect.com>B# Subject: Re: patches, graphic cards 2 Message-ID: <394A771F.DB6836B3@clarityconnect.com>  < This is an S3 Trio32/64 PCI Graphics Adapter and it uses theH SYS$GQADRIVER.EXE image.  For OpenVMS V7.1 there are no patches for this graphics adapter driver.   Phillip Helbig wrote:b > E > In article <3948E36C.63CEF181@clarityconnect.com>, "Mark D. Jilson"l$ > <jilly@clarityconnect.com> writes: >  > >$ ANA/SYSTEMl > >CLUE CONFIG > > 8 > >See if SDA lists a Gnn: device and what SDA calls it. >  > No Gnn:  Closest is: > " > 6000   14 GQA:   13 S3 Trio32/64 >  > What's that?   -- )D Jilly	- Working from Home in the Chemung River Valley - Lockwood, NY0 	- jilly@clarityconnect.com			- Brett Bodine fan. 	- Mark.Jilson@Compaq.com			- since 1975 or so, 	- http://www.jilly.baka.com               -   ------------------------------  % Date: Fri, 16 Jun 2000 21:06:38 +0200r= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com> < Subject: Re: Request information about GNU C for VMS (Alpha)) Message-ID: <394A7ABE.8D6EE322@gtech.com>i  
 Jon wrote:? >    We're looking for GNU C for VMS or a similar product.  Anyi# > pointers would be great.  Thanks!s  1 Go to http://www.levitte.org/~ava/vms_gnu.htmlx !_   Arne   ------------------------------  % Date: Fri, 16 Jun 2000 16:53:20 +0200 = From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>e1 Subject: Re: Searching for  " in files using dcl.O) Message-ID: <394A3F60.47CE0920@gtech.com>s   "Waard, D.G.A. de" wrote:<F > How can I locate the position of " (double quote)using dcl lexicals.+ > Im trying to do this using the following:h > / > a = f$locate(""",line) ; but this won't work.c   You need f$locate("""", !i   The rule is:   - double all quotes-   - put quotes around-   so " -> "" -> """"   Arne   ------------------------------  % Date: Fri, 16 Jun 2000 16:30:29 -0400A  From: kuff@tessco.com (Hal Kuff)D Subject: Storage Works / Snapshots / Maybe it's time to skip OpenVMSO Message-ID: <AD1D026356980F99.DCBA09BFEBD043F6.617FC9F13925FF9E@lp.airnews.net>m  7    IS there any kind of snapshot supported by OpenVMS? cE    We'd sure like to buy about $200,000 worth of new Storage, but theuH ability to do snapshots from OpenVMS is not negotiable. This needs to be2 scripted and run several times per day unattended.  A    Why this is available on NT now with 1,000 times the installede4 StorageWorks base on OpenVMS starving is beyond.....   ------------------------------  # Date: Fri, 16 Jun 2000 23:26:20 GMTa9 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen) H Subject: Re: Storage Works / Snapshots / Maybe it's time to skip OpenVMS+ Message-ID: <+bFQPWuk$s8t@eisner.decus.org>r  r In article <AD1D026356980F99.DCBA09BFEBD043F6.617FC9F13925FF9E@lp.airnews.net>, kuff@tessco.com (Hal Kuff) writes:  9 >    IS there any kind of snapshot supported by OpenVMS? a  > Since you go on to talk about disks, I will presume you mean a@ snapshot aimed at disk backup, not the older VMS "snapshot" used for booting more quickly.e  ? Check your archives of this newsgroup for a recent announcementn; of the forthcoming "SDK" for the next version of VMS.  Thatt= SDK should contain everything coming to VMS in the forseeable  future.   C >    Why this is available on NT now with 1,000 times the installed:6 > StorageWorks base on OpenVMS starving is beyond.....  1 NT has no requirement to run in a (real) cluster.    ------------------------------   Date: 16 Jun 2000 23:14:58 GMT2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)- Subject: Unix to OpenVMS portability concerns2, Message-ID: <8iecdi$ib7@gap.cco.caltech.edu>  G I finally got around to writing down some information for people new ton' doing Unix to OpenVMS ports.  It's at: h  :    http://seqaxp.bio.caltech.edu/www/UNIX_TO_VMS_NOTES.TXT  J Please email me any additions, corrections, clarifications which may be in	 order.   a   Regards,   David Mathog mathog@seqaxp.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech     ------------------------------  # Date: Fri, 16 Jun 2000 22:43:16 GMTt2 From: "Zane H. Healy" <healyzh@shell1.aracnet.com> Subject: Re: VAX on Intel?4 Message-ID: <84y25.73$6C1.12907@typhoon.aracnet.com>  5 In comp.os.vms Bill Todd <billtodd@foo.mv.com> wrote:qI > So that's one reason that an 'affordable' VMS in the Windows (let aloneSH > Linux) cost range might be hard to achieve (it works with the hobbyistJ > license precisely because such people don't need support, but that's not  > true of commercial customers).  E Perhaps it would make sense to sell varying levels of support.  I'm a-B Hobbyist, BUT I'd be very happy to buy a full blown system withoutL restrictions if it was affordable.  I don't need support, with the exception of access to the patches.   K Still, I've got to wonder if there would be enough demand to make something'G like this work Compaq's time.  Let's be realistic.  How many people arebG there activelly using OpenVMS at home?  A few thousand I'd guess.  MostaC people want more, and understandably.  I was working on my main VMS K Workstation at home last night and had to switch over to my Mac in order toi- be able to run a chat client I needed to run.2  I > Recently, it's been easy to pick up inexpensive VAXen with licensed VMSlI > systems on them for a few hundred dollars (comparable to an entry-leveluL > Win98 home system).  This suggests strongly how much demand there would beN > for very-low-end VMS systems if Compaq went to the trouble of creating them: > just about none.  L Maybe the people wanting VMS systems at home don't want an inexpensive VAX. J My main system is a PWS433au w/320MB RAM, 24-bit colour, and UW-SCSI.  I'dL rather not use a VAX for anything other than a server, and my main server isK a AlphaStation 200 4/233.  You should consider the availability of Alpha's,n- and the cost of systems such as the PWS433au.w  F > To me, this suggests that expansion of the VMS user base (which I doL > consider a desirable goal, and suspect Larry does as well) is not going toL > come via the route you're advocating - at least not in the present climateJ > and with VMS's present (lack of) industry visibility and acceptance.  WeK > should be thankful that it's possible to purchase a new (DS10) VMS systemII > for $5000 - $6500 (depending on options), which puts it well within theaF > range of any small business that needs VMS, and look to other marketH > segments (all of which offer greater margins and profit potential) for > expansion opportunities.  L How much better would a system like this without any support do if it was inI the $2-3000 range?  That would put it in the range of a good Windows PC.  K Personally I don't expect *any* support on a Windows PC.  Sell them at that D price range, and have them flagged as being ineligible for a supportJ contract.  I'd love a DS10, but unless there is a significant price drop IJ won't be able to think about affording them until they start showing up in' significant numbers on the used market.   D How many business would find it practical to buy a system that isn'tH eligible for a maintenance contract?  Of course such a restriction would: probably just result in some 3rd party offering just that.   			Zaneu   ------------------------------  % Date: Fri, 16 Jun 2000 19:51:49 -0400R* From: David A Froble <davef@tsoft-inc.com> Subject: Re: VAX on Intel?- Message-ID: <394ABD95.AD4B70C5@tsoft-inc.com>i   "Zane H. Healy" wrote: > 7 > In comp.os.vms Bill Todd <billtodd@foo.mv.com> wrote: K > > So that's one reason that an 'affordable' VMS in the Windows (let alone J > > Linux) cost range might be hard to achieve (it works with the hobbyistL > > license precisely because such people don't need support, but that's not" > > true of commercial customers). > G > Perhaps it would make sense to sell varying levels of support.  I'm amD > Hobbyist, BUT I'd be very happy to buy a full blown system withoutN > restrictions if it was affordable.  I don't need support, with the exception > of access to the patches.  > M > Still, I've got to wonder if there would be enough demand to make somethingwI > like this work Compaq's time.  Let's be realistic.  How many people areiI > there activelly using OpenVMS at home?  A few thousand I'd guess.  MostsE > people want more, and understandably.  I was working on my main VMSdM > Workstation at home last night and had to switch over to my Mac in order to2/ > be able to run a chat client I needed to run.   N For personal VMS usage, there must be personal productivity applications.  NotN really there now.  Definitely won't be until there is a market for them.  sort of a catch-22 situation.  K > > Recently, it's been easy to pick up inexpensive VAXen with licensed VMS K > > systems on them for a few hundred dollars (comparable to an entry-leveltN > > Win98 home system).  This suggests strongly how much demand there would beP > > for very-low-end VMS systems if Compaq went to the trouble of creating them: > > just about none.  J For the decent systems, that don't require 3-phase power, there is still aO decent demand.  VAXstation 3100s and older are more suitable as museum pieces. eJ Before making statements about this, I'd like to see the numbers for hobby  licenses.  Might be interesting.  M > Maybe the people wanting VMS systems at home don't want an inexpensive VAX.oL > My main system is a PWS433au w/320MB RAM, 24-bit colour, and UW-SCSI.  I'dN > rather not use a VAX for anything other than a server, and my main server isM > a AlphaStation 200 4/233.  You should consider the availability of Alpha's,r/ > and the cost of systems such as the PWS433au.s  O Now, there's something I think the MS crowd doesn't do as much.  Servers in thesN home.  Hey, could be wrong.  With wintel, a server is usually for file storageO and centralizing printers.  On VMS, the server can be used for all activities. oA Of course, there's the previously mentioned lack of applications.u  H > > To me, this suggests that expansion of the VMS user base (which I doN > > consider a desirable goal, and suspect Larry does as well) is not going toN > > come via the route you're advocating - at least not in the present climateL > > and with VMS's present (lack of) industry visibility and acceptance.  WeM > > should be thankful that it's possible to purchase a new (DS10) VMS system0K > > for $5000 - $6500 (depending on options), which puts it well within the:H > > range of any small business that needs VMS, and look to other marketJ > > segments (all of which offer greater margins and profit potential) for > > expansion opportunities. > N > How much better would a system like this without any support do if it was inJ > the $2-3000 range?  That would put it in the range of a good Windows PC.  M Actually, I'd bet you could get a DS10 for $3000.  Street prices can be lowerhK than retail, and retail is only about $3,500 for the hardware.  VMS on thisu1 system shouldn't cost the same as VMS on an ES40.c  M > Personally I don't expect *any* support on a Windows PC.  Sell them at thatpF > price range, and have them flagged as being ineligible for a support > contract.S  H Oh, eligible is Ok, but not for free for 3 years the same as a much more expensive system.t  A   I'd love a DS10, but unless there is a significant price drop I L > won't be able to think about affording them until they start showing up in) > significant numbers on the used market.r  N Well, how about a currently fictional DS02?  PC type price.  PC type service. O PC type warranty.  Nothing Compaq will make big money on, but how much money doeM you derive solely from spending money on a page in the WSJ?  Not the businessuP generated, just for spending the money.  There's more than one method to use forL advertising, and there's more than promotional benefit in having hundreds ofO thousands of VMS systems in places where people will see them.  Use Monitor for O the screen saver, and the up-time will be readily visable. Of course, this will P not work with those who insist on powering off their computers when not in use.  Can't win 'em all.  F > How many business would find it practical to buy a system that isn'tJ > eligible for a maintenance contract?  Of course such a restriction would< > probably just result in some 3rd party offering just that. >  >                         ZaneN No such thing.  Compaq may decrease their warranty coverage, but for a serviceF organization to refuse to sell a maintenance contract?  Bad statement.   Dave   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com6 T-Soft, Inc.  170 Grimplin Road  Vanderbilt, PA  15486   ------------------------------  % Date: Fri, 16 Jun 2000 23:09:08 -0500 7 From: "David J. Dachtera" <djesys.nospam@earthlink.net>  Subject: Re: VAX on Intel?- Message-ID: <394AF9E4.D0F7BDDC@earthlink.net>a   Larry Kilgallen wrote: > i > In article <39498CB2.6C9BD0D1@earthlink.net>, "David J. Dachtera" <djesys.nospam@earthlink.net> writes:s > H > > For example: how would YOU go about expanding the OpenVMS user base? > >nF > > ...or is it that you are not convinced that it SHOULD be expanded? > C > As with Macintosh, I am convinced that one of the major strengthseB > of VMS is tight integration with the hardware and the ability of* > the OS group to influence the hardware.   ; This, however, increases hardware costs and keeps OpenVMS'slF cost-effectiveness in the unfavorable range. Note that most MACs todayH use common VGA-type monitors, "standard" SCSI devices, and focus less onC proprietary network protocols and media and more on UTP and TCP/IP.d   > I believe that throwings+ > that away is to lessen the value of VMS, y  D Obviously, I disagree. In the first place, no one is "throwing away"C anything. What this would actually do is to put OpenVMS on a "levelsF playing field" with other o.s.-es which are more tolerant of "diverse"H hardware. Strange that diversity among the work-force is viewed as good,F while diversity in the technical world is viewed as bad. Being able toG sucessfully use more third-party hardware with OpenVMS systems can only F help expand the user/installed base, while at the same time preserving@ the level of quality we've all come to expect. That is, expect aH high-quality device, but do not reject something which "plays nice" mostE of the time, and let the user be responsible for the results of using- lower quality gear.-   > and I am not interested0@ > in approaches that widen the interest in VMS at the expense of > decreasing the quality..  H In what way would increasing the user/installed base in any way threaten quality?  D > Widening the VMS user base is a great goal, but should not be done( > at the expense of existing strengths.    Same question again...   > Personally, I would suggesttC > they go for a "quality" approach, but that might be attributed to.; > my fantasy that the world will give up on trash products.t  C Agreed. However, quality can no more be "legislated" than morality.0  pD > Of course for Marketing the appropriate characteristic is "in yourA > face", and "remember the brand", so that means I do not want myDE > all-time favorite commercial spokesman (Stan Freberg) pitching VMS.r  F Actually, marketing can simply mean sufficient repetition to embed the message.  2 Here's a classic example - complete this sentence:  8 "Winston taste good like a ______________  ___________."  E That ad has been off the air for almost thirty years; yet, I'll bet abG good portion of the US folks on the newsgroup were able to complete theg! sentence almost without thinking.m   -- i David J. Dachtera. dba DJE Systemsd" http://home.earthlink.net/~djesys/  : Unofficial Affordable OpenVMS Home Page and Message Board:+ http://home.earthlink.net/~djesys/vms/soho/    ------------------------------  # Date: Sat, 17 Jun 2000 01:24:34 GMT?+ From: Jonas Lindholm <jlindholm@nyc.rr.com>e Subject: Re: Vest problemt* Message-ID: <394AD30A.455074C7@nyc.rr.com>   Larry Kilgallen wrote:   > In article <1137A4A23A51D311B2D600105A1D5213019AEEA5@seantexch.unitedad.com>, Terry Marosites <TMarosites@unitedad.com> writes:, > J > >        I am migrating from a VAX to a Alpha and I am trying to convertO > > some .EXE files ,( that I don't have source code for ) using vest . Some ofcN > > the programs fails the convert with a remark about the VMS version 1.1 . IK > > find it hard to believe the programs are that old. But in any case theyaK > > won't convert using vest, is there a min version that vest and convert?l >t? > VEST was specifically built such that it cannot deal with thei? > older image header format that was used prior to VMS V4.0. Ati? > a US DECUS Symposium in 1990 or 1991 Dick Sites and other DECiA > folks asked a bunch of customers whether that was an acceptable-< > tradeoff, and we all said yes.  (They asked a lot of otherA > questions also, related to the not-yet-announced Alpha system.)r   Hmm, I was able to vest an image that was compiled and linked on VAX/VMS 3.7 and the program run perfect on the Alpha, 6.1. I did, this 94/95.2   /Jonas Lindholmm   ------------------------------  % Date: Fri, 16 Jun 2000 19:15:39 -0500q, From: "Glenn C. Everhart" <Everhart@GCE.com>/ Subject: Re: VMS File Caching Futures and so on ' Message-ID: <394A7CDB.16344BC4@GCE.com>V   Bill Todd wrote: > 7 > Glenn C. Everhart <Everhart@GCE.com> wrote in messagep# > news:3946A50F.5D5BEAFD@GCE.com...  > > A bit of summary.-A > > 1. VMS caching has been a problem for some time, and is beingm? > > addressed. The forthcoming 7.3 fieldtest will see the firstcC > > public light for the new caching system. There are still things75 > > that can improve but it should be worth the wait.  > > A > > 2. On the Veritas homepage are some whitepapers detailing they@ > > unix caching scheme (and their replacement scheme). Note how= > > in both, user data is not written to disk till long afteruB > > the program is done; there is no mention of even ensuring that/ > > data is written before closing the program,- > L > While I haven't checked to make sure it was still there for Glenn to find,N > there used to a page indicating that whether files were flushed on close wasH > a mount-time option.  So the rest of his comments don't apply - unlessL > that's the behavior desired when the file system was mounted (which is, ofL > course, the expected behavior in the Unix environment, so really shouldn'tM > surprise anyone:  people there understand the need to flush files at pointsd- > where guaranteed persistence is important).T  : The Veritas system is of course another third party system= that has the problem of being in bed with the elephant, as it < has been called. Eventually the elephant rolls over; in this@ case Sun can at any time supply some enhancements to the Solaris= filesystems and Veritas' third party filesystem gets crushed.P; Presuming that won't happen due to sheer cussedness, it canB< still happen that some changes in Solaris, perhaps connected= with security, could force filesystem changes and cause Sun'so: people to decide they don't want their system's structures decided by third parties.  >  >  so in principle> > > a program written for the default situation in unix cannot? > > know that its data was written correctly. I have seen casesmA > > where treating user data this way, for an app written without D > > being laced with explicit flushes, can lose application metadata- > > and royally corrupt its data as a result.  > N > Such cases have a technical term in this industry:  they're called 'bugs' inL > the application, and have been known to occur even in the VMS environment.  A Not always. They also arise from user error. Write never so clearaC a manual and there will still be people doing things, non defaultedn things, which have dangers.a   >  >  Admittedly the VeritasoA > > f/s allows one to specify write-through; it appears not to be  > > the default. > N > No, it's not the default, any more than writing through RMS sequential filesM > is:  in both cases, the only way the user has to ensure that the data is on.J > disk is to flush it (or, on Unix, 'synch' the file) explicitly.  The oneJ > difference is that in the Unix environment dirty data can be shared withL > another program without necessarily forcing it to disk - behavior which isJ > well-defined in that environment and (as with all well-defined behavior)/ > causes problems only if used inappropriately.s >  > >d= > > 3. When writing to nla0:, every record results in a $qio.'A > > When writing to a ramdisk, RMS and language runtime librariesh? > > can be between your app and the disk. Ditto for hard disks,sB > > though default behavior can change; RMS has lots of variations< > > in its defaults depending on device characteristic bits. > M > Dear me:  that sounds like at least as much a recipe for user confusion and 0 > consequent disaster as the Unix behavior does. > B > > It is well to count I/O operations (show dev/full shows these)? > > to see whether every record is writing disk or every block. @ > > VMS underlying block I/O has generally been faster than unix@ > > block I/O in tests at low level, but a user app that invokesC > > multiple levels of additional interpretation slows things down.b> > > VMS engineering is I believe well aware of this. Still one> > > should not expect default VMS behavior to ever change to a$ > > setting that can lose user data. > M > Nor would any sensible person advocate such a change.  However, it's always H > possible to define new environments (or possibly file attributes) thatG > support new default behavior without affecting existing applications.o >  > - bill >  > >e> > > 4. The VMS compilers I've seen have had the most wonderfulB > > diagnostics I've encountered, pointing out errors which sailed@ > > by compilers from old RSX, from Microsoft, from Absoft, fromB > > Sun. While I must confess (in a joking way here) that it wouldA > > be refreshing for a C dialect to allow itself to be expressedY@ > > in Macro-11, allowing some of that good old code to migrate,? > > C and its descendants (both legitimate and Morganatic) havea= > > proven their worth. Much more C has been ported in nativeg0 > > format than has Macro-11 (or pal-11r or...). > > < > > 5. Now that QT is ported to VMS, anyone up for compiling > > the K office?  > >V > > Glenn Everhart   ------------------------------  % Date: Fri, 16 Jun 2000 21:21:08 -0400p' From: "Bill Todd" <billtodd@foo.mv.com>e/ Subject: Re: VMS File Caching Futures and so onh( Message-ID: <8iejj1$61d$1@pyrite.mv.net>  5 Glenn C. Everhart <Everhart@GCE.com> wrote in messaget! news:394A7CDB.16344BC4@GCE.com...e > Bill Todd wrote: > >o9 > > Glenn C. Everhart <Everhart@GCE.com> wrote in message % > > news:3946A50F.5D5BEAFD@GCE.com...p > > > A bit of summary. C > > > 1. VMS caching has been a problem for some time, and is being A > > > addressed. The forthcoming 7.3 fieldtest will see the firsteE > > > public light for the new caching system. There are still thingsa7 > > > that can improve but it should be worth the wait.  > > >fC > > > 2. On the Veritas homepage are some whitepapers detailing the B > > > unix caching scheme (and their replacement scheme). Note how? > > > in both, user data is not written to disk till long afteriD > > > the program is done; there is no mention of even ensuring that1 > > > data is written before closing the program,f > > H > > While I haven't checked to make sure it was still there for Glenn to find,0L > > there used to a page indicating that whether files were flushed on close wasnJ > > a mount-time option.  So the rest of his comments don't apply - unlessK > > that's the behavior desired when the file system was mounted (which is,' ofD > > course, the expected behavior in the Unix environment, so really	 shouldn'tgH > > surprise anyone:  people there understand the need to flush files at points/ > > where guaranteed persistence is important).t > < > The Veritas system is of course another third party system? > that has the problem of being in bed with the elephant, as it > > has been called. Eventually the elephant rolls over; in thisB > case Sun can at any time supply some enhancements to the Solaris? > filesystems and Veritas' third party filesystem gets crushed.f= > Presuming that won't happen due to sheer cussedness, it cano> > still happen that some changes in Solaris, perhaps connected? > with security, could force filesystem changes and cause Sun's < > people to decide they don't want their system's structures > decided by third parties.o  K I'm not sure whether to dignify the above by calling it FUD or call a spadeo# a spade and say it's pure bullshit.   E You might as well say it's equally dangerous to use VMS as a databasesK platform, because any such database must be purchased from some third partyeL whom Compaq may screw at any time by changing some documented VMS interface.  J Third-party file systems are common in Unix environments, and the internalD interfaces both above and below are well-defined and subject to onlyJ carefully-scrutinized changes (if changed at all, which is rare).  VeritasJ is almost certainly more important to Sun than Oracle is to VMS, so figure? out what the chances of the described problem are likely to be.   L The consequences to Sun from its customers would be at least as catastrophicL as would, say, accompany a Compaq decision to drop indexed file support fromK RMS:  sure, not *everyone* uses indexed files, but a whole lot of customersS7 would just stop running and know exactly whom to blame.i   > >m > >  so in principle@ > > > a program written for the default situation in unix cannotA > > > know that its data was written correctly. I have seen casesiC > > > where treating user data this way, for an app written withoutnF > > > being laced with explicit flushes, can lose application metadata/ > > > and royally corrupt its data as a result.i > >sF > > Such cases have a technical term in this industry:  they're called	 'bugs' innA > > the application, and have been known to occur even in the VMSt environment. >uC > Not always. They also arise from user error. Write never so clearoE > a manual and there will still be people doing things, non defaulted  > things, which have dangers.s  K Wrong.  Protecting against user error is the joint responsibility of the OSaD and the application, and the buck stops at the latter (as long as OSK behavior is well defined and sufficient to give the application the controlaJ it needs).  Failure to include a $FLUSH operation in a VMS application canJ as easily allow a user to corrupt application-level meta-data under VMS asE failure to include an fsync in a Unix application can allow a user to L corrupt application-level meta-data there:  there may be more frequent needsL for fsyncs than $FLUSHes, but all that means is that a buggy application mayH luck out a higher percentage of the time under VMS - which is not what I; call 'reliability' (and is what I call a bug, as I stated).r   - bill   ------------------------------  % Date: Fri, 16 Jun 2000 13:54:29 -0400r" From: Dan Sugalski <dan@sidhe.org>" Subject: Re: VMS Security features: Message-ID: <4.3.2.7.0.20000616135344.00c66b70@24.8.96.48>  1 At 05:50 PM 6/16/00 +0200, Jan Vorbrueggen wrote:v  >briggs@eisner.decus.org writes: >rJ > > > I was under the impression that VMS set the fault-on-execute bit on  > pages  > > > allocated as data + > > On VAX, at least, there is no such bit.r >s* >I do not think there is on Alpha, either.  B Yup, there is. See the Alpha version of the I&DS for more details.   					Dan  I --------------------------------------"it's like this"-------------------y2 Dan Sugalski                          even samurai? dan@sidhe.org                         have teddy bears and evene;                                       teddy bears get drunke   ------------------------------   End of INFO-VAX 2000.336 ************************