1 INFO-VAX	Thu, 23 Aug 2001	Volume 2001 : Issue 468       Contents:- 164LX system and QLA1080 SCSI host controller 1 Re: 164LX system and QLA1080 SCSI host controller  Re: A Primrose Path... Re: BIG cluster problem $ Re: Compaq as a technology leader???  Re: DECWindows and Reflections X  Re: DECWindows and Reflections X  Re: DECWindows and Reflections X  Re: DECWindows and Reflections X Employee Screening Service Employee Screening Service" Get the file-id for an opened file& Re: Get the file-id for an opened file* Re: Re: Get the file-id for an opened file& Re: Get the file-id for an opened file2 Re: GTK and VMS: current status and future plans ?/ Re: How is cluster id generated= csid 00010001? ; I need VAX/VMS + _C_ + TCP/IP developer for a project in NJ ! Re: keyboard controls in Open VMS  linking with xlib  Re: linking with xlib  Re: linking with xlib $ LPR from VMS 7.2-1 to Win2000 Server( Re: LPR from VMS 7.2-1 to Win2000 Server( Re: LPR from VMS 7.2-1 to Win2000 Server( Re: LPR from VMS 7.2-1 to Win2000 Server MLU and VMS 7.3  Re: More Alpha rubbish in print  Re: Nits in Slides Re: Nits in Slides5 Re: Nuts-n-bolts in News (was: Re: Nits in Slides...) 5 Re: Nuts-n-bolts in News (was: Re: Nits in Slides...) 5 Re: Nuts-n-bolts in News (was: Re: Nits in Slides...) 5 Re: Nuts-n-bolts in News (was: Re: Nits in Slides...)  Re: PGP for VMS 6.2  Re: Printing Through A Firewall  Rob: Close but no beer Re: Rob: Close but no beer Re: Rob: Close but no beer RWMBX 3 RWMBX - What to do, What to do! Help me if you can. 7 Re: RWMBX - What to do, What to do! Help me if you can. 7 Re: RWMBX - What to do, What to do! Help me if you can. 7 Re: RWMBX - What to do, What to do! Help me if you can. 7 Re: RWMBX - What to do, What to do! Help me if you can.  Re: Some postive points I hope.  Re: Some postive points I hope.  Re: Some postive points I hope.  Re: Some postive points I hope.  Re: Some postive points I hope.  Re: Sqlsrv_manage  Re: Take it elsewhere  Re: TAR files in VMS Unexpected DIAGPACK error  Re: Unexpected DIAGPACK error  Re: V5.5-2 Password Recovery Re: VAX: DEC-C float & double , Re: VMS high reliability needed by Air Force, Re: VMS high reliability needed by Air Force, Re: VMS high reliability needed by Air Force, Re: VMS high reliability needed by Air Force# Re: VMS/Alpha backend for GCC 3.x ?  Re: Wailing and Moaning.... 4 Re: Wailing at Eunuchs (was: Wailing and Moaning...)4 Re: Wailing at Eunuchs (was: Wailing and Moaning...)4 re: Wailing at Eunuchs (was: Wailing and Moaning...)4 Re: Wailing at Eunuchs (was: Wailing and Moaning...)4 Re: Wailing at Eunuchs (was: Wailing and Moaning...)4 Re: Wailing at Eunuchs (was: Wailing and Moaning...)4 Re: Wailing at Eunuchs (was: Wailing and Moaning...)4 Re: Wailing at Eunuchs (was: Wailing and Moaning...)4 Re: Wailing at Eunuchs (was: Wailing and Moaning...)4 Re: Wailing at Eunuchs (was: Wailing and Moaning...)  F ----------------------------------------------------------------------  % Date: Thu, 23 Aug 2001 14:18:25 +0100 $ From: Dave Dumolo <dave_d@cwcom.net>6 Subject: 164LX system and QLA1080 SCSI host controller) Message-ID: <3B8502A0.C964FF1E@cwcom.net>   C I've been using my 164LX based Alpha for about 18 months. It's been H running Debian Linux over a Tekram SCSI host controller. I recently cameH by a couple of Qlogic QLA1080 controllers - KZPCA in DEC speak, I think.  B Being curious I thought I would see if I could now load Alpha/VMS.F However, I seem to have fallen at the first hurdle - the box can't seeH the new SCSI host controller. As both controllers were in their unbrokenF antistatic sleeves and the vendor represented them as new and unused IG find it difficult to believe that both are DOA although I don't dismiss F the possibility. Does anyone know if these controllers can be detected% on a 164LX system (running V5.8 SRM)?    ------------------------------  % Date: Thu, 23 Aug 2001 08:14:58 -0500 1 From: Robert Alan Byer <byer@mail.ourservers.net> : Subject: Re: 164LX system and QLA1080 SCSI host controller3 Message-ID: <3B84BB82.7EAB6D20@mail.ourservers.net>    > E > I've been using my 164LX based Alpha for about 18 months. It's been J > running Debian Linux over a Tekram SCSI host controller. I recently cameJ > by a couple of Qlogic QLA1080 controllers - KZPCA in DEC speak, I think. > D > Being curious I thought I would see if I could now load Alpha/VMS.H > However, I seem to have fallen at the first hurdle - the box can't seeJ > the new SCSI host controller. As both controllers were in their unbrokenH > antistatic sleeves and the vendor represented them as new and unused II > find it difficult to believe that both are DOA although I don't dismiss H > the possibility. Does anyone know if these controllers can be detected' > on a 164LX system (running V5.8 SRM)?  >   F I'm in the same boat and good luck trying to make it work, I tried for, three weeks to get one working with no luck.   --    @  +------------------+--------------------------+---------------+@  | Robert Alan Byer | byer@mail.ourservers.net | ICQ #65926579 |@  +------------------+--------------------------+---------------+@  | Send an E-mail request to obtain a copy of my PGP key.      |@  +-------------------------------------------------------------+@  | "It is by caffeine alone I set my mind in motion.  It is by |@  |  cans of cola the thoughts aquire speed, the hands aquire   |@  |  shakes, the shakes become a warning.  It is by caffeine    |@  |  alone I set my mind in motion."                            |@  +-------------------------------------------------------------+   ------------------------------  % Date: Thu, 23 Aug 2001 07:12:18 -0600  From: yyyc186@mindspring.com Subject: Re: A Primrose Path... ; Message-ID: <3b850157$1$lllp186$mr2ice@nntp.mindspring.com>   3 In <3B57496D.3E86FFBE@videotron.ca>, on 07/19/2001  =    at 04:56 PM, JF Mezei <jfmezei.spamnot@videotron.ca> said:   R >> > -> No, the ONLY thing DEC EVER produced that EVER became an industry standard >> > -> was the VT-100.   J >Didn't DEC also design the 105 key keyboard ? (I know that the wintel and= >MAC world adopted a slightly different placement of the page % >up/down/cut/paste keys but still...)   ; Dec's COBOL compiler pretty much was the COBOL-85 standard.    --  ; -----------------------------------------------------------  yyyc186@mindspring.com; -----------------------------------------------------------    ------------------------------  # Date: Thu, 23 Aug 2001 10:58:37 GMT . From: Burnie M <burniem.NOSPAM@ozemail.com.au>  Subject: Re: BIG cluster problem8 Message-ID: <seo9ot03ia143pthvjpb6lehn2lls0c0ba@4ax.com>  C On 22 Aug 2001 09:27:25 -0700, vmendham@altavista.com (Vic Mendham)  wrote:  G >Using other posting on the group, I have managed to rename the SCSNODE D >and SCSSYSTEMID to what I think they should be. I have reloaded the >licenses as needed. > F >Now I need to set the disks from production $1$ to $2$ for the backup >system.6 >Is there a specific sysgen parameter which does this? > 	 >Regards,  >victor.mendham@emergis.com     	 ALLOCLASS   = You are going to find a few other small gotcha along the way. = Paying for a VMS Sys Man for a day would be a quicker option.    ------------------------------  # Date: Thu, 23 Aug 2001 16:39:56 GMT . From: "Duane Sand" <duane.sand@mindspring.com>- Subject: Re: Compaq as a technology leader??? < Message-ID: <wlah7.3640$sa.1757008@news1.rdc1.sfba.home.com>  
 > mulp wrote: L > > Ok, who thinks that Compaq was or is a leader in introducing technology?L > > What are some examples of these technologies since I can't think of any?   Here's something new !! 2       http://bbspot.com/News/2001/08/keyboard.html   ------------------------------  % Date: Thu, 23 Aug 2001 10:03:09 +0100 ( From: Nic Clews <sendspamhere@127.0.0.1>) Subject: Re: DECWindows and Reflections X ) Message-ID: <3B84C6CD.6A44EB9D@127.0.0.1>    Dave Pampreen wrote: > 2 > When I start a DECterm  I get the error message: >          Can't find > > font -*-terminal-Bold-R-Normal--18-140-100-100-c-0-ISO8859-* > N > Does anyone else get this?  If so how do I fix it?  When I installed Motif I3 > installed everything...Or is this on the PC side?  > J > I have Reflections X 7.2,  AlphaStation 250 4/266, VMS 7.2-1, DEC AXPVMS > DWMOTIF V1.2-5.   & You guess right, it is on the PC side.  E When reflections was installed, you also need to install the fonts it 9 requires, which vaguely match up with the contents of the E sys$sysdevice:[vms$common.sysfont.decw...] or at least those you use.   G This also apply to other third party X windowing software, e.g. Exceed.   A For X terminals, you start a font server process and the terminal G actually loads the fonts from the VMS host, but PC's don't generally do G this. I'm not aware of any way of serving or loading the fonts to a PC, 7 products such as Xcursion install the fonts by default.   F I guess a PC reinstall is in order and probably selecting the 'custom' option.  --  ( Regards, Nic Clews CSC Computer Sciences nclews at csc dot com    ------------------------------  % Date: Thu, 23 Aug 2001 16:58:54 +0200 & From: Michael Joosten <joost@c-lab.de>) Subject: Re: DECWindows and Reflections X $ Message-ID: <3B851A2E.500F@c-lab.de>   Nic Clews wrote: >    > C > For X terminals, you start a font server process and the terminal I > actually loads the fonts from the VMS host, but PC's don't generally do I > this. I'm not aware of any way of serving or loading the fonts to a PC, 9 > products such as Xcursion install the fonts by default.  >   G But most PC Xservers also have options to use a X fontserver, including ( Exceed and Reflection very likely, too.   F In some cases you really want to do this, Java/Swing applications come
 into my mind.   H > I guess a PC reinstall is in order and probably selecting the 'custom'	 > option.   H You probably mean the 'Xserver software', not just the whole PC, do you?F 8-) Sometimes the fonts  are already installed but not yet selected in the Font Configuration menues.  F Another opportunity is to define font aliases. Most DCE font names are= just aliases into the common font names. Have a look into the B documentation as also on the vendors website, there are often fontA updates for new releases of DCE on different platforms and so on.    --  * Michael Joosten, SBS C-LAB, joost@c-lab.de* Fuerstenallee 11, 33094 Paderborn, Germany, Phone: +49 5251 606127, Fax: +49 5251 6060658 C-LAB is a cooperation of University Paderborn & SIEMENS   ------------------------------  # Date: Thu, 23 Aug 2001 16:54:34 GMT 3 From: Tim Llewellyn <tim.llewellyn@cableinet.co.uk> ) Subject: Re: DECWindows and Reflections X / Message-ID: <3B85337D.FD664E91@cableinet.co.uk>    Michael Joosten wrote: >  > Nic Clews wrote: > >  >  > > E > > For X terminals, you start a font server process and the terminal K > > actually loads the fonts from the VMS host, but PC's don't generally do K > > this. I'm not aware of any way of serving or loading the fonts to a PC, ; > > products such as Xcursion install the fonts by default.  > >   H decw$xfs does (try to) serve fonts. I had this working with SCO XVision,? but didn't find it very reliable, the font server kept dropping 
 connections.  H Compaq services offered me a patch but it wasn't compatible with VMS 7.1 and A I wasn't able to upgrade, so I don't know if the patch fixed this  problem. The latest Motif didn't.   > I > But most PC Xservers also have options to use a X fontserver, including ) > Exceed and Reflection very likely, too.    agreed       --   Tim.Llewellyn@cableinet.co.uk     C Standard disclaimer applies. My views in no way represent those of  ! my employers or service provider.    ------------------------------  % Date: Thu, 23 Aug 2001 13:43:56 -0400 5 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com> ) Subject: Re: DECWindows and Reflections X 2 Message-ID: <ahbh7.673$bB1.31389@news.cpqcorp.net>  E There is a fix (dunno the version) to work around a compiler bug seen  loading specific fonts.     E Tim Llewellyn wrote in message <3B85337D.FD664E91@cableinet.co.uk>...  >  >  >Michael Joosten wrote:  >> >> Nic Clews wrote:  >> > >> >> >F >> > For X terminals, you start a font server process and the terminalL >> > actually loads the fonts from the VMS host, but PC's don't generally doL >> > this. I'm not aware of any way of serving or loading the fonts to a PC,< >> > products such as Xcursion install the fonts by default. >> > > I >decw$xfs does (try to) serve fonts. I had this working with SCO XVision, @ >but didn't find it very reliable, the font server kept dropping
 >connections. I >Compaq services offered me a patch but it wasn't compatible with VMS 7.1  >andB >I wasn't able to upgrade, so I don't know if the patch fixed this	 >problem.  >The latest Motif didn't.  >  >>J >> But most PC Xservers also have options to use a X fontserver, including* >> Exceed and Reflection very likely, too. >  >agreed  >  >  >--  >Tim.Llewellyn@cableinet.co.uk > C >Standard disclaimer applies. My views in no way represent those of " >my employers or service provider.   ------------------------------   Date: Thu, 23 Aug 2001 10:11:29 0 From: "Profile Strategies" <jobfitnow@orgio.net># Subject: Employee Screening Service 8 Message-ID: <iss.8b6.3b853711.a4cc8.1@mx2.west.saic.com>  2 We proudly represent Profiles International, Inc. . specializing in the world's finest employment + screening and employee evaluation products.   2 Which of the following employee related activities is your highest priority?    *Hire the Right People *Reduce Turnover *Improve Productivity  *Minimize Workplace Theft 0 *Identify People With Work Ethic and Reliability *Motivate Employees  *Increase Sales  *Make Managers More Effective  *Team Compatibility  *Customer Service   : Profile Strategies offers pre and post employment testing 3 and assessment tools for employee screening of job  8 candidates and managing current employees.  These tools 3 are used by managers and human resource personnel.    5 Pre employment screening include the measurement of   : job matching, aptitude, abilities, interest, personality, 6 sales skills, soft skills, integrity, work ethic, and 9 evaluation of competencies during the employee selection   process.  B Post employment assessments include team building, team analysis, ; customer service, succession planning, coaching, training,  : and 360 multi rater feedback performance reviews used for & management and leadership development.  + We are the experts!  Our clients love us.     & TAKE THE GUESS WORK OUT OF HIRING AND  MANAGING YOUR EMPLOYEES!  ; For more information, simply reply to this email or call us  at 1-805-238-0910.  Thank you.  > ==============================================================> This message was intended for any business owners and managers= that may be consindering to hire new employees.  If you feel  8 you've received this message in error or do not wish to B receive future communications from Profile Strategies via e-mail, ; please accept our sincere apologies. To discontinue future  = e-mails please reply back to this message and simply type in  4 the subject line UNSUBSCRIBE followed by the address; that you wish to have removed from our list.  We honor All   Remove requests.> ==============================================================        ------------------------------  % Date: Thu, 23 Aug 2001 10:38:43 +0001 2 From: Profile Strategies <jobfitnow@techemail.com># Subject: Employee Screening Service - Message-ID: <0GIJ008AV76J1P@mx.east.saic.com>   2 We proudly represent Profiles International, Inc. . specializing in the world's finest employment + screening and employee evaluation products.   2 Which of the following employee related activities is your highest priority?    *Hire the Right People *Reduce Turnover *Improve Productivity  *Minimize Workplace Theft 0 *Identify People With Work Ethic and Reliability *Motivate Employees  *Increase Sales  *Make Managers More Effective  *Team Compatibility  *Customer Service   : Profile Strategies offers pre and post employment testing 3 and assessment tools for employee screening of job  8 candidates and managing current employees.  These tools 3 are used by managers and human resource personnel.    5 Pre employment screening include the measurement of  l: job matching, aptitude, abilities, interest, personality, 6 sales skills, soft skills, integrity, work ethic, and 9 evaluation of competencies during the employee selection s process.  B Post employment assessments include team building, team analysis, ; customer service, succession planning, coaching, training, e: and 360 multi rater feedback performance reviews used for & management and leadership development.  + We are the experts!  Our clients love us.     & TAKE THE GUESS WORK OUT OF HIRING AND  MANAGING YOUR EMPLOYEES!  ; For more information, simply reply to this email or call us  at 1-805-238-0910.  Thank you.  > ==============================================================> This message was intended for any business owners and managers= that may be consindering to hire new employees.  If you feel 28 you've received this message in error or do not wish to B receive future communications from Profile Strategies via e-mail, ; please accept our sincere apologies. To discontinue future i= e-mails please reply back to this message and simply type in  4 the subject line UNSUBSCRIBE followed by the address; that you wish to have removed from our list.  We honor All e Remove requests.> ==============================================================        ------------------------------  + Date: Thu, 23 Aug 2001 06:41:22 +0000 (UTC):- From: "Liss Ake" <Ake.Liss@avestapolarit.com> + Subject: Get the file-id for an opened file H Message-ID: <DE53479E81510942887FDB07FD2290EB4E52F7@av0es001.cssdom.com>   Hi,   E How can I get the file-id for a file that is "locked by another user"c from a FORTRAN program?   D I have a program that uses fortran-OPEN with USEROPEN-param. In this: useropen-function I connect a NAM-block to FAB and call=20C SYS$OPEN. Then I can access nam$w_fid fields to obtain the file-id.oD This works OK for closed files, but I'm interested in a section-fileH connected to a global section created with $crmpsc.  The problem is that% sys$open gives me the RMS$_FLK error.&  E I know (of course) that I could do a $dir/file and parse that output.a' That is not the answer I'm looking for.r   Best regards	 -Ake Lissr BlueMaster Datar     -- y Posted from [193.15.230.232] l1 via Mailgate.ORG Server - http://www.Mailgate.ORG.   ------------------------------  % Date: Thu, 23 Aug 2001 20:05:10 +0010E% From: paddy.o'brien@zzz.tg.nsw.gov.aue/ Subject: Re: Get the file-id for an opened file)5 Message-ID: <01K7HHTHS8TE004AM6@tgmail.tg.nsw.gov.au>   F >How can I get the file-id for a file that is "locked by another user" >from a FORTRAN program?  1 In each program on the OPEN statement use SHARED.i   Regards, Paddy   ------------------------------  + Date: Thu, 23 Aug 2001 11:33:24 +0000 (UTC)a- From: "Liss Ake" <Ake.Liss@avestapolarit.com>i3 Subject: Re: Re: Get the file-id for an opened file-H Message-ID: <DE53479E81510942887FDB07FD2290EB4E565B@av0es001.cssdom.com>   Hmm,  E I would like to check any file, no matter if they are open, closed or F open/shared. If I can do it with DCL dir/file then I should be able toG do it from a program.  Remember I can not control how files are opened.   4 What I really want is the opposit to lib$fid_to_name   Best regards	 -Ake Lisso BlueMaster Datao     -- Z Posted from [193.15.230.232] e1 via Mailgate.ORG Server - http://www.Mailgate.ORGa   ------------------------------  % Date: Thu, 23 Aug 2001 09:14:11 -0400c% From: Bob Harris <harris@zk3.dec.com> / Subject: Re: Get the file-id for an opened filet3 Message-ID: <230820010914115603%harris@zk3.dec.com>n  / [[ This message was both posted and mailed: seea;    the "To," "Cc," and "Newsgroups" headers for details. ]]e  
 In articleF <DE53479E81510942887FDB07FD2290EB4E52F7@av0es001.cssdom.com>, Liss Ake# <Ake.Liss@avestapolarit.com> wrote:e  H X  How can I get the file-id for a file that is "locked by another user" X  from a FORTRAN program? X  sG X  I have a program that uses fortran-OPEN with USEROPEN-param. In this.; X  useropen-function I connect a NAM-block to FAB and call aF X  SYS$OPEN. Then I can access nam$w_fid fields to obtain the file-id.G X  This works OK for closed files, but I'm interested in a section-filetK X  connected to a global section created with $crmpsc.  The problem is that ( X  sys$open gives me the RMS$_FLK error. X  IH X  I know (of course) that I could do a $dir/file and parse that output.* X  That is not the answer I'm looking for.  E It has been a long time since I wrote this, _AND_ I do not know if itaE will actually do what you want (plus I work on Tru64 UNIX these days,tC so I can not even test to see if it does anything close to what youcF want), and it is written in C, not Fortran (I used to use Fortran as aF system implementation language from '79-'85, but that was 16 years agoD and now I would be hard pressed to remember a lot of the details :-)  , The sys$search() may be all you really need.  2                                         Bob Harris  G /*----------------------------------------------------------------------G     XQP_IS_FILE_OPEN.C - this program is a Quick and Dirty hack to test|G         if the specified file is opened on the current node.  This willtA         not detect if the file is opened by a different node of asC         cluster.  The advantage of this approach is that it doesn't B         interfer with any applications that would like to open theD         file.  But it must be run on the same node of a cluster that?         the application trying to open the file is also running-2         on, otherwise it will not detect the open.                8E         In order to understand all that is going on with this it will5A         require carefull reading of the "VMS I/O User's Referencei9         Manual: Part I", Chapter 1, "ACP-QIO Interface". eG ---------------------------------------------------------------------*/r #include <rms.h> #include <stdio.h> #include <ssdef.h> #include <stsdef.h>' #include <iodef.h> #include <fibdef.h>( #include <atrdef.h>l #include <sbkdef.h>t   typedef unsigned char byte;n typedef unsigned short word; typedef unsigned long longword;   " #define BYTE(f,o) *(byte *)(f+(o))" #define WORD(f,o) *(word *)(f+(o))& #define LONG(f,o) *(longword *)(f+(o))   struct z_atr {     word size;     word type;     void *addr;  };         main( int ac, char **av )- {-     longword status;     char *file;      longword d_dev[2];     longword d_fib[2];     word iosb[4];      word chan;     struct fibdef fib;     struct FAB fab;t     struct NAM nam;a     struct z_atr atr[4];     struct sbkdef sbk;     char esa[NAM$C_MAXRSS];      G /*-------------------------------------------------------------------*/D       if ( ac != 2 )        usage();e       file = av[1];e /*   Find the file  */     fab = cc$rms_fab;      nam = cc$rms_nam;r          fab.fab$l_nam = &nam;p     fab.fab$l_fna = file;c!     fab.fab$b_fns = strlen(file);i     nam.nam$l_esa = esa;      nam.nam$b_ess = sizeof(esa);     %     status = sys$parse( &fab, 0, 0 );l$     if ( !(status & STS$M_SUCCESS) )     {s         printf( "%s\n", file );          lib$stop( status );-     }-&     status = sys$search( &fab, 0, 0 );$     if ( !(status & STS$M_SUCCESS) )     {n         printf( "%s\n", file );t         lib$stop( status );-     }a /*    Assign a channel to the device */      d_dev[0] = nam.nam$t_dvi[0];!     d_dev[1] = &nam.nam$t_dvi[1];e.     status = sys$assign( d_dev, &chan, 0, 0 );     if ( !(status & 1) )     {n         printf( "%s\n", file );s         lib$stop( status );p     }o /*C   Initialize the XQP File Information Block with the File ID of thee
   target file0 */#     memset( &fib, 0, sizeof(fib) );s     d_fib[0] = sizeof(fib);R     d_fib[1] = &fib;     ;     fib.fib$r_fid_overlay.fib$w_fid[0] = nam.nam$w_fid[0]; r;     fib.fib$r_fid_overlay.fib$w_fid[1] = nam.nam$w_fid[1]; s;     fib.fib$r_fid_overlay.fib$w_fid[2] = nam.nam$w_fid[2]; l /*@   Build an Attribute list to fetch the file's Record Attributes. */"     memset( atr, 0, sizeof(atr) );      atr[0].size = ATR$S_STATBLK;      atr[0].type = ATR$C_STATBLK;     atr[0].addr = &sbk;A     "     memset( sbk, 0, sizeof(sbk) ); /*D   Fetch the file statistics block for the local node of the cluster. */8     status = sys$qiow( 0, chan, IO$_ACCESS, iosb, 0, 0, F                                                 d_fib, 0, 0, 0, atr, 0 );     if ( (status & 1) )w         status = iosb[0];u     if ( !(status & 1) )     {t         printf( "%s\n", file );e         lib$stop( status );l     }s /* */     if ( sbk.sbk$w_acnt != 0 )     {c=         printf( "File %s is opened %d times on this node\n", l#             file, sbk.sbk$w_acnt );y     }v     else     {e=         printf( "File %s is NOT open on this node\n", file );y     }t /* */     sys$dassgn( chan );n /* */     exit(1); }t       usage()' {p, #define FP(_s) fprintf( stderr, "%s\n", _s )C  FP("Usage: XQP_IS_FILE_OPEN :== $disk:[dir]XQP_IS_FILE_OPEN.EXE"); (  FP("       XQP_IS_FILE_OPEN filename");  FP("");A  FP("       filename - specifies the file to test to see if it is  open");g;  FP("                  on the current node of a cluster.");   FP("");  exit( SS$_BADPARAM ); }e   ------------------------------  % Date: Thu, 23 Aug 2001 08:12:53 -0500t1 From: Robert Alan Byer <byer@mail.ourservers.net>r; Subject: Re: GTK and VMS: current status and future plans ?d3 Message-ID: <3B84BB05.6F661423@mail.ourservers.net>    > , > What is the current status of GTK on VMS ? > L > I know that there is a port of 1.2-8; has anybody formed an opinion on how > stable it is ? > 4 > Is anyone planning to port later versions to VMS ? >  > Thanks for any information,o >   E It's used in the OpenVMS port of FreeCiv and works just fine, haven't  had a problem with it.   -- e  @  +------------------+--------------------------+---------------+@  | Robert Alan Byer | byer@mail.ourservers.net | ICQ #65926579 |@  +------------------+--------------------------+---------------+@  | Send an E-mail request to obtain a copy of my PGP key.      |@  +-------------------------------------------------------------+@  | "It is by caffeine alone I set my mind in motion.  It is by |@  |  cans of cola the thoughts aquire speed, the hands aquire   |@  |  shakes, the shakes become a warning.  It is by caffeine    |@  |  alone I set my mind in motion."                            |@  +-------------------------------------------------------------+   ------------------------------  # Date: Thu, 23 Aug 2001 15:40:04 GMTR2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)8 Subject: Re: How is cluster id generated= csid 00010001?2 Message-ID: <ot9h7.655$bB1.30985@news.cpqcorp.net>  i In article <8b51ed8.0108221052.3be8ab00@posting.google.com>, vmendham@altavista.com (Vic Mendham) writes:u  D :Looking at an old boot log, the system usually booted with an id ofD :00010002 from the message Node "rebuilt_nodename" (csid 0010002) isF :now a VAXcluster member. I'm using an image of another (primary) node= :in the cluster and from boot logs it boots as csid 00100001.i :eE :While testing the rebuilt system, it is not connected to any networkeF :and boots with a VOTE of 5. The csid is 0010001, which is the same as :the production system.o :fC :Question: Is it the same as the production system, due to the factuF :that the image backup is from the production system? Or is it 0010001E :because currently, it is the only node in the cluster. When I put it.C :on the production network, should it normally go to 0010002, or is ? :there a file that needs to be changed to allow this to happen?   I   Um, you would not be attempting any "creative" cluster configurations,     would you?  I   All nodes sharing a CI, DSSI or MC communications interconnect must be i   members of the same cluster.  G   On cluster interconnects managed by the Port Emulator (PE) layer (eg:oF   the NI and FDDI interconnects), the cluster group number is the key J   setting -- this value is used to generate the network address that will ,   be used among members of the same cluster.  I   The Cluster System Identification (CSID) is a unique value intended to nJ   differentiate hosts within the same cluster.   Once allocated, the CSID J   value is only deallocated if the cluster member node exits and the laterL   rejoins the cluster, with the node identity determined by the SCSSYSTEMID K   and SCSNODE parameters -- a node that is rejoining the cluster is issued  J   a new (and unique) CSID value as the original CSID is deallocated.  LikeH   the PID, it takes quite some time to roll through all CSIDs, and thereE   will be no duplicate CSIDs issued together within the same cluster.i  E   Please see the OpenVMS FAQ for details of setting the VOTES and theXI   EXPECTED_VOTES system parameters, and particularly for some details of eJ   how you can corrupt your data when these parameters are set incorrectly.   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: 23 Aug 2001 10:28:57 -0700. From: dpotapenko@emplifi.com (Denis Potapenko)D Subject: I need VAX/VMS + _C_ + TCP/IP developer for a project in NJ= Message-ID: <a7aad5c8.0108230928.2b9b8210@posting.google.com>r   Hi   subj  B if you are interested - please send me your resume + contact phone numbers + salary/rate req'so     Regards,+ Denis Potapenko, Global Recruiting Divisiona  9 EMPLIFI (an iGATE Capital company) http://www.emplifi.comC  9 1000 Commerce Drive, Suite 500, Pittsburgh, PA 15275, USAp   e-mail : DPotapenko@emplifi.como- phone : 412-490-5511 / 888-330-5497 ext 55-11a   ------------------------------  # Date: Thu, 23 Aug 2001 15:53:21 GMTm2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)* Subject: Re: keyboard controls in Open VMS2 Message-ID: <RF9h7.657$bB1.31137@news.cpqcorp.net>  o In article <BFBEDDF2CFEDD411917400508BF3A6FF027C20@EXCHSVR.FSC.COM.FJ>, Inosa Qativi <Inosa@fsc.com.fj> writes:c  M :In digital VAX system, how do we disable F5 key and other F keys for 101-keym6 :and 104-key keyboards  in Open VMS operating system ?  =   Without some additional background, an answer is difficult.e  5   Which specific terminal device(s) or emulator(s)?  O  I   I'll also ask that you post the OpenVMS VAX version, but (unless it is UA   an unusually old version) that probably won't be a factor here.   I   In the specific case of the F1 through F5 keys, these can be processed =G   directly (and locally) by the terminal or terminal emulator, and (in oG   various cases) can or cannot be particularly disabled.  Please check oG   the setup screens for the specific terminal(s) or emulator(s) in use.v  J   The reference to a 101-key keyboard would likely be to the LK443, PCCAL,I   PCXAL, LK471-series or similar PC-layout keyboard, which was used by a tL   variety of devices -- but is not necessarily the best choice for use with N   native OpenVMS application expectations.  I'll assume the 104-key reference M   was intending to be to the 105-key LK201, LK250 and LK451-series or to the  L   108-key LK401 series and related keyboards.  (That said, the answer still F   depends on the terminal device(s) or terminal emulator(s) involved.)  I   Within OpenVMS, the DEFINE/KEY command is used, and the SMG and similarrJ   calls can also be used to process keypresses.  The default settings for I   keys such as F12 can also be controlled using the SET TERMINAL command lH   (and specifically the line-editing-related mechanisms).  An example ofJ   capturing a function key keypress via a $qio extended read is available 
   at the URL:t  -     http://www.openvms.compaq.com/freeware/ -v1       freeware50/srh_examples/qio_extended_read.cr  A   SMG and similar libraries can also be used for this, of course.a  N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  + Date: Thu, 23 Aug 2001 13:01:14 +0000 (UTC) - From: Asbjoern Djupdal <djupdal@stud.ntnu.no>r Subject: linking with xlib- Message-ID: <9m2uqq$raa$1@tyfon.itea.ntnu.no>    Hi.m  8 I am completely new to vms, so I have a simple question:   How do I link to xlib?  K I have made a little c-program using xlib, that compiles fine with cc, but tD I don't know how to run the linker, so that xlib gets linked with it# to produce an executable program.      Asbjrnf   ------------------------------    Date: 23 Aug 2001 08:59:27 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen)= Subject: Re: linking with xlib3 Message-ID: <WhYZ1YV6$Dr9@eisner.encompasserve.org>a  ] In article <9m2uqq$raa$1@tyfon.itea.ntnu.no>, Asbjoern Djupdal <djupdal@stud.ntnu.no> writes:a > Hi.r > : > I am completely new to vms, so I have a simple question: >  > How do I link to xlib? > M > I have made a little c-program using xlib, that compiles fine with cc, but cF > I don't know how to run the linker, so that xlib gets linked with it% > to produce an executable program.  h  B The shareable image name is DECW$XLIBSHR.  Link to it as you wouldH to any other shareable image.  The relevant section on the documentation
 CD-ROM is:  1 	/VMSDOC073/v73/4548/4548pro_001.html#obj_mod_secC   ------------------------------  + Date: Thu, 23 Aug 2001 14:58:51 +0000 (UTC)a- From: Asbjoern Djupdal <djupdal@stud.ntnu.no>7 Subject: Re: linking with xlib- Message-ID: <9m35nb$21m$1@tyfon.itea.ntnu.no>o  . Larry Kilgallen <Kilgallen@spamcop.net> wrote:  D > The shareable image name is DECW$XLIBSHR.  Link to it as you wouldJ > to any other shareable image.  The relevant section on the documentation > CD-ROM is:  3 > 	/VMSDOC073/v73/4548/4548pro_001.html#obj_mod_secd   Thanks.    Asbjrnl   ------------------------------  % Date: Thu, 23 Aug 2001 07:35:56 -0600t4 From: "Michael D. Ober" <mdo.@.wakeassoc.com.nospam>- Subject: LPR from VMS 7.2-1 to Win2000 Serverc2 Message-ID: <2F7h7.150$Xx3.137223@news.uswest.net>  H I need to send an LPR Print job from AlphaVMS 7.2-1, TCPIP v5.0A ECO2 toK Windows 2000 Server SP1.  VMS requires an IP and a port, but the W2K serverGL uses named services for it's LPD service.  The IP address for the W2K server	 is fixed.   H How do I use a name instead of a port number when I define the VMS print
 queue?  orI How can I force W2K Server to use a fixed port number for the LPD serviceuG for the printer?  Also, for the W2K people, is there anyway I can avoide8 using a NullSessionShare to print in this configuration?   -- Thanks, 
 Mike Ober.   ------------------------------  % Date: Thu, 23 Aug 2001 16:42:32 +02001< From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <noone@home.com>1 Subject: Re: LPR from VMS 7.2-1 to Win2000 Servern( Message-ID: <3B851658.91B742BC@home.com>  ? In NT4 threre was an optional "package" that you could install. 7 Called "TCP/IP Printing Services" or something similar.p@ It contained an LPD module. With that running you could just setF up the PRINTCAP file on you VMS box using remote "node" and "printer".   Jan-Erik Sderholm   ------------------------------  # Date: Thu, 23 Aug 2001 16:45:11 GMTa3 From: Tim Llewellyn <tim.llewellyn@cableinet.co.uk> 1 Subject: Re: LPR from VMS 7.2-1 to Win2000 Server / Message-ID: <3B853150.F83CD01F@cableinet.co.uk>    "Michael D. Ober" wrote: > J > I need to send an LPR Print job from AlphaVMS 7.2-1, TCPIP v5.0A ECO2 toM > Windows 2000 Server SP1.  VMS requires an IP and a port, but the W2K server N > uses named services for it's LPD service.  The IP address for the W2K server > is fixed.e > J > How do I use a name instead of a port number when I define the VMS print > queue?  orK > How can I force W2K Server to use a fixed port number for the LPD servicetI > for the printer?  Also, for the W2K people, is there anyway I can avoid : > using a NullSessionShare to print in this configuration?   $ mcr ucx$lprsetup  2 (or tcpip$lprsetup if you have TCP/IP Services V5)  E then add a remote queue using the remote machine name (rm) of the W2k  server.oH Print to that queue, lpd will create it for you on startup. If you don't likeD the lprsetup utility you can edit the printcap file directly but I'd stick_" with the utility if you're unsure.  H Don't see why ports are relevant here, unless you are playing games with lpd.   regards  --   Tim.Llewellyn@cableinet.co.uk     C Standard disclaimer applies. My views in no way represent those of (! my employers or service provider.s   ------------------------------  % Date: Thu, 23 Aug 2001 11:48:32 -0600a4 From: "Michael D. Ober" <mdo.@.wakeassoc.com.nospam>1 Subject: Re: LPR from VMS 7.2-1 to Win2000 Server 2 Message-ID: <Tlbh7.162$5d2.124863@news.uswest.net>   Thanks,s  I The ports aren't relevent, except that TELNETSYM requires port numbers toe3 print to IP printers and I had never used LPRSETUP.  --
 Mike Ober.  @ "Tim Llewellyn" <tim.llewellyn@cableinet.co.uk> wrote in message) news:3B853150.F83CD01F@cableinet.co.uk...s >Q >k > "Michael D. Ober" wrote: > > L > > I need to send an LPR Print job from AlphaVMS 7.2-1, TCPIP v5.0A ECO2 toH > > Windows 2000 Server SP1.  VMS requires an IP and a port, but the W2K serverI > > uses named services for it's LPD service.  The IP address for the W2Kc server
 > > is fixed.t > >nL > > How do I use a name instead of a port number when I define the VMS print > > queue?  orE > > How can I force W2K Server to use a fixed port number for the LPDt service K > > for the printer?  Also, for the W2K people, is there anyway I can avoids< > > using a NullSessionShare to print in this configuration? >  > $ mcr ucx$lprsetup >e4 > (or tcpip$lprsetup if you have TCP/IP Services V5) >aG > then add a remote queue using the remote machine name (rm) of the W2k-	 > server.cJ > Print to that queue, lpd will create it for you on startup. If you don't > likeF > the lprsetup utility you can edit the printcap file directly but I'd > sticky$ > with the utility if you're unsure. >nJ > Don't see why ports are relevant here, unless you are playing games with > lpd. > 	 > regardsD > -- > Tim.Llewellyn@cableinet.co.uko >(D > Standard disclaimer applies. My views in no way represent those of# > my employers or service provider.n >c   ------------------------------  % Date: Thu, 23 Aug 2001 15:29:22 +0000d= From: Thomas Schildknecht <thomas.schildknecht@aiub.unibe.ch>a Subject: MLU and VMS 7.3- Message-ID: <3B852152.F5160F81@aiub.unibe.ch>   G I was using the 'mlu' utility to select tapes from a TZ857 minilibrary. E This worked fine until I moved from VMS AXP V7.1 to VMS AXP V7.1! The- problem is the following:2  E The robotic device is accessed via a pseudo-device on SCSI sub-lun 1. - The device is created (e.g. for MKA100) with:   <    mcr sysman io connect mka101 /noadapt/driver=sys$gkdriver  A BUT: V7.3 does not anymore allow this because 'there is already a_1 different driver defined for the device type MK'!f  G Does anybody know a workaround? Any other utility to command the robot?    Thanks,e   Thomas    i -- pH ************************************************************************G Thomas Schildknecht            eMail: thomas.schildknecht@aiub.unibe.ch B Astronomical Institute         www:      http://www.aiub.unibe.ch/! University of Berne              0B Sidlerstrasse 5                Phone:        +41-31-631 85 94 (91)= CH-3012 Bern, Switzerland      Fax:          +41-31-631 38 69rH ************************************************************************   ------------------------------  % Date: Thu, 23 Aug 2001 18:42:47 +02000= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>o( Subject: Re: More Alpha rubbish in print) Message-ID: <3B853287.1EDF4731@gtech.com>m   Richard Tomkins wrote:J > The point I was stumbling towards was that 2 good word processors and 58K > lame word processors for a system in comparison to 2 good word processorseN > and 3 lame word processors for another system, does not make the system with4 > 60 word processors better than the one that has 5. > G > There is no need to have thousands of applications and programs for aeM > system. There is just a need to have good quality applications andsolutionssI > available, and in that department, across all those different Operating M > Systems, there were some good quality applications and solutions available.   ; Noone is questioning the quality of the available software.   ? We are questioning, whether the software people want to use are 
 available.  G You mentioned word-processors yourself. The 3 big one is MS Word, Lotus ? Word and Corel Wordperfect. Which are available for Alpha (in a  recent version !) ?   F When we talk about email the 3 big ones are: MS Exchange, Lotus Domino5 and Novell Groupwise. Which are available for Alpha ?    Arne   ------------------------------  % Date: Thu, 23 Aug 2001 09:27:58 -0400F* From: John Reagan <john.reagan@compaq.com> Subject: Re: Nits in Slides_) Message-ID: <3B8504DE.7020704@compaq.com>e   Larry Kilgallen wrote: > ; > Were those added for the sake of performance (after EV4), E > or were they there from the beginning for the sake of correctness ?  >   G There were there in EV4 and my recollection was for correctness.  I do  F remember that the need for them came up late in the process after the H VEST folks realized that they couldn't do something without having them.   John Reagan    ------------------------------    Date: 23 Aug 2001 09:01:57 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen)  Subject: Re: Nits in Slidess3 Message-ID: <mECjdktwjOOj@eisner.encompasserve.org>o  V In article <3B8504DE.7020704@compaq.com>, John Reagan <john.reagan@compaq.com> writes: > Larry Kilgallen wrote: >>  < >> Were those added for the sake of performance (after EV4),F >> or were they there from the beginning for the sake of correctness ? >> e > I > There were there in EV4 and my recollection was for correctness.  I do -H > remember that the need for them came up late in the process after the J > VEST folks realized that they couldn't do something without having them.  F Ok.  Andrew's original claim was that changes were introduced to AlphaD to make VESTed images perform better.  I am sure he was referring toC changes _after_ Alpha and Alpha VMS 1.0 were released to customers.n   ------------------------------  % Date: Thu, 23 Aug 2001 13:49:41 +0100-( From: Nic Clews <sendspamhere@127.0.0.1>> Subject: Re: Nuts-n-bolts in News (was: Re: Nits in Slides...)) Message-ID: <3B84FBE5.B6253678@127.0.0.1>    andrew harrison wrote: >> lF >>   This assumes we use the identical configuration and the identicalG >>   translation mechanisms, of course.  And having seen ample evidencecG >>   of the designs of various x86 applications -- timing loops, directbG >>   hardware access, etc -- I'm not sure I'd generalize based on that.a >> d >r7 >It does but any otimism you might want to inject into b5 >this discussion needs to be countered with the fact h5 >that Intel have tried as much as they can to improvei6 >the performance of the x86 emulation but that it has 9 >proved to be very hard to emulate x86 and other non ILP  # >binaries into an ILP architecture.e  $ Thats like saying that a performance3 car manufacturer is busy altering valve timings andu< tweaking inlet pressures, when you're fuelling it with dregs= of French wine and measuring actual performance against that.t  = The VMS guys are software engineers, I'd hardly describe whaty5 Micro$oft cobble together as 'engineered', would you?c  A >>   Compaq plans to have EV7 and systems based on EV7 available.  > = >Quite so people will be migrating off EV7 onto IA64 my pointe; >exactly and there is no current indication that IA64 will e >be faster.   = Quite a number of implementations we have don't even need the ? EV6 performance, in the majority of cases it won't be an issue. > EV7 may be too fast, and it won't be the first time I've had a' complaint that a systems is *too* fast.t  ? But to requote myself quoting from Hitchhikers, the performancee; issue is not with Deep Thought, but the computer that is ton come AFTER it.  8 To reiterate, these guys are software engineers. They'veA addressed Alpha [NUMA] performance issues in the operating systemt@ before now, I'm sure tweaking the kernel for IPF will just be in a days work for them.   ? What you should actually be worried about, is that these peoples/ may just turn out an operating system that willi: run on an USIII. It may run like a dog, but, it'll be VMS.  F Customers buy Solaris from you, the fact it runs on your processors is? more of a co-incidence than intent. Further proof of buying theOF operating system rather than the platform should be evidenced that the industry is not 'buying' linux.y  A VMS is currently undergoing a transformation that could level ther playing field.  = Someone here stated that Alpha (64bit) to Playstation 2 (128)>9 would be an upgrade.  Despite the IO limitations, and thee7 'commercial advantages' of a PS2 port of VMS, it is not-@ impossible. I expect the ODS2/5 implementation for a memory card= will be fine, finding/swapping memory cards would be a littlea
 inconvenient.c  C Andrew, I really fear your missing many of the points here (in many7B posts) trying to twist technicalities to resemble arguments. I can< only assume you have never been given the task of writing an: _operating_system_ in a high level language such as BASIC C _under_  a timeharing operating system. Ever played seriously with a stacks? Self modifying code?  A If you had, you'd know all this was possible. If you want a bit 0 B or 1, it's up to the engineer writing the code. If they want a bitD of storage, they just declare it, on chip, off chip, doesn't matter.  E As to the reality of performance, I seriously doubt that the majority-; of commercial applications are actually taking advantage of2> the chip features, they may as well just be serial processors.> When do application writers ever listen to software engineers?   In another post, you also said:t  7 >Both HP-UX and Tru64 will have much larger catalogues M2 >of natively compiled apps than OpenVMS so if you 6 >have to rely on emulation it had better be real fast.  6 What is there to stop VMS creating a native 'shell' in4 which to run these applications? All of a sudden the< balance shifts... A bit like running your MSDOS applications6 under an NT shell. Anything is possible when bit comes to push.  < I don't know if you really believe anything I have said, but7 the only obstacles are commercial realitites as opposede4 to massaging the technology into what you want it to deliver.  5 I hope you get/have had the opportunity to spend timea7 with your Solaris engineers, observe the magic they canl7 weave, much like the folks you're conversing with here.h  ! [Not me, I 'just' USE the stuff].   C I for one really appreciate the time they are taking out discussinge7 what they are doing here, a personal thank-you to them.t   -- r( Regards, Nic Clews CSC Computer Sciences nclews at csc dot coms   ------------------------------  % Date: Thu, 23 Aug 2001 14:04:19 +0100i0 From: andrew harrison <andrew.nospam@uk.sun.com>> Subject: Re: Nuts-n-bolts in News (was: Re: Nits in Slides...)* Message-ID: <3B84FF53.73AB4179@uk.sun.com>   Hoff Hoffman wrote:t > _ > In article <3B83E70A.E12AE091@uk.sun.com>, andrew harrison <andrew.nospam@uk.sun.com> writes:O8 > :It does but any otimism you might want to inject into6 > :this discussion needs to be countered with the fact7 > :that Intel have tried as much as they can to improvee7 > :the performance of the x86 emulation but that it hasi: > :proved to be very hard to emulate x86 and other non ILP% > :binaries into an ILP architecture.i > D >   Cool.  Too bad we're interested in Alpha and VAX, and not IA-32. >   7 Cool, so how what is it that is unique to Alpha and VAx_6 that makes you confident that you won't have the same & issues that x86 emulation on IA64 has.   > .. > > > :Yes but there is nothing in Mckinley for example that would: > :lead anyone to believe that emulated code would run any; > :faster than the overall improvements in performance thato; > :it will deliver. So for example if Mckinley is 2x faster 8 > :than Itanium then the current x86 emulation will also; > :run 2x faster or for some apps at Pentium 200 Mhz speed.  > D >   Cool.  Too bad we're interested in Alpha and VAX, and not IA-32.  7 Cool, so how what is it that is unique to Alpha and VAxo6 that makes you confident that you won't have the same ' issues that x86 emulation on IA64 has ?   < > :Now this may be good enough performance for some apps but8 > :it does depend very much on the proportion of apps on; > :the new platform that have to be translated/emulated and > > :the number that are native. If all your apps are translated: > :then this will be a huge problem if 99% are native then > :it will not.: > ? >   Cool.  Too bad most folks don't look first to use emulatione@ >   or translation.  I'd hope that most folks haven't lost their >   source code. >   
 You hope !  8 > :Running Oracle on a CPU that was in effect delivering7 > :200 Mhz Pentium levels of performance because Oraclee7 > :hadn't ported native would I suggest be unacceptablei > :to most of you customers. > ? >   Cool.  Too bad most folks don't look first to use emulationa@ >   or translation.  I'd hope that most folks haven't lost their >   source code. >   
 You hope !9 Remember you are the company thats putting its customers n8 through a transition that requires them to locate their 8 source code. A transitioon which if the posters to this 3 group are indicative the majority think is totally n unnecessary.  ; > :The biggest job that Compaq has is to try to ensure that 7 > :the translator/emulator never gets used in anger. Itl6 > :remains to be seen if Compaq have the ability to do > :this. > 6 >   Though not if you can help it, I would expect. :-) >   6 Personnally or from Sun's standpoint. Personnaly there5 are no Alpha boxes of any kind in the retail company d2 I advise and the investment bank where I perform a3 similar role has a long standing policy of ridding m6 themselves of Alpha servers because they are no longer3 fit for purpose. So personally I have no influence  # over what happens to Alpha systems.>  1 Sun is already targetting your customer base who -3 are unlikely to be molified by much of what Compaq  2 has published either officially or unofficially in this newsgroup.o  7 Incedentally I was slightly involved in the investment f6 banks decision to switch from Compaq x86 desktops and 3 servers to another supplier. You can't have people c2 supplying a commodity platform which has a longer 3 lead time than an E10K and which is more expensive  ! than the other commodity vendors.     ? > :Quite so people will be migrating off EV7 onto IA64 my pointi< > :exactly and there is no current indication that IA64 will
 > :be faster.  > ? >   Ok, first terminology.  IA-64 is the architecture.  Itanium > >   is the first implementation.  IPF is the Itanium Processor? >   Family.  Itanium may or may not have an advantage over EV7.cC >   Given that Itanium is shipping and EV7 isn't, EV7 will probablys >   be faster. >   > In your homily you forgot ILP (Instruction Level Parallelism).  9 Incedentally I did not say which IA64 implimentation you p: would be making your customers migrate to because it isn't; clear which it will be. It apparently isnt Itanium which isl as much as anyone knows.   > :>9 > :> :2.     The delta between a natively compiled app ons8 > :> :       IA64 and an emulated/translated one is huge- > :> :       this wasn't the case with Alpha.n > :>E > :>   I'm not sure there is enough information to substantiate that.M > :>   Or to deny it.p > :b? > :There is if you compare native ported x86 apps (on Alpha NT)TD > :with FX!32 translated apps on Alpha, in some cases the difference > :was not that marked.b > :eA > :Where as the delta between x86 (emulated code) and native IA64r > :code is huge. > D >   Cool.  Too bad we're interested in Alpha and VAX, and not IA-32. >   7 Cool, so how what is it that is unique to Alpha and VAX 6 that makes you confident that you won't have the same ' issues that x86 emulation on IA64 has ?c  9 > :> :3.     The Alpha designers were carefull to providem< > :> :       good mapping of VAX capabilities (with a couple; > :> :       of significant exceptions which were rectifiedu: > :> :       later) no such care has been taken for IA-64. > :>D > :>   Interesting.  We haven't seen any of this.  (You clearly haveF > :>   far more knowledge of the VAX, Alpha, and IPF architectures andH > :>   the specific implementations.  Would you mind providing specificsE > :>   of where the resulting performance problems will lurk in IPF?)- > :> > :-> > :Page protection for example. 8bit in VAX and Alpha I assume >DI >   The Alpha Architecture actually doesn't have eight protection bits innK >   its Page Table Entry (PTE) design.  It doesn't have any protection bits I >   architected.  (Though Alpha as used by OpenVMS does have eight bits.)e >    Architected ? or implimented ?  H >   It's really cool how we got OpenVMS to work on Alpha, isn't it then? >   C It does however as you know perfectly well have support for VESTED kC applications specifically the RC and RS instructions, strange that oD you forgot to mention this. It also has support for byte allignment / because certain VAX apps ran poorly without it.e     Andrew Harrisons Enterprise IT Architecti   ------------------------------  % Date: Thu, 23 Aug 2001 15:02:28 +0100I( From: Nic Clews <sendspamhere@127.0.0.1>> Subject: Re: Nuts-n-bolts in News (was: Re: Nits in Slides...)) Message-ID: <3B850CF4.9D558647@127.0.0.1>P   andrew harrison wrote: [required snip]o > You hope !: > Remember you are the company thats putting its customers9 > through a transition that requires them to locate theirp9 > source code. A transitioon which if the posters to this 4 > group are indicative the majority think is totally > unnecessary. [another required snip]n  B We have clients happily running stable superceded operating systemE software on hardware that itself was superceded by hardware which hasbG gone 'end of life'. I'm not talking small numbers of systems, even if IhG took my shoes and socks off to assist the counting, I'd still be a good 3 few dozen arms and legs short of the total numbers.(  ? Nobody is forcing anybody to do anything with existing systems.   F I presume that means you are 'somebody', because you are not 'nobody',C or 'anybody', and I'm pretty certain you're not 'everybody' either.e  = A fair number of new implementations are from scratch, so old3H applications aren't an issue. Even if your argument had any credibility,G you've failed to take account of the brand new installations. If you'vel9 ever been involved with such a project, you'll know this.s   -- l( Regards, Nic Clews CSC Computer Sciences nclews at csc dot com    ------------------------------    Date: 23 Aug 2001 11:38:03 -05001 From: Graham Allan <allan@physics.umn.edu.nospam>k> Subject: Re: Nuts-n-bolts in News (was: Re: Nits in Slides...)2 Message-ID: <w534rqyag0k.fsf@pitcairn.spa.umn.edu>  2 andrew harrison <andrew.nospam@uk.sun.com> writes:   > Hoff Hoffman wrote:e > > F > >   Cool.  Too bad we're interested in Alpha and VAX, and not IA-32. > >  > 9 > Cool, so how what is it that is unique to Alpha and VAx 8 > that makes you confident that you won't have the same ( > issues that x86 emulation on IA64 has.  B Well, one obvious thought to me is that, if Alpha-to-IA64 uses theD same combination of emulation and translation as FX!32 did, then youG can ultimately end up with IA64 instructions that are optimised in some*G way. I wouldn't claim to know anything about it, but I would think that B this is a completely different situation from the IA64 silicon (orC microcode, whatever) having to emulate x86 instructions on the fly.    Graham -- gI -------------------------------------------------------------------------  Graham Allan9 School of Physics and Astronomy - University of MinnesotarI -------------------------------------------------------------------------    ------------------------------  # Date: Thu, 23 Aug 2001 13:38:40 GMT # From: Carl Peto <cpeto@bigfoot.com>e Subject: Re: PGP for VMS 6.2* Message-ID: <3B85155A.5040206@bigfoot.com>  H Very long delay in answering this for you but if you're still curious I H would swear blind that I got PGP 2.6.3i to compile on vms 5, I think it C was 5-5.2 or some funny number like that - it was... oh... 1997 or cI something like that so my memory is hazy!  I also used a script for text  A based eve that would encrypt your editor contents at a key-press oC (written by someone else) and I modified it into a DEC-Windows TPU rI compatible version with widgets and stuff.  Have been meaning to post it r! somewhere useful for 4 years!  :)t   Charles Aoki wrote:w  	 > Hi All,e > H > I understand the PGP v2.6.2/3i compiles and runs on OpenVms 6.2.  DoesD > anyone know of the most curent version that will run on a Vax (vms > 6.2).  > 	 > Thanks,1	 > Charles0 > : > Charles Aoki, Univ of Hawaii - Kapiolani Comm Coll, IMTS: > caoki@hawaii.edu  Voice:(808)734-9369  Fax:(808)734-9287 >    ------------------------------    Date: 23 Aug 2001 08:04:24 -0500 From: briggs@encompasserve.org( Subject: Re: Printing Through A Firewall3 Message-ID: <CBceytb8MwrB@eisner.encompasserve.org>r  Y In article <9lvqek$19hq@newton.cc.rl.ac.uk>, "Richard Brodie" <R.Brodie@rl.ac.uk> writes:  > ? > "Paul Anderson" <paul.r.anderson@compaq.com> wrote in messagen7 > news:210820011123510998%paul.r.anderson@compaq.com...- > L >> > Is there anyway to lock down the outgoing port number to just one port? >>I >> You must specify the printer's correct port number, in this case 9100,rJ >> in DCPS$STARTUP.COM.  This number will not change, so I'm confused with3 >> your reference to port numbers of 1100 and 1101.  > I > I assume he means the local port number, which is probably unspecified.oP > Normally it's not important but it would make the firewall rule more specific:V > i.e. "Allow outgoing from x to 9100", rather than "Allow outgoing from any to 9100".  C As I'm sure you know, this doesn't provide much additional securityrE against a cracker who knows how your rules are crafted.  The attackerdE is in control of the source port number.  But, against someone trying * a port sweep, it could be quite effective.   	John Briggs   ------------------------------  % Date: Thu, 23 Aug 2001 14:13:49 +0100-( From: Nic Clews <sendspamhere@127.0.0.1> Subject: Rob: Close but no beerm) Message-ID: <3B85018D.3AA3013F@127.0.0.1>c   :-(N   Got this yesterday.s  ( From: Customer Services @ Streets Online9 Subject: Order Number 2154856 - Item No Longer Available.>   Dear Nic Clews,u  < With reference to order 2154856.  Unfortunately we have been unsuccessfulD in sourcing "VAXCLUSTER PRINCIPLES" Book from our various suppliers.  E This item has now been removed from your order and of course you havet not been charged.'  F We do apologise for this discrepancy, but unfortunately publishers and distributorsH do on occasion cancel products without informing retailers.  However, in manyE instances the manufacturers have released an updated version which ism available to order.  F If you are unable to find an updated version, we do hope that you will find a" suitable alternative on our sites:   ...t   Boo.   --  ( Regards, Nic Clews CSC Computer Sciences nclews at csc dot coml   ------------------------------  + Date: Thu, 23 Aug 2001 07:28:18 -0600 (MDT)t" From: John Nebel <nebel@csdco.com># Subject: Re: Rob: Close but no beerdF Message-ID: <Pine.OSF.4.21.0108230725210.8678-100000@athena.csdco.com>   Nic,  ' There is a used one on Amazon for $130.   2 "Condition: Very Good - Minor dust soil to covers"  
 John Nebel  % On Thu, 23 Aug 2001, Nic Clews wrote:e   > :-(c >  > Got this yesterday.  > * > From: Customer Services @ Streets Online; > Subject: Order Number 2154856 - Item No Longer Available.h >  > Dear Nic Clews,w > > > With reference to order 2154856.  Unfortunately we have been > unsuccessfulF > in sourcing "VAXCLUSTER PRINCIPLES" Book from our various suppliers. > G > This item has now been removed from your order and of course you have  > not been charged.  > H > We do apologise for this discrepancy, but unfortunately publishers and > distributorsJ > do on occasion cancel products without informing retailers.  However, in > manyG > instances the manufacturers have released an updated version which is  > available to > order. > H > If you are unable to find an updated version, we do hope that you will > find a$ > suitable alternative on our sites: >  > ...u >  > Boo. >  > -- m* > Regards, Nic Clews CSC Computer Sciences > nclews at csc dot comt >    ------------------------------  + Date: Thu, 23 Aug 2001 09:46:11 -0700 (PDT)h. From: Fabio Cardoso <fabiopenvms@yahoo.com.br># Subject: Re: Rob: Close but no beerr@ Message-ID: <20010823164611.72165.qmail@web20206.mail.yahoo.com>   Well    ! After years I believe I will havet- my OpenVMS Cluster course at Compaq / Brazil. 4 If there is quorum for the class ... let's pray !=202 What I dont believe is:  the price of a VMS course0 in Brazil is twice "cheaper" than in USA and UK.0 I would like to go abroad for the Cisco courses, but I am over now ....   Regards    FC=20a      - --- Nic Clews <sendspamhere@127.0.0.1> wrote:t > :-(n >=20 > Got this yesterday.r >=20* > From: Customer Services @ Streets Online0 > Subject: Order Number 2154856 - Item No Longer > Available. >=20 > Dear Nic Clews,  >=204 > With reference to order 2154856.  Unfortunately we > have beenh > unsuccessful3 > in sourcing "VAXCLUSTER PRINCIPLES" Book from oura > various suppliers. >=204 > This item has now been removed from your order and > of course you have > not been charged.e >=20+ > We do apologise for this discrepancy, butt > unfortunately publishers and > distributors2 > do on occasion cancel products without informing > retailers.  However, in2 > many6 > instances the manufacturers have released an updated > version which is > available to > order. >=205 > If you are unable to find an updated version, we dot > hope that you will > find a$ > suitable alternative on our sites: >=20 > ...a >=20 > Boo. >=20 > --=20o* > Regards, Nic Clews CSC Computer Sciences > nclews at csc dot como     =3D=3D=3D=3D=3DvL =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D. F=E1bio dos Santos Cardoso OpenVMS System Manager Petrobras / Rio de Janeiro BrazilL =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3Db  2 __________________________________________________ Do You Yahoo!?H Make international calls for as low as $.04/minute with Yahoo! Messenger http://phonecard.yahoo.com/e   ------------------------------  % Date: Thu, 23 Aug 2001 15:48:51 +0200m< From: Georges SZAFRANSKI <szafranski@lutece.evt.cpqcorp.net> Subject: RWMBX6 Message-ID: <3B8525E3.6BC639C7@lutece.evt.cpqcorp.net>  F As a quick fix, you can try to increase DEFMBXBUFQUO SYSGEN parameter. It's dynamic, BUT system wide.eH When you find the suitable value, I'd suggest you to read the I/O user's< guide(chap 4 Mailbox driver) and to recode your application.   ------------------------------  % Date: Thu, 23 Aug 2001 07:48:49 -0500w( From: "Bill Ames" <billames@accunet.net>< Subject: RWMBX - What to do, What to do! Help me if you can./ Message-ID: <to9urmfl18vj34@corp.supernews.com>a  A Well I wrote my routine to check on RWMBX states and report to memK periodically.  Does anyone know it RWMBX is normal or if there is something F I can do (a setting) to keep them from popping up so often. I have oneK machine that hits this state often throughout the day.  I realize the RWMBXr? is a mail box wait state.  So can I increase the mail box size?t   Thanks,  Bill   FYITD I work for a company that has their business system running on 7 AXPH servers.  I develop programs in BASIC, COBOL and C which generally deals, with EDI. So my VMS and DCL are a bit rough.   ------------------------------    Date: 23 Aug 2001 08:45:07 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen)e@ Subject: Re: RWMBX - What to do, What to do! Help me if you can.3 Message-ID: <83O1TmF9Vl+E@eisner.encompasserve.org>d  Z In article <to9urmfl18vj34@corp.supernews.com>, "Bill Ames" <billames@accunet.net> writes:C > Well I wrote my routine to check on RWMBX states and report to meeM > periodically.  Does anyone know it RWMBX is normal or if there is something H > I can do (a setting) to keep them from popping up so often. I have oneM > machine that hits this state often throughout the day.  I realize the RWMBX A > is a mail box wait state.  So can I increase the mail box size?   F > I work for a company that has their business system running on 7 AXPJ > servers.  I develop programs in BASIC, COBOL and C which generally deals. > with EDI. So my VMS and DCL are a bit rough.  0 But this problem has very little to do with DCL.  C Look at your BASIC, COBOL or C code and adjust the parameter BUFQUOm4 in your call to SYS$CREMBX that creates the mailbox.   ------------------------------  # Date: Thu, 23 Aug 2001 14:18:19 GMTa1 From: "Mark D. Jilson" <jilly@clarityconnect.com>M@ Subject: Re: RWMBX - What to do, What to do! Help me if you can.1 Message-ID: <3B851145.C673B0A@clarityconnect.com>t  F RWMBX may be normal for an account that is underconfigured for BYTLM. H For example, if the process is trying to issue a mailbox IO for 10 bytesC and BYTLM says there are only 9 bytes available and the process hasaD other outstanding buffered IO then the process will go into RWMBX toD wait for any of those other outstanding buffered IO to complete thusC returning some BYTLM.  This may also happen if the process wants toeG write a n byte message to a mailbox that only has <n bytes free but the D mailbox has a non-zero message count.  The process will be placed inE RWMBX to wait for messages to be removed from the mailbox which wouldiE allow space for this message to be written.  DSNlink has an excellent 2 article in it for troubleshooting the RWMBX state.   Bill Ames wrote: > C > Well I wrote my routine to check on RWMBX states and report to me M > periodically.  Does anyone know it RWMBX is normal or if there is somethingsH > I can do (a setting) to keep them from popping up so often. I have oneM > machine that hits this state often throughout the day.  I realize the RWMBXeA > is a mail box wait state.  So can I increase the mail box size?  > 	 > Thanks,i > Bill >  > FYIuF > I work for a company that has their business system running on 7 AXPJ > servers.  I develop programs in BASIC, COBOL and C which generally deals. > with EDI. So my VMS and DCL are a bit rough.   --  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: Thu, 23 Aug 2001 14:21:30 GMT = From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-)t@ Subject: Re: RWMBX - What to do, What to do! Help me if you can.0 Message-ID: <00A00F45.D6F9CFC5@SendSpamHere.ORG>  c In article <83O1TmF9Vl+E@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes:e[ >In article <to9urmfl18vj34@corp.supernews.com>, "Bill Ames" <billames@accunet.net> writes:dD >> Well I wrote my routine to check on RWMBX states and report to meN >> periodically.  Does anyone know it RWMBX is normal or if there is somethingI >> I can do (a setting) to keep them from popping up so often. I have oneeN >> machine that hits this state often throughout the day.  I realize the RWMBXB >> is a mail box wait state.  So can I increase the mail box size? >IG >> I work for a company that has their business system running on 7 AXP K >> servers.  I develop programs in BASIC, COBOL and C which generally dealsh/ >> with EDI. So my VMS and DCL are a bit rough.n >m1 >But this problem has very little to do with DCL.  > D >Look at your BASIC, COBOL or C code and adjust the parameter BUFQUO5 >in your call to SYS$CREMBX that creates the mailbox.   I In addition, find out WHY you are getting a RWMBX.  Is it that the writereH is writing to the mailbox faster than the reader reads from the mailbox?H Is the reader (perhaps a separate process) dying and causing the writer H (perhaps another process) to write until it fills the mbx and thus, getsI itself wedged into the RWMBX state?  If you don't understand the cause, afJ BUFQUO increase may NOT correct the problem, only postpone its occurrance.   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMe            XJ   "And of course, I'm a genius, so people are naturally drawn to my fiery I   intellect.  Their admiration overwhelms their envy!" -- Calvin & Hobbes    ------------------------------  % Date: Thu, 23 Aug 2001 14:04:27 -0400b- From: JF Mezei <jfmezei.spamnot@videotron.ca> @ Subject: Re: RWMBX - What to do, What to do! Help me if you can., Message-ID: <3B8545A8.B18F3009@videotron.ca>   Bill Ames wrote: > C > Well I wrote my routine to check on RWMBX states and report to metM > periodically.  Does anyone know it RWMBX is normal or if there is somethingoH > I can do (a setting) to keep them from popping up so often. I have oneM > machine that hits this state often throughout the day.  I realize the RWMBXaA > is a mail box wait state.  So can I increase the mail box size?a  L Does the process "unfreeze" itself ? If so, then you are correct, increasingG the mailbox buffer when the program creates the mailbox will allow more-G mailbox messages to be queued/written to the mailbox before the reading J process reads the messages. It is a sign that processes are writing to theI mailbox faster than the reading process can take them out of the mailbox.   J If, however, the process hangs and stays in RWMBX forever, the trick is toJ ANA/SYSTEM, SET PROCESS <name of process in RWMBX), show proc/CHANNELS and+ that will tell you the mailbox name/number.h  4 Then, you can write a simple DCL procedure tsuch as:   $OPEN/READ temp MBAxxx:s $loopa $READ/end=finish temp test
 $goto loop $finish: $exit   K The above will effectively empty the buffer of a mailbox by reading all theeG stuff from it, thus allowing processed that had tried to write to it tob complete their IO and continue.m  J Note that the above procedure should be SPAWN/NOWAIT  to run as a separateE process which you can kill (STOP/ID) once it has emptied the mailbox, L otherwise it will continue to consume messages instead of allowin the server to get them.   ------------------------------  % Date: Thu, 23 Aug 2001 11:23:02 +0100y% From: Alan Greig <a.greig@virgin.net>i( Subject: Re: Some postive points I hope.8 Message-ID: <qkl9otkjvo39606bhgeggc5a0acc4kuofk@4ax.com>  , On Wed, 22 Aug 2001 22:12:31 -0400, JF Mezei% <jfmezei.spamnot@videotron.ca> wrote:l    M >Generally, system managers of serious systems tend to design and prepare forCM >D-R much better whereas companies generally hire cheap newbies to handle the-I >systems that they bought because they were cheap. But put an experienced.N >manager at the helm of an NT system, and you might get some fairly impressive@ >results if the later can plan well, inside of NT's limitations.  E We do have experienced NT managers. Problem is that DR on NT involves ? first install the OS, then install the backup software disasterAF recovery module then load up the indices, then start recovering on theE large number of NT servers you need because trying to have one or twotE NT servers do everything (as you can do with VMS or Unix) is a recipea for disaster in itself.t  E Then you find all the minor incompatibilities between your supposedly E identical DR hardware which cause networks, graphics etc to misbehave*C or crash. You can't get anywhere near VMS's ability to just restoretE the OS plus data to any old system with the power an disk capacity tor handle the load.  E So unless you actually buy duplicate cold standby systems at the time-F of original purchase with absolutely everything identical you stand no= chance of just restoring images and expecting things to work.i  D I know we've been through all this with the DR company supplying theD kit and they confirm that NT *always* takes far longer to recover inB their experiences than VMS or Unix. VMS winds hands down virtually every where they go.   >DJ >The best surprise is no surprise. If you know what to expect during a D-RK >plan, and if you train and test at regular intervals, then the process, nocZ >matter how long it takes (minutes, days) should be smooth and go as was planned/expected.  F We plan/expect NT recovery to be relatively slow and laborious. That'sF one of th reasons why absolutely critical systems are on VMS then Unix for the most part.     -- Alan   ------------------------------  % Date: Thu, 23 Aug 2001 16:25:18 +0200n& From: Michael Joosten <joost@c-lab.de>( Subject: Re: Some postive points I hope.$ Message-ID: <3B85124E.6956@c-lab.de>   David Mathog wrote:v >  > john nixon wrote:s >    > N > This is exactly what I would have predicted.  And had you thrown NT into the > mix I'd expect that the1O > small army of sysops needed to keep them running would still be struggling toa  > get all their systems back up. > G And if it would only because the PC server refused to boot, though they0E insisted that everything was right with disk - only to discover later C that the Disk Manager silently resetted the active flag of the bootgD partition. Yes, the little green cross was just too small. Nice GUI, guys...l This actually happened here.    Q > Over the weekend the IBM Deskstar 75 Gb drive in my Solaris box died.  (After al" > piddling 4 months of service).    E I really wonder how large the failure rate of ten year old 3.5" disks G will be. I'm about to send two old M2366 (1GB, 5.25") bricks to someone1A who likes to have a fitting old system. Very few of them actuallycG failed. Contrast that with the 3.5" disks at least until '96 - very few  of them survived !  *   Restoring the system wasn't particularly1 > difficult in terms of the operations - in fact,mQ > it was very much like restoring a VMS system (although I've only ever done thatrM > from a local drive). There were some extra steps on the Solaris end though.p >  > VMS  Solaris= > +         +           open the box, replace the failed diskaG > +         +           boot from CDROM to a running version of the OS.nP > -         +           start up enough of a network to read the tape across the > local net-8 > -         +           partition the drive  (2 minutes)I > -         +           create file systems on the partitions (2 minutes)mK > +         +           restore the drive/partition using backup/ufsrestore M > -         +           run fsck on the restored partition (times varied, <10m
 > minutes)A You could skip that. Unmounting would be enough before rebooting.   Q > -         +           installboot (set the disk to boot the correct partition 1o	 > minute)uM > +         -           set SRM to boot from the disk (if replacing an entireu	 > system)  > +         +           reboot > O > Also not indicated in this is that for VMS you'd just do the one BACKUP/IMAGEn > per disk, whereasoL > for Solaris (and other Unices) you generally have to backup each partition > separately, ie, steps 5-7iI > have to be done N times per disk, where N is the number of partitions..n >  Sometimes that's M > a good thing (one corrupted partition) and usually it's not (I usually losed > whole disks, not > single partitions).a > N > The thing I did learn the hard way from this experience is that ufsdump runsQ > several fold faster than does ufsrestore.  So if you're expecting that since iteL > took 5 hours to make all the backup tapes, it will take 5 hours to restoreQ > them, you're in for another 10 hours or so of unexpected down time.  And if youuL > were expecting to use the second 5 hours to restore another system, you'reH > probably going to have to sleep on the computer room floor.  BACKUP isM > apparently able to keep a DLT streaming both when writing to, and restoringeK > from a save set, and my recollection is that making an image save set and O > restoring one take pretty much the same time.  ufsrestore, on the other hand, N > has some intrinsic performance problems.  There are other unix backup tools,N > variants of tar, for instance, but last I looked none of the free ones  know5 > how to handle an ACL (which we use on this system).c >   H The main problem with ufsrestore is, that it uses the normal file systemE routines. So, the whole data goes through the buffer cache and so on.vE There is/was a small program, 'fastfs', that switched the sync. inoden: updates off. VERRY useful for restoring or copying disks. F OTOH, this is also an advantage - you can restore a ?fsdump on another filesystem, even via NFS.c  H Joerg Schili's 'star' seems to be an alternative, though I'm not sure ifG he already has integrated ACLs. And yes, streaming with the native UNIXmH backup tools is somewhat 'underdeveloped'. There are 'buffer' and 'team'D to be placed in pipes before/after dump and restore. 'star' even has it's own in-memory buffering.r  G For disaster recovery, one should always prepare an own spare disk withaF the neccesary stuff. Booting ANY system via CD is tedious, and what doD you do when your backups are made by Networker or such ? (Note: withE Solaris, I'd recommend a single partition for / and /usr at least for-E the recovery disk. Once you change controllers, shift them around they< PCI slots, you are SOL when trying to reboot from an image!)  M > It's really odd to me that after all these years there's still no Unix toolC > around that's smart enoughM > to backup a full disk, and give the option of restoring the disk (partitionn( > table, fs, and data) or just restoringL > one partition.   After all, if this Uber tool just stored: partition info,H > boot info (if any), dump of partition1, dump of partition2..., dump ofN > partitionN, it would seem to have all of the data required to rebuild a fullM > disk automatically, or just a subset of partitions,  using nothing but a shy$ > script and all the existing tools.  F Right, but this is quite system dependant (though SGI's Backup/RestoreC scripts do exactly this, IIRC), as it involves disk labels and boot C block installations. Unfortunately this wasn't really in the set of=@ little tools a SystemV or BSD4.3 base 'distribution' contained.   ? And... there is that 'differentiation' effect, with many system3C engineers knowing exactly that they could do it better than others, A notably the competition. (it's not only the marketing...). Not to 8 mention the software houses that jumped into that niche.  8 Well, and now excuse the UNIXish/Solarisish rant here...     -- _* Michael Joosten, SBS C-LAB, joost@c-lab.de* Fuerstenallee 11, 33094 Paderborn, Germany, Phone: +49 5251 606127, Fax: +49 5251 6060658 C-LAB is a cooperation of University Paderborn & SIEMENS   ------------------------------  % Date: Thu, 23 Aug 2001 16:45:11 +0200c& From: Michael Joosten <joost@c-lab.de>( Subject: Re: Some postive points I hope.$ Message-ID: <3B8516F7.1372@c-lab.de>   Alan Greig wrote:U >    > G > We do have experienced NT managers. Problem is that DR on NT involveseA > first install the OS, then install the backup software disasterxH > recovery module then load up the indices, then start recovering on theG > large number of NT servers you need because trying to have one or twocG > NT servers do everything (as you can do with VMS or Unix) is a recipe  > for disaster in itself.@ >   H NT/Win2000 not being able to boot even minimally from CDROM is truckload of nonsense by itself. But... H I've made quite good experiences with external disks containing just theF minimal set of tools for reconfiguration and the backup software. IwasB surprised to see that such a SCSI disk was even able to boot off aD different SCSI controller provided the driver was already installed.F It might even usefull to designate an old EIDE disk for this, providedF that one could fit that disk sufficiently easy back into the server...  G > Then you find all the minor incompatibilities between your supposedly-G > identical DR hardware which cause networks, graphics etc to misbehavepE > or crash. You can't get anywhere near VMS's ability to just restoredG > the OS plus data to any old system with the power an disk capacity tog > handle the load. >   F Yes... So far about cheaper controllers and the like. You are probablyD well advised to buy a second set at the SAME time when buying the PCH server. Just in case, because a year later you won't be able to get themA anymore - or only with different/incompatible firmware/hardware.  A See also the 3COM 3C905b/c problems recurring in alt.solaris.x86..  G > So unless you actually buy duplicate cold standby systems at the timemH > of original purchase with absolutely everything identical you stand no? > chance of just restoring images and expecting things to work.d >    Chance yes, but no gurarantee.  H You are from GB, why not offer bets for the admins on the amount of timeE a recovery will take? Granted, the NT ones will need a better rate ort' handicap... Give them an incentive! 8-)a   -- e* Michael Joosten, SBS C-LAB, joost@c-lab.de* Fuerstenallee 11, 33094 Paderborn, Germany, Phone: +49 5251 606127, Fax: +49 5251 6060658 C-LAB is a cooperation of University Paderborn & SIEMENS   ------------------------------  % Date: Thu, 23 Aug 2001 08:11:04 -0700 ' From: David Mathog <mathog@caltech.edu>r( Subject: Re: Some postive points I hope.+ Message-ID: <3B851D08.363E4A0A@caltech.edu>    Bill Todd wrote:  6 > "David Mathog" <mathog@caltech.edu> wrote in message' > news:3B843379.9D4EB6B8@caltech.edu...p >e > ...  >iI > > Over the weekend the IBM Deskstar 75 Gb drive in my Solaris box died.t
 > (After a" > > piddling 4 months of service). > L > I remember reading somewhere that the 75 GB IBM drives made in Hungary hadG > defects leading to very short service lives.  You might want to use aw" > different disk as a replacement. >s  M My experience has been that if you have a drive fail early and you replace it  with another drive of the G same type,  that will fail early too.  (Ie, most drive failures are badeM design or bad production runs),  So there's a 100 Gb Western Digital drive inmL there now.  These things are so inexpensive these days that it's ridiculous,L 100 Gb for $300 retail at the local computer superstore.  Conversely, a tape  drive big enough and fast enoughM to back it up is going to run into the thousands, and tape media large enough  to hold that much data costsH a substantial fraction of the drive price, with tape cartridges that areL roughly the same size as the disks.   Our existing DLT4000s don't really cutM it for disks of this size (although they were great back when disks of 4 or 9o Gb were the the norm.)  G Looks like I'm soon going to be using whole disks as backups instead ofy tapes.   Regards,   David Mathog mathog@caltech.edu   ------------------------------  % Date: Thu, 23 Aug 2001 13:57:40 -0400-- From: JF Mezei <jfmezei.spamnot@videotron.ca>u( Subject: Re: Some postive points I hope., Message-ID: <3B854412.494A2CB6@videotron.ca>   Alan Greig wrote:aG > Then you find all the minor incompatibilities between your supposedlySG > identical DR hardware which cause networks, graphics etc to misbehaveiE > or crash. You can't get anywhere near VMS's ability to just restorevG > the OS plus data to any old system with the power an disk capacity toW > handle the load.  N That is what tests are for. A disaster recovery plan is much more than writingH a set of procedures to use in case of a disaster (or in the case of VMS,F setting up a dual site VMS cluster), it involves a lot more, includingR telecom, operator consoles etc, but more importantly, it involves regular testing.  L And it is during that testing that you will find out that some of the backupN boxes aren't quite compatible, and once you have discovered this, you can takeL paliative actions such as buying compatible hardware cards that you can plugP in during D-R activity so that the backup system will run the production system.  M But in a true D-R plan, shouldn't you just have to move your application data N over to the backup machine with the remainder already being there and ready to use ?   L Many in the VMS community are spoiled brats because D-R is so easy if you'reM given a pair of fibres between two buildings. But if you don't have access toiI that and volume shadowing is not an option, then you start having similareL problems to other other platforms even though they may be solvable in a moreI elegant solution (a single backup for instance instead of multiple ones)..   ------------------------------  % Date: Thu, 23 Aug 2001 16:48:59 +1000E/ From: "Phil Howell" <phowell@snowyhydro.com.au>  Subject: Re: Sqlsrv_manage1 Message-ID: <lI1h7.1404$vI2.78764@ozemail.com.au>a  3 "A Bonaveidogo" <Asena@fsc.com.fj> wrote in message-, news:01C12BE5.44AEDC20@PATRICK.FSC.COM.FJ... >0L > We're using Oracle RDB SQL V7.0-5 on Open VMS V7.1-2.  Where can I get the above utility ie SQLSRV_MANAGE?! >l > AB >o It should be in sys$system  & SYS$COMMON:[SYSEXE]SQLSRV_MANAGE.EXE;1
 ANALYZ A07-04I  (         Image Identification Information  +                 image name: "SQLSRV_MANAGE"k4                 image file identification: "V7.0-00"3                 image file build identification: ""t7                 link date/time: 28-SEP-1996 00:35:57.66u/                 linker identification: "A11-14"   > or else you will have to reinstall some Rdb components from cd Phil   ------------------------------    Date: 23 Aug 2001 08:36:47 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen)o Subject: Re: Take it elsewhere3 Message-ID: <KaNxP8zT6HZG@eisner.encompasserve.org>e  [ In article <3B8472BA.9F87376E@fsi.net>, "David J. Dachtera" <djesys.nospam@fsi.net> writes:t > Larry Kilgallen wrote: >>  } >> In article <JsNg7.10291$2u.73324@www.newsranger.com>, Simon Clubley <simon_clubley@remove_me.excite.com-Earth.UFP> writes:R >>  J >> > Actually, what may be a better idea is to follow the example of otherL >> > newsgroups and put forward a proposal to create comp.os.vms.discuss forM >> > political/policy discussions and leave comp.os.vms for technical issues.n >> >M >> > What do other people think ? Note that I am _NOT_ offering to drive thislO >> > proposal, but am offering it as a suggestion in case others want to create  >> > a formal proposal.t >> m7 >> The "powers that be" frown on having any newsgroups:  >> n. >>                 This.That.Theother.Subtopic >> o >> at the same time as:i >>  % >>                 This.That.Theother  >> g& >> also exists.  Pascal ran into this. > 5 > Dunno 'bout that. Among my newsgroup subscriptions:u > comp.dcom.modems > comp.dcom.modems.cable > F > ...and others, mostly Linux groups that I seldom look at these days.  C I don't know what it took to create those (or when).  I do know the1( grief the Pascal folks were given. YMMV.   ------------------------------    Date: 23 Aug 2001 08:14:55 -0500 From: briggs@encompasserve.org Subject: Re: TAR files in VMSs3 Message-ID: <SU75zdGypPX9@eisner.encompasserve.org>s  b In article <3b83ff93$1@news.si.com>, "Brian Tillman" <tillman_brian@notnoone.notnohow.com> writes:J >>Is there a utility that will just copy one TAR tape to another TAR tape,E >>without creating the intermediate VS file?  Of course, I need a VMSs
 > utility. > ; > Assume the tapes are $10$MKA100 and $10$MKA200.  DO this:b > ' > $ mount $10$mka200 whateverthelabelisn  > $ mount $10$mka200: otherlabel' > $ copy $10$mka100:[]*.* $10$mka200:[]e > $ dism $10$mka100: > $ dism $10$mka200:  G If we're talking about standard tar tapes as might be created on a Unix D system, for instance, then the tapes will not be labelled.  Instead,C they will simply contain a series of 10240 byte records followed byrH a tape mark  (The last record might be a smaller multiple of 512 bytes).  . To copy a single tar archive from such a tape:  ( $ MOUNT $10$MKA100 /FOREIGN /BLOCK=10240( $ MOUNT $10$MKA200 /FOREIGN /BLOCK=10240# $ COPY /LOG $10$MKA100: $10$MKA200:t# $ COPY /LOG $10$MKA100: $10$MKA200:e  G The first copy takes the archive from one tape to the other.  Check thegF statistics produced by /LOG to get a sanity check that you've actuallyH got the archive and not just some descriptive text file that some admins: (such as myself) sometimes put on the front end of a tape.  C The second copy probably just copies the zero record "file" between B the two tape marks that conventionally indicate end of volume.  IfE the statistics on that copy indicate anything other than zero records H copied then you've got multiple files on your source tape.  One approachE in that case is to simply repeat the COPY operation to copy each filetF across until you get a zero record file (two adjacent tape marks whichG by convention indicate end of volume) or a parity error (if you run off-E the end of the portion of the tape that was written, you get a parityr error).e  5 Other useful commands on a /FOREIGN mounted tape are:n   	$ DUMP device-name: 	$ SET MAG /REWIND device-name  $ 	and, if you know what you are doing  D 	$ SET MAG /SKIP=FILE=n (positive or negative both work) device-name   	John Briggs   ------------------------------  " Date: Thu, 23 Aug 01 11:58:42 +100 From: rok@nuk.uni-lj.sib" Subject: Unexpected DIAGPACK error& Message-ID: <3b84eff9.0@NUK.Uni-Lj.Si>  . AlphaServer 4100 5/400 4MB, OpenVMS AXP V7.2-1    I get  + %INIT-F-DIAGPACK, disk is a diagnostic packo  4 when I try to initialize an device type DEC RZ1DF-CB5 (DS-RZ1DF-WW 9.1GB) disk which used to be volume 0 oft. multidisk ODS-2 converted to ODS-5 volume set.  2  As I do not like the suggested user action at all4 (Retry the operation using another volume) I will be gratefull for any hints.   Regards,  D Rok Vidmar                       Internet:  rok.vidmar@nuk.uni-lj.siM National and University Library  Phone:     +386 1 421 5461 <-- changed againtM Turjaska 1, SI-1000 Ljubljana    Fax:       +386 1 425 7293 <-- changed again  Slovenia   ------------------------------  # Date: Thu, 23 Aug 2001 14:11:20 GMTu1 From: "Mark D. Jilson" <jilly@clarityconnect.com>r& Subject: Re: Unexpected DIAGPACK error2 Message-ID: <3B850FA0.859781BD@clarityconnect.com>  G Redo the INIT command with /NOVERIFIED qualifier.  It appears that INITeE is getting confused by what it is attempting to read as the bad block D data.  If this INIT works then retry it again without the qualifier.   rok@nuk.uni-lj.si wrote: > 0 > AlphaServer 4100 5/400 4MB, OpenVMS AXP V7.2-1 >  >  I get > - > %INIT-F-DIAGPACK, disk is a diagnostic packe > 6 > when I try to initialize an device type DEC RZ1DF-CB7 > (DS-RZ1DF-WW 9.1GB) disk which used to be volume 0 of 0 > multidisk ODS-2 converted to ODS-5 volume set. > 4 >  As I do not like the suggested user action at all6 > (Retry the operation using another volume) I will be > gratefull for any hints. > 
 > Regards, > F > Rok Vidmar                       Internet:  rok.vidmar@nuk.uni-lj.siO > National and University Library  Phone:     +386 1 421 5461 <-- changed againuO > Turjaska 1, SI-1000 Ljubljana    Fax:       +386 1 425 7293 <-- changed again 
 > Slovenia   --  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: Thu, 23 Aug 2001 10:22:18 +0000 (UTC)e' From: david20@alpha2.mdx.ac.uk (D.Webb) % Subject: Re: V5.5-2 Password Recovery)+ Message-ID: <9m2lgq$ah9$1@aquila.mdx.ac.uk>l  i In article <3B8407D4.A1659DFA@gtech.com>, Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com> writes:f >Carl Karcher wrote:D >> You can't without resorting to brute force (trying every possibleL >> password character combination). Read access to the SYSUAF.DAT file wouldG >> be required. As an exercise, I found cracking a 6 character password 1 >> takes about 6 days on an Alphaserver 1200/533.h >s8 >Before trying systematic combinations, then trying with= >a list of known words is probably optimal (f.ex. 50000 knowny= >words is very small compared to all 6 character combinationsn+ >and the chance of a hit is relative high).  >t >Arne   K Unless of course the words you choose are in the system dictionary in whicht> case you have (on most VMS systems) just wasted a lot of time.  I (probably best would be checking against a list of words but appending ort* prepending a couple of numerics to them.).  
 David Webb VMS and Unix team leader CCSS Middlesex University   ------------------------------    Date: 23 Aug 2001 07:48:34 -0500 From: briggs@encompasserve.org& Subject: Re: VAX: DEC-C float & double3 Message-ID: <1cHVytKyut22@eisner.encompasserve.org>h  _ In article <21AUG200122365393@eql14.caltech.edu>, rankin@eql14.caltech.edu (Pat Rankin) writes:d6 > In article <Db5zI9d2wYEe@eisner.encompasserve.org>,\% >  briggs@encompasserve.org writes...tI >> The in-memory layout of the first four bytes of a D-Floating double isgM >> identical to the in-memory layout of the first four bytes of an F-Floating'H >> float.  Bytes 5-8 of the D-Floating value are just low order mantissa >> bits. >>H >> If you treat a D as an F, the effect is an implicit truncation of theG >> ignored low order bits.  If you treat an F as a D, the effect is thesD >> implicit appending of whatever random bits appear next in memory. > = >      That's true, but irrelevant.  The %f specifier is usedi= > to format doubles, and the value of `sky' which gets passed-6 > to printf will be converted to double by the caller.  G You are, of course, correct.  I had completely forgotten about floatinge3 point promotion in the context of subroutine calls.a   	John Briggs   ------------------------------  % Date: Thu, 23 Aug 2001 08:47:23 -0400g5 From: David Beatty <David.Beatty@qwertysasasdfgh.com>l5 Subject: Re: VMS high reliability needed by Air ForceA2 Message-ID: <FvuEOzrbjhPNnu6j=KuWnenM6m6p@4ax.com>  7 On Wed, 22 Aug 2001 22:37:24 -0500, "David J. Dachtera"t <djesys.nospam@fsi.net> wrote:   >Randy Burlew wrote: >> fL >> In article <3B82C1F4.CE30489F@videotron.ca>, jfmezei.spamnot@videotron.ca
 >> says... >> >P >> >But in light of the Alpha murder, consider that this deal was in the works aO >> >long long time ago, having a *NEW* deal signed in light of the alpha murder) >> >would have far more value. >>  < >> I'm a big supporter of free speech and all, but this talk> >> of the Alpha "murder" is getting a wee little bit tiresome. >> n: >> Besides, computer language is already too violent (kill# >> the process, abort, crash, etc.)l >h) >Story told by a co-worker some time ago:o >lH >Overheard at a party: a tech. support person on the phone to a computerH >operator: "It did, huh? Well, o.k. Just take a dump and I'll look at it >in the morning. Bye."  B ... that one's right up there with "Take an aspirin and call me in< the morning", although opposite ends of the intestinal tract" are taking the appropriate action!   ------------------------------  % Date: Thu, 23 Aug 2001 09:28:55 -0400 + From: Brendan Welch <brendan_welch@uml.edu>m5 Subject: Re: VMS high reliability needed by Air Forceo' Message-ID: <3B850517.A8B385E8@uml.edu>F  9 > Besides, computer language is already too violent (kille" > the process, abort, crash, etc.) > A You forgot the obvious,  EXECUTE  the program !   We execute bothu& directives and living things.  tee hee   --? Brendan Welch, system analyst, Univ. of Massachusetts - Lowell,o W1LPGr   ------------------------------  % Date: Thu, 23 Aug 2001 09:57:13 -0600g4 From: "Michael D. Ober" <mdo.@.wakeassoc.com.nospam>5 Subject: Re: VMS high reliability needed by Air Forceo2 Message-ID: <uJ9h7.182$Xx3.241544@news.uswest.net>  F My favorite was when our Honeywell systems crashed, the first four hex3 characters in the crashdump were frequently "DEAD".c --
 Mike Ober.  8 "Brendan Welch" <brendan_welch@uml.edu> wrote in message! news:3B850517.A8B385E8@uml.edu...  >f; > > Besides, computer language is already too violent (kille$ > > the process, abort, crash, etc.) > >lC > You forgot the obvious,  EXECUTE  the program !   We execute bothn( > directives and living things.  tee hee >u > --A > Brendan Welch, system analyst, Univ. of Massachusetts - Lowell,a > W1LPGw   ------------------------------  % Date: Thu, 23 Aug 2001 12:09:03 -0400 - From: "John Eisenschmidt" <jeisensc@aaas.org>d5 Subject: Re: VMS high reliability needed by Air Forcen+ Message-ID: <sb84f275.078@AAASMTA.aaas.org>   J IBM hardware (and I see now Alphas) when they come up init everything to = 0xdeadbeef.    "Waiting for deadbeef"  K >>> "Michael D. Ober" <mdo.@.wakeassoc.com.nospam> 08/23/2001 11:57:13 AM =  >>>tF My favorite was when our Honeywell systems crashed, the first four hex3 characters in the crashdump were frequently "DEAD".  --
 Mike Ober.  8 "Brendan Welch" <brendan_welch@uml.edu> wrote in message! news:3B850517.A8B385E8@uml.edu...  >c; > > Besides, computer language is already too violent (killn$ > > the process, abort, crash, etc.) > >hC > You forgot the obvious,  EXECUTE  the program !   We execute bothG( > directives and living things.  tee hee >m > --A > Brendan Welch, system analyst, Univ. of Massachusetts - Lowell,e > W1LPGt   ------------------------------  % Date: Thu, 23 Aug 2001 18:57:35 +0200L= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com> , Subject: Re: VMS/Alpha backend for GCC 3.x ?) Message-ID: <3B8535FF.F3EC3CC3@gtech.com>o   John Malmberg wrote: > Simon Clubley wrote:A > >>>Does anybody know of any moves to add a VMS/Alpha backend tog; > >>>GCC 3.x along with any necessary changes to binutils ?s  H > If the GNAT frontend produces the GNU assembler format on output, thenG > all you really need to port for the backend is the assembler program.e > J >  From the last time I played with the assembler (on VAX) it was supplied8 > and built separately from the rest of the GCC package.  ' The GNU Assembler is part of binutils !w   Arne   ------------------------------  % Date: Thu, 23 Aug 2001 13:34:53 -0400 5 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com>i$ Subject: Re: Wailing and Moaning....2 Message-ID: <H8bh7.668$bB1.31448@news.cpqcorp.net>  A David J. Dachtera wrote in message <3B8474A7.291B0EAB@fsi.net>...a >eA >If I pulled my punches because I was concerned about what othersiA >thought, I'd have to crawl in a hole and hate myself for being a.I >hypocrite. I wouldn't respect someone who did, I would expect no respectf# >from anyone who saw me do it here.s >rI >A mentor of mine (Denis Waitley) said, "If you don't stand for something  >you'll fall for anything".s >s    H "It is better to stay silent and perhaps be thought a fool, than to open) ones mouth and remove all doubt."  Annon.d   ------------------------------  % Date: Thu, 23 Aug 2001 02:37:22 -0400o( From: Hamlyn Mootoo <univms@bigfoot.com>= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...)h+ Message-ID: <3B84A4A2.DAD78BE8@bigfoot.com>b   Rob Young wrote: [snip]@   Two skilled Unix "persons" making a mistake in a shell script.O > >>         Two high profile clients dead in the water for a day or two.  Why?  > >>, > >>         # rm -r ${variable_goes_here}/* > >R; > > Couldn't have been too "skilled" to make that mistake !6 > >o > M >         Absolutely skilled.  One of the people is a 15 year veteran, admin,wI >         system configurator... one of the brighter guys you would know.s > ! >         Trust me... bad design.'  B All this example serves to illustrate is the power of the the UnixG command syntax.  And whoever coded the above example, regardless of his.G years of systems management experience, is obviously not experienced atnG writing shell scripts.  Actually, it's doubtful that whoever wrote thateD had many months, let alone years of scripting experience. Any decentB shell scripter would NEVER, I repeat NEVER use that combination ofH syntax for just the exact reason of the variable being undefined.  Don'tH take my word for it. Post this example in any of the Unix newsgroups andE when they get done laughing maybe they'll respond if anyone can stillg# manage to type without cracking up.aH    Especially when it comes to text processing a Unix one-liner can takeA the place of a 20 line DCL procedure in some cases.  I have quiteoG frequently needed to massage some textual data, and FTPed the file to aeG Unix system to accomplish in seconds, what would have taken a half hourrA on VMS to do with DCL (because a whole procedure would have to beiH written and debugged).  Piping in Unix is extremely powerful; that's whyG VMS has its poor-man's equivalent today.  One thing I would like to seetG in VMS is a straight redirection symbol (">") without having to use thenG "PIPE" command in front of it.  Also, the ability to take the result of-G any expression and assign it to a variable without having to go throughr/ intermediate steps would also be very useful.  u   HM   ------------------------------  % Date: Thu, 23 Aug 2001 19:46:29 +0010A% From: paddy.o'brien@zzz.tg.nsw.gov.aui= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...) 5 Message-ID: <01K7HH6BC7IA0049M4@tgmail.tg.nsw.gov.au>W   Rob Young wrote: [snip]@   Two skilled Unix "persons" making a mistake in a shell script.K > >>         Two high profile clients dead in the water for a day or two.  d Why? > >>, > >>         # rm -r ${variable_goes_here}/* > >_; > > Couldn't have been too "skilled" to make that mistake !" > >  > G >         Absolutely skilled.  One of the people is a 15 year veteran, i admin,I >         system configurator... one of the brighter guys you would know.r > ! >         Trust me... bad design.d  B All this example serves to illustrate is the power of the the UnixG command syntax.  And whoever coded the above example, regardless of hisoG years of systems management experience, is obviously not experienced ataG writing shell scripts.  Actually, it's doubtful that whoever wrote thataD had many months, let alone years of scripting experience. Any decentB shell scripter would NEVER, I repeat NEVER use that combination ofH syntax for just the exact reason of the variable being undefined.  Don'tH take my word for it. Post this example in any of the Unix newsgroups andE when they get done laughing maybe they'll respond if anyone can still># manage to type without cracking up.eH    Especially when it comes to text processing a Unix one-liner can takeA the place of a 20 line DCL procedure in some cases.  I have quitebG frequently needed to massage some textual data, and FTPed the file to afG Unix system to accomplish in seconds, what would have taken a half hourrA on VMS to do with DCL (because a whole procedure would have to be H written and debugged).  Piping in Unix is extremely powerful; that's whyG VMS has its poor-man's equivalent today.  One thing I would like to see G in VMS is a straight redirection symbol (">") without having to use theeG "PIPE" command in front of it.  Also, the ability to take the result ofyG any expression and assign it to a variable without having to go through1/ intermediate steps would also be very useful.  -   HM  , Return-path: <Info-VAX-Request@mvb.saic.com>< Received: from vaccine.tg.nsw.gov.au (vaccine.tg.nsw.gov.au  [203.32.225.150])-.  by tgmail.tg.nsw.gov.au (PMDF V5.2-32 #33154)I  with ESMTP id <01K7HAG2VY7G004A6U@tgmail.tg.nsw.gov.au> for obrien@GECKOe7  (ORCPT rfc822;paddy.o+27brien@zzz.tg.nsw.gov.au); Thu,s  23 Aug 2001 16:33:32 +10:000 Received: from scarab.tg.nsw.gov.au (unverified)=  by vaccine.tg.nsw.gov.au (Content Technologies SMTPRS 4.2.1)t@  with SMTP id <T558cad2835cb20e1960e7@vaccine.tg.nsw.gov.au> forC  <paddy.o'brien@zzz.tg.nsw.gov.au>; Thu, 23 Aug 2001 16:32:58 +1000yJ Received: by scarab.tg.nsw.gov.au; (5.65v4.0/1.3/10May95) id AA29211; Thu,  23 Aug 2001 16:45:00 +1000aH Received: from mvb.saic.com by scarab.tg.nsw.gov.au (smtpxd); id XA20741% Date: Thu, 23 Aug 2001 02:37:22 -0400a( From: Hamlyn Mootoo <univms@bigfoot.com>= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...)i To: Info-VAX@Mvb.Saic.Comf, Reply-to: Hamlyn Mootoo <univms@bigfoot.com>+ Message-id: <3B84A4A2.DAD78BE8@bigfoot.com>GJ Organization: Independent VMS and UNIX Systems Manager and Programmer for  Hire
  (consultant)  MIME-version: 1.00& X-Mailer: Mozilla 4.77 [en] (Win98; U), Content-type: text/plain; charset="us-ascii" Content-transfer-encoding: 7bit + Nntp-Posting-Date: 23 Aug 2001 06:32:27 GMToI X-Trace: UmFuZG9tSVaMc8r99sQdCeRj32+9/clKPtsCtICpj4CDLS5FdY9uwzx5abrcYNhSc X-Complaints-To: abuse@rcn.com X-Accept-Language: en  X-Gateway-From: mvb.saic.com	 Lines: 40e X-Gateway-source-info: USENETh X-Newsgroups: comp.os.vmsn: Original-recipient: rfc822;paddy.o'brien@zzz.tg.nsw.gov.au   ------------------------------  % Date: Thu, 23 Aug 2001 11:19:59 +0100l/ From: Nigel Arnot <sysmgr@maxwell.ph.kcl.ac.uk>n= Subject: re: Wailing at Eunuchs (was: Wailing and Moaning...)f6 Message-ID: <00A00F4E.025D0B42.4@maxwell.ph.kcl.ac.uk>  B >   Two skilled Unix "persons" making a mistake in a shell script.Q > > >>         Two high profile clients dead in the water for a day or two.  Why?g > > >>. > > >>         # rm -r ${variable_goes_here}/* > > >o= > > > Couldn't have been too "skilled" to make that mistake !v > > >e > > O > >         Absolutely skilled.  One of the people is a 15 year veteran, admin,vK > >         system configurator... one of the brighter guys you would know.  > > # > >         Trust me... bad design.o > D > All this example serves to illustrate is the power of the the Unix > command syntax.    [snip] >  > HM >   + Sorry, but I have to disagree *absolutely*.I  H What this illustrates is the fundamental good design that went into VMS.G By delimiting all directory specifications with [], and by calling the u6 top-level directory 000000 rather than <null string>, C many catastrophic errors are converted into harmless syntax errors.gN Another example is the mandatory version number on a delete command, making itE impossible for an intended DELETE *ABC.DEF;* to accidentally DELETE *UD when a CR is accidentally pressed or delivered by noise on the line.  A Note also that it's not a DCL/shell issue. It's more fundamental:gL VMS filenames have(had?) a restricted character set and a defined, parseableI syntax. Unix file names have neither.  A rogue program can create a file,:? or indeed a folder, using "*" or "/*" as the name. I wonder howyF many shell scripts can safely hack the processing thereof .... shades  of Alice's "DRINK ME" potion.n  E At a deeper level is the set of VMS privileges compared to the singleIF all-encompassing unix "root". Yes, it is true that having obtained oneE of many of the VMS privileges, you can use/abuse it to obtain all theIB others, and therefore the raft of privilege bits isn't primarily aJ defense against crackers. Rather, it's a Molly-guard, protecting you from  *un*intentional disaster.o  M Buffer overruns, and strings by descriptor rather than reference, are a third / much-discussed example of VMS getting it right.i  F Engineers know well that powerful buttons and levers must be protectedE by Molly-guards, interlocks, padlocks, and suchlike. The designers ofuI unix failed to appreciate this, and the above is a typical example of the  consequences.    	Yours,1
 		Nigel Arnotj- 		NRA@MAXWELL.PH.KCL.AC.UK                   g  7 		"In the beginning there was nothing, which exploded."C   ------------------------------  % Date: Thu, 23 Aug 2001 15:54:56 +0200o& From: Michael Joosten <joost@c-lab.de>= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...) $ Message-ID: <3B850B30.63DE@c-lab.de>   Hamlyn Mootoo wrote: >  > Rob Young wrote: > [snip]B >   Two skilled Unix "persons" making a mistake in a shell script.Q > > >>         Two high profile clients dead in the water for a day or two.  Why?. > > >>. > > >>         # rm -r ${variable_goes_here}/* > > >r= > > > Couldn't have been too "skilled" to make that mistake !  > > >o > >tO > >         Absolutely skilled.  One of the people is a 15 year veteran, admin,AK > >         system configurator... one of the brighter guys you would know.t > >s# > >         Trust me... bad design.l > D > All this example serves to illustrate is the power of the the UnixI > command syntax.  And whoever coded the above example, regardless of hiswI > years of systems management experience, is obviously not experienced atcI > writing shell scripts.  Actually, it's doubtful that whoever wrote thatuF > had many months, let alone years of scripting experience. Any decentD > shell scripter would NEVER, I repeat NEVER use that combination ofJ > syntax for just the exact reason of the variable being undefined.  Don'tJ > take my word for it. Post this example in any of the Unix newsgroups andG > when they get done laughing maybe they'll respond if anyone can still5% > manage to type without cracking up.j  C Right. This wasn't scripted, it was scribbled. Especially if it wascH written to be executed as root. A more skilled shell scripter would haveA used at least a check if the variable was set before hand or usedY conditional evaluation, like6    rm -rf ${variable_goes_here-"UNDEFINED_DIRECTORY!"}@ assuming of course that is no dir called /UNDEFINED_DIRECTORY! .    J > written and debugged).  Piping in Unix is extremely powerful; that's whyI > VMS has its poor-man's equivalent today.  One thing I would like to seetI > in VMS is a straight redirection symbol (">") without having to use thefI > "PIPE" command in front of it.  Also, the ability to take the result ofhI > any expression and assign it to a variable without having to go through / > intermediate steps would also be very useful.c  E Yes, exactly. While seeming awkward at first glance, pipes, sed, awk,sF regexp are extremelt usefull and powerfull. When I had to tinker whithE DCL while using at the same time UNIX, I was appalled by its ancient, E fortranesque (IV, not 77) control flow. I must say, though, that this- was with VMS 4.5 in '88...F OTOH, you cannot easly bring a shell script from one UNIX to an other,F as the exact functionality of ksh, sed and awk is sometimes different.C The best advise for someone who wants to build platform independant C scripting know-how is probably to use Perl for everything. And yes,n2 these F$ lexcial functions are also nice, I agree.  E I was just thinking about how autoconf could be adapted to VMS, and Iy> wondered how parsing the configure.in could be accomplished... -- d* Michael Joosten, SBS C-LAB, joost@c-lab.de* Fuerstenallee 11, 33094 Paderborn, Germany, Phone: +49 5251 606127, Fax: +49 5251 6060658 C-LAB is a cooperation of University Paderborn & SIEMENS   ------------------------------    Date: 23 Aug 2001 10:20:07 -0500+ From: young_r@encompasserve.org (Rob Young)s= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...)l3 Message-ID: <t$BwtiInwAdN@eisner.encompasserve.org>   M In article <3B850B30.63DE@c-lab.de>, Michael Joosten <joost@c-lab.de> writes:o > Hamlyn Mootoo wrote: >> u >> Rob Young wrote:l	 >> [snip]tC >>   Two skilled Unix "persons" making a mistake in a shell script.aR >> > >>         Two high profile clients dead in the water for a day or two.  Why? >> > >>a/ >> > >>         # rm -r ${variable_goes_here}/*- >> > >> >> > > Couldn't have been too "skilled" to make that mistake ! >> > > >> >P >> >         Absolutely skilled.  One of the people is a 15 year veteran, admin,L >> >         system configurator... one of the brighter guys you would know. >> >$ >> >         Trust me... bad design. >> )E >> All this example serves to illustrate is the power of the the UnixuJ >> command syntax.  And whoever coded the above example, regardless of hisJ >> years of systems management experience, is obviously not experienced atJ >> writing shell scripts.  Actually, it's doubtful that whoever wrote thatG >> had many months, let alone years of scripting experience. Any decentoE >> shell scripter would NEVER, I repeat NEVER use that combination of K >> syntax for just the exact reason of the variable being undefined.  Don't K >> take my word for it. Post this example in any of the Unix newsgroups andnH >> when they get done laughing maybe they'll respond if anyone can still& >> manage to type without cracking up. > E > Right. This wasn't scripted, it was scribbled. Especially if it wasgJ > written to be executed as root. A more skilled shell scripter would haveC > used at least a check if the variable was set before hand or usedU > conditional evaluation, like8 >    rm -rf ${variable_goes_here-"UNDEFINED_DIRECTORY!"}B > assuming of course that is no dir called /UNDEFINED_DIRECTORY! . >   = 	Notice I was careful enough to NOT say it was something thats@ 	could NOT be avoided.  But when you look at what took place, itD 	was the transposition of a single letter in one case and all drivesC 	were wiped out.  Good shell programming technique would have savednC 	the day, as you point out.  But what you are also doing is raisingyC 	the bar to a certain level (a level that may not always be reached G 	via an oversight or a rush job) that may not always be there.  TestinghB 	for the variable may have occured, can't recall.  You can test onB 	one line and make a typo on the next line... happens all the time
 	you know.  F 	Transposition of a single letter and all your drives are wiped clean.   	Bad design ... trust me.t   					Rob   ------------------------------  % Date: Thu, 23 Aug 2001 12:53:18 -0400a( From: Hamlyn Mootoo <univms@bigfoot.com>= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...)0+ Message-ID: <3B8534FE.CC397217@bigfoot.com>e   Nigel Arnot wrote: > D > >   Two skilled Unix "persons" making a mistake in a shell script.S > > > >>         Two high profile clients dead in the water for a day or two.  Why?w > > > >>0 > > > >>         # rm -r ${variable_goes_here}/* > > > > ? > > > > Couldn't have been too "skilled" to make that mistake !o > > > >  > > >eQ > > >         Absolutely skilled.  One of the people is a 15 year veteran, admin,aM > > >         system configurator... one of the brighter guys you would know.A > > >a% > > >         Trust me... bad design.  > >eF > > All this example serves to illustrate is the power of the the Unix > > command syntax.- > [snip] > >  > > HM > >2 > - > Sorry, but I have to disagree *absolutely*.  > J > What this illustrates is the fundamental good design that went into VMS.H > By delimiting all directory specifications with [], and by calling the7 > top-level directory 000000 rather than <null string>, E > many catastrophic errors are converted into harmless syntax errors.tP > Another example is the mandatory version number on a delete command, making itG > impossible for an intended DELETE *ABC.DEF;* to accidentally DELETE *eF > when a CR is accidentally pressed or delivered by noise on the line. > , > Note also that it's not a DCL/shell issue.  = Wrong.  There is nothing requiring you to use the DCL commandnC interpreter on VMS, nor is there anything requiring you to use RMS!-D Ergo, you can use ANY characters you wish to name a file without theG burden of RMS. That's WHY they're called shells.  Similarly on Unix, nol6 one forces the user to run ANY of the standard shells.    It's more fundamental:tN > VMS filenames have(had?) a restricted character set and a defined, parseableK > syntax. Unix file names have neither.  A rogue program can create a file, A > or indeed a folder, using "*" or "/*" as the name. I wonder howlG > many shell scripts can safely hack the processing thereof .... shades  > of Alice's "DRINK ME" potion.e > G > At a deeper level is the set of VMS privileges compared to the singlehH > all-encompassing unix "root". Yes, it is true that having obtained oneG > of many of the VMS privileges, you can use/abuse it to obtain all themD > others, and therefore the raft of privilege bits isn't primarily aK > defense against crackers. Rather, it's a Molly-guard, protecting you fromt > *un*intentional disaster.   > Any VMS user holding just the BYPASS privilege can do the sameH destruction to the file system as can "root" under Unix.  And since mostF programmers on VMS don't take the time to make sure their applicationsF can be installed with privs so that non-priv'ed users can just do whatG they're supposed to, an incredibly LARGE percentage of user accounts onbE VMS systems are running around with near full privs, or at least ones A that will allow their accounts to get the rest with minimal work.e   > O > Buffer overruns, and strings by descriptor rather than reference, are a thirdt1 > much-discussed example of VMS getting it right.h > H > Engineers know well that powerful buttons and levers must be protectedG > by Molly-guards, interlocks, padlocks, and suchlike. The designers ofbK > unix failed to appreciate this, and the above is a typical example of the  > consequences.m  F If you knew the history or Unix from its Multics roots, you would know& that they appreciated this quite well!   >  >         Yours, >                 Nigel Arnoth* >                 NRA@MAXWELL.PH.KCL.AC.UK > G >                 "In the beginning there was nothing, which exploded."s   ------------------------------    Date: 23 Aug 2001 12:07:28 -0500+ From: young_r@encompasserve.org (Rob Young)A= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...)a3 Message-ID: <4bRpgYd83Wot@eisner.encompasserve.org>   V In article <3B8534FE.CC397217@bigfoot.com>, Hamlyn Mootoo <univms@bigfoot.com> writes: > @ > Any VMS user holding just the BYPASS privilege can do the sameJ > destruction to the file system as can "root" under Unix.  And since mostH > programmers on VMS don't take the time to make sure their applicationsH > can be installed with privs so that non-priv'ed users can just do whatI > they're supposed to, an incredibly LARGE percentage of user accounts onaG > VMS systems are running around with near full privs, or at least ones.C > that will allow their accounts to get the rest with minimal work.o >   : 	There are two answers to this that point a finger back to< 	poor administration.  While it is true protections may leadA 	one to the easier out of doling out BYPASS, the best solution isy@ 	to set protections if you can but sharing across groups may not@ 	allow this.  The second way if setting of global protections is8 	an issue is to place and grant ACLs.  I have done both.  G 	However, at one time, I came into a situation that was totally out of dG 	control, users had BYPASS.  Looking at undoing it meant weeks of work eA 	and tedium.  I left it as it was.  I wasn't worried as all user tB 	accounts were CAPTIVE.  No sweat.  My intention was if I ever gotA 	around to it (nothing of a higher priority) was to throw ACLs ate( 	the problem.  I never got around to it.  < 	To suggest there are large numbers of VMS systems out thereC 	with BYPASS for users and they can get to a $ sign is nonsense andfA 	a tangent to the fact that Unix's hiearchial filesystem is a bad  	design.   >> nP >> Buffer overruns, and strings by descriptor rather than reference, are a third2 >> much-discussed example of VMS getting it right. >> nI >> Engineers know well that powerful buttons and levers must be protectedPH >> by Molly-guards, interlocks, padlocks, and suchlike. The designers ofL >> unix failed to appreciate this, and the above is a typical example of the >> consequences. > H > If you knew the history or Unix from its Multics roots, you would know( > that they appreciated this quite well! >   = 	Eh?  What does history have to do with allowing poor design?F 	Elucidate, please.t   				Rob<   ------------------------------  % Date: Thu, 23 Aug 2001 13:51:19 -0400p- From: JF Mezei <jfmezei.spamnot@videotron.ca>u= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...)v, Message-ID: <3B854295.B71E9583@videotron.ca>   Nigel Arnot wrote:H > Engineers know well that powerful buttons and levers must be protectedG > by Molly-guards, interlocks, padlocks, and suchlike. The designers ofvK > unix failed to appreciate this, and the above is a typical example of the  > consequences.0  N Perhaps the VMS command "INIT" might be given some molly-guards ? Even on DOS,C FORMAT would prompt you if you are really sure you want to do this.o  F So, UNIX has some commands that can easily wipe out your data, and theL equivalent commands on VMS require more typing. So, it is a little less easyM on VMS to zap your data. But everyone knows that you should only use your allSN mighty privileges when absolutely necessary and the rest of the time, use onlyN normal user privileges. (well, that is *supposed* to be like that, right ? :-)  L And a good system manager wouldn't really try an untested program or commandI file without taking a look at it , or first testing it on an unprivilegedh account, would he ?a   ------------------------------  % Date: Thu, 23 Aug 2001 13:41:07 -040045 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com>o= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...).2 Message-ID: <yebh7.672$bB1.31397@news.cpqcorp.net>  A Hamlyn Mootoo wrote in message <3B8534FE.CC397217@bigfoot.com>...a >k > ? >Any VMS user holding just the BYPASS privilege can do the sameeI >destruction to the file system as can "root" under Unix.  And since mostxG >programmers on VMS don't take the time to make sure their applicationsrG >can be installed with privs so that non-priv'ed users can just do what H >they're supposed to, an incredibly LARGE percentage of user accounts onF >VMS systems are running around with near full privs, or at least onesB >that will allow their accounts to get the rest with minimal work. >     H Remind me not to work on any of your systems.  Except for the standaloneJ systems that I self manage, I neither have privledges, or have them turnedI on if I do, unless I need them.  I would venture to say that outside of a J development environment, most production VMS shops give the minimal set of< required privledges to users other than the system managers.  J Installing random images written by sloppy programmers sounds dangerous onI any OS.  Especially when all the facilities exist to allow them to do thed# right thing (unlike, say, Windows).e   ------------------------------    Date: 23 Aug 2001 12:24:51 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen)*= Subject: Re: Wailing at Eunuchs (was: Wailing and Moaning...)r3 Message-ID: <u3lZKctSo8MC@eisner.encompasserve.org>'  V In article <3B8534FE.CC397217@bigfoot.com>, Hamlyn Mootoo <univms@bigfoot.com> writes:  I > they're supposed to, an incredibly LARGE percentage of user accounts onAG > VMS systems are running around with near full privs, or at least ones C > that will allow their accounts to get the rest with minimal work.o  ! Not on systems with which I deal.e   Your mileage may vary.   ------------------------------   End of INFO-VAX 2001.468 ************************