1 INFO-VAX	Thu, 06 Jul 2000	Volume 2000 : Issue 374       Contents: %IMGACT-F-SYMVECMIS ???? Re: %IMGACT-F-SYMVECMIS ????/ Re: BACKUP under VMS 7.2-1 and ODS5 file system / Re: BACKUP under VMS 7.2-1 and ODS5 file system ' better tool to indicate magtape status? + Re: better tool to indicate magtape status? ! Re: CD burners for OpenVMS Alpha? " Re: CMQ/Dec with a sense of Humor? Re: DCL vs. "real" programming Re: DCL vs. "real" programming DEC C Question *variable ++  Re: DEC C Question *variable ++ < Re: Directory Sizes (was: good news (for me,  I think) . . .' Re: Multilingual System Dictionary(ies) ' Re: Multilingual System Dictionary(ies) ' Re: Multilingual System Dictionary(ies) ( Re: Netscape Fasttrack for OpenVMS 7.2-12 Re: OpenVMS loses big, was:  RE: Compaq advertises2 Re: OpenVMS loses big, was:  RE: Compaq advertises2 Re: OpenVMS loses big, was:  RE: Compaq advertises1 Re: OpenVMS loses big, was: RE: Compaq advertises % Re: RMS indexed file instatly growing % Re: SCSI hard drive 4 VAX 4700 needed 1 Re: Selling software to corporation AND hobbyists  Re: Trapping a STOP/ID Re: Trapping a STOP/ID Re: Trapping a STOP/ID Re: Trapping a STOP/ID Re: Trapping a STOP/ID Re: Trapping a STOP/ID Re: Tutorial1 Re: Upgrade to VMS 7.2 fails on Help Message File  Re: Very Large RMS Files Re: Very Large RMS Files VMS employment Re: VMS employment We need VMS people Re: We need VMS people Re: We need VMS people RE: We need VMS people0 Re: What is an ACP? (was Re: Trapping a STOP/ID)  F ----------------------------------------------------------------------  # Date: Thu, 06 Jul 2000 03:51:05 GMT  From: rcyoung@groupz.net! Subject: %IMGACT-F-SYMVECMIS ???? ) Message-ID: <8k0vn7$92d$1@nnrp1.deja.com>   C %IMGACT-F-SYMVECMIS, shareable image's symbol vector table mismatch   E Ran into this error on an Alpha when running 3rd party software. Have F other copies of software running on other systems, both Vax and Alpha,- and have never run into it before. Any ideas?     & Sent via Deja.com http://www.deja.com/ Before you buy.    ------------------------------  % Date: Wed, 05 Jul 2000 21:18:03 -0700 ! From: Shane.F.Smith@Healthnet.com % Subject: Re: %IMGACT-F-SYMVECMIS ???? C Message-ID: <OF0C69CADF.C73DF668-ON88256914.00172606@HEALTHNET.COM>   K An RTL the software's calling isn't the version it's expecting. How depends J on whether the RTL is part of the 3rd party software, in which case it's aG messed up installation, or it's a system RTL. If it's a system RTL this F probably means the software was linked on a later version of VMS, or a' machine with different patches applied.   ? What's the full text of the error, it should give the RTL name.    Shane           , rcyoung@groupz.net on 07/05/2000 08:51:05 PM   To:   Info-VAX@Mvb.Saic.Com  cc:   " Subject:  %IMGACT-F-SYMVECMIS ????    C %IMGACT-F-SYMVECMIS, shareable image's symbol vector table mismatch   E Ran into this error on an Alpha when running 3rd party software. Have F other copies of software running on other systems, both Vax and Alpha,- and have never run into it before. Any ideas?     & Sent via Deja.com http://www.deja.com/ Before you buy.    ------------------------------   Date: 5 Jul 2000 19:15:25 GMT 2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)8 Subject: Re: BACKUP under VMS 7.2-1 and ODS5 file system6 Message-ID: <8k01gd$evv$1@mailint03.im.hou.compaq.com>  N In article <QNndL5foE3Lt@belo>, "Bernie Milo" <milosavljevi@usp.ac.fj> writes:J :I have just found out that our VMS 7.2-1 system is not doing incremental E :backups on my ODS5 file system files. Basically, it is ignoring the   :/since=backup.  : : :Is this a know problem and if so, how do I get around it?     Not yet clear...  3   What is the exact BACKUP command being used here? ?   What are you seeing happen when the /SINCE=BACKUP is ignored? =   Have you applied any of the available ECO kits to V7.2-1?     N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  $ Date: Wed, 5 Jul 2000 11:11:12 +1200, From: "Bernie Milo" <milosavljevi@usp.ac.fj>8 Subject: Re: BACKUP under VMS 7.2-1 and ODS5 file system Message-ID: <OWbuZrLctAVH@belo>   K A follow-up. I did a backup on the top level directory with /record and the O /since=backups now appear to be working properly on lower level directories and  files.   Bernie.     O Bernie Milo <milosavljevi@usp.ac.fj> wrote in message news:QNndL5foE3Lt@belo...  > Hi, J > I have just found out that our VMS 7.2-1 system is not doing incremental backups L > on my ODS5 file system files. Basically, it is ignoring the /since=backup. > ; > Is this a know problem and if so, how do I get around it?  > 	 > Bernie.  >  >    ------------------------------   Date: 5 Jul 2000 18:02:15 GMT 2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)0 Subject: better tool to indicate magtape status?, Message-ID: <8jvt77$583@gap.cco.caltech.edu>  F Has anybody got a tool which will give a better indication of magtape ; status (for a DLT) on OpenVMS?  Right now I'll I've got is     $ show dev/full mkb500  H and that doesn't give me any idea at all what the tape position is.  TheB Quantum DLT4000 on our system sounds and looks the same when it isK rewinding as it does when it is streaming forward, and I can't see the tape 6 roll for even an eyeball estimate of what's going on.   G If nothing else, I'd at least like to know for sure which direction the  tape was moving!     Thanks,    David Mathog mathog@seqaxp.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech     ------------------------------  % Date: Wed, 05 Jul 2000 19:28:33 -0500 7 From: "David J. Dachtera" <djesys.nospam@earthlink.net> 4 Subject: Re: better tool to indicate magtape status?- Message-ID: <3963D2B1.259EF6E3@earthlink.net>    David Mathog wrote:  > G > Has anybody got a tool which will give a better indication of magtape < > status (for a DLT) on OpenVMS?  Right now I'll I've got is >  > $ show dev/full mkb500 > J > and that doesn't give me any idea at all what the tape position is.  TheD > Quantum DLT4000 on our system sounds and looks the same when it isM > rewinding as it does when it is streaming forward, and I can't see the tape 7 > roll for even an eyeball estimate of what's going on.  > I > If nothing else, I'd at least like to know for sure which direction the  > tape was moving!  H I can't imagine that such info. is even available from teh drive without& getting into the realm of IO$DIAGNOSE.  ( Have you tried http://www.quantum.com/ ?   --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/    ------------------------------  % Date: Wed, 05 Jul 2000 20:40:02 -0500 % From: Chris Scheers <asi@airmail.net> * Subject: Re: CD burners for OpenVMS Alpha?O Message-ID: <3C84AF76276C38CF.AC6B81F3F2A8C28A.058CCF60DA22AA6C@lp.airnews.net>   J > FYI, even with a second *NEW* disk installed in the toy box, Weendoze'98I > would not install.  It saw the Weendoze NT disk and refused to install. F > The disk was physically disconnected to get it to install.  There isH > still some braindamage as Weendoze'98 insists that D: is a CDrom when, > in fact, F: is the CD. :(   G Some MS CDs are meant for use with new systems and do a check to see if F you already have Windows installed.  If they find Windows, they refuse to install.   C I don't know if any of the Win98 CDs are like this, but some of the  Win95 CDs are.  A For the CDROM drive letter, there is no "in fact", just differing F opinions.  <grin>  In fact, you can set the drive letter to any unused: letter.  I set it to H: on both my Windows and NT systems.  E For Win98, go to Control Panel =>  System.  Select the device manager E tab and then find the CDROM drive.  Select its properties page and at 7 the bottom of the page you can change the drive letter.   : For WinNT, use Administrative Toold => Disk Administrator.  F What probably happened is that you installed Win98 with the other diskG disconnected.  Win98 found your first disk (C:) and made the CDROM D:.  ? Then you connected the other drive where Win98 found additional G partitions making them E:, etc. since D: was already used.  (That's why E I set the CDROM to H:.  It gives me room to add additional partitions " without things getting too weird.)  
 Good luck!  G FWIW:  I run MSDOS 6.20, Win 3.11, Win98, WinNT 3.51, and WinNT 4.0 all 5 off the same disk.  It's messy, but usefull at times.   G ----------------------------------------------------------------------- $ Chris Scheers, Applied Synergy, Inc.  G 817-237-3360 (Voice)    817-237-3074 (Fax)    Internet: asi@airmail.net    ------------------------------  # Date: Wed, 05 Jul 2000 21:47:29 GMT % From: Joel Garry <jgarry@my-deja.com> + Subject: Re: CMQ/Dec with a sense of Humor? ) Message-ID: <8k0ade$rv9$1@nnrp1.deja.com>   , In article <39488701.2259.14D557@localhost>,)   Ken Robinson <ksrobin@erenj.com> wrote: / > On 14 Jun 2000, at 19:42, Dan O'Reilly wrote:  > F > > I don't know if VMS Datatrieve had this message, but Datatrieve-11 had E > > an error message, in response to a user trying "AT MIDDLE OF PAGE  > > <dosomething>"H > > that was "AT MIDDLE OF PAGE is innovative, but, alas, illegal".  One of my G > > all-time faves.  Along with, of course, "LINK-F-Illegal Error" from  the  > > RT-11 linker...  > C > Speaking of Datatrieve... Does anybody remember what you got when B > you typed "Help wombat" when you were in Datatrieve and you wereE > using a terminal that recognized REGIS graphics???  The last time I  tried > > the command, at a recent Decus Symposium, it was gone... :-(  C Whoa, I never had anything better than a CIT-101 until I got a 241. 9 Wasn't that help a line drawing of the Australian animal?   H Speaking of Datatrieve and humor, I always thought the syntax for nestedH queries was kind of bizarre, like FIND FIND FIND FIND ALL ALL ALL ALL...D but I wound up using other languages before I could delve too deeplyC into that... does anyone know where I might find online examples of , ridiculously complicated Datatrieve queries?   >  > Ken Robinson > ksrobin@erenj.com  >  >    jg -- These opinions mine ! mailto:joel-garry@nospam.home.com  Remove nospam to mail 3 http://ourworld.compuserve.com/homepages/joel_garry     & Sent via Deja.com http://www.deja.com/ Before you buy.    ------------------------------   Date: 5 Jul 2000 18:15:30 GMT 2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)' Subject: Re: DCL vs. "real" programming 6 Message-ID: <8jvu02$e18$2@mailint03.im.hou.compaq.com>   helbig@astro.rug.nl writes... H :For example, when writing an application which processes a lot of MAIL C :files, would one achieve much better performance by writing it in  C :Fortran or whatever and using the callable MAIL interface than by   :writing .COM files etc in DCL?   ?   The answer is, of course, the deliberately ambiguous "yes".     G   There are always trade-offs.  Each application and each customer has  C   its own desired/appropriate/expected performance -- there is the  A   time-to-run performance, the time-to-solution performance, the  B   time-to-fix (support performance), and various other things all C   interchangeably lumped under the heading of "better performance".   B   I cover some of the trade-offs involved in the use of DCL in theD   introduction of the Writing Real Programs in DCL book -- DCL is a ?   good choice for quick-to-implement (and potentially somewhat  F   inefficient use of system resources) solutions, and not so good for F   other classes of problems, or for situations where system resources    (or time) is tight.     C   The age-old interpreter vs compiler discussions, and the age-old  B   "good" vs "good enough" discussions...  Folks have spent days orD   weeks "bumming" instructions out of certain code paths -- the realD   question (at least in the commercial space) involves the expected F   payback...  Payback in the open source and hobbyist spaces often hasC   other ways of gaining (potentially different sorts of) payback...   N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------   Date: 5 Jul 2000 15:22:24 PDT T From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515)' Subject: Re: DCL vs. "real" programming 3 Message-ID: <CM5HKOuMNtxQ@mccdev.slac.stanford.edu>   / In article <8job8r$j8u$1@info.service.rug.nl>,  1     	helbig@astro.rug.nl (Phillip Helbig) writes:  [...] I > It's not a one-shot (for which I would certainly use DCL), so it looks  I > like it's F$ time.  This reminds me, someone once mentioned here a DCL  I > compiler which would take a .COM file and convert it to the equivalent  G > FORTRAN with system calls etc and compile that.  Who can point me to   > this?   H         There is (or  was)  a  company  called Channel Islands Software,H     really  just one guy, who had a product that did this.  I can't  forH     the life of me  remember  its  name.   Something  like  DCLFort?   IH     dunno...   It  did,  in  fact,  read  DCL  command  files and outputH     Fortran.  It was a bit pricey, like  US$900 or so for a single smallH     system.   Something  for the "big boys" who had huge DCL  procedures     that needed to be sped up.  H         I talked to the  owner/programmer  on  the  phone a time or two.H     One  of  the  issues  was that the source code  it  output  was  notH     particularly "maintainable".  Obfuscated I  take  it.   Another  wasH     that  you  needed  the  product's  RTL,  which provided the callableH     replacements for all  the  lexical  functions  that aren't otherwiseH     available  (of  course).  In other words, there was a lot more  thanH     just calls to VMS RTL and System Services routines.  At the  time  IH     couldn't justify the cost (I mostly wanted to just try it out, but aH     "demo"  was  really  available).   I  think there may have been someH     issues of keeping the product up  to  date with VMS releases as well*     (i.e., adding the new lexicals, etc.).  H         If there's interest, I'll see if  I  still have a flier or otherA     reference information, but I expect it's a long-dead product.            -Ken --  M  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edu :  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------  % Date: Wed, 05 Jul 2000 22:05:18 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> $ Subject: DEC C Question *variable ++, Message-ID: <3963E95E.E16E1E80@videotron.ca>   myfunction (int *length) {  *length ++ ; return ; }    or   myfunction (int *length) {  *length = *length + 1 ;  return ; }   A Is there a reason why only the second variant works as expected ?   + What does  " *length ++ "  supposed to do ?      DEC-C 6.0, VAX (VMS 7.2).    ------------------------------  % Date: Wed, 05 Jul 2000 20:08:46 -0700 5 From: "Larry D Bohan, Jr" <LBohan@dbc.spam_less..com> ( Subject: Re: DEC C Question *variable ++2 Message-ID: <7=ZjOYvbEAf7V8uNA1WxpwnhdG2k@4ax.com>  , On Wed, 05 Jul 2000 22:05:18 -0400, JF Mezei% <jfmezei.spamnot@videotron.ca> wrote:   $ in the first example, try instead:       (*length)++    >myfunction (int *length)  >{
 >*length ++ ; 	 >return ;  >} >  >or  >  >myfunction (int *length)  >{   >*length = *length + 1 ;	 >return ;  >} > B >Is there a reason why only the second variant works as expected ? > , >What does  " *length ++ "  supposed to do ? >  >  >DEC-C 6.0, VAX (VMS 7.2).   ------------------------------  % Date: Wed, 05 Jul 2000 18:30:05 -0500 7 From: "David J. Dachtera" <djesys.nospam@earthlink.net> E Subject: Re: Directory Sizes (was: good news (for me,  I think) . . . - Message-ID: <3963C4FD.E5EF8F4A@earthlink.net>    Jan Vorbrueggen wrote: > ; > "David J. Dachtera" <djesys.nospam@earthlink.net> writes:: > H > > Very true! ...and for those OpenVMS and VAX/VMS versions which don'tK > > support CREATE/DIRECTORY/ALLCOATION=nnnn, I have a DCL proc. which will D > > pre-allocate directories so they are not (seriously) fragmented. > @ > 'scuse me - don't ODS2 directory files have to be contiguous!?  F Quite true! I was actually thinking about cases where directories needF to be pre-allocated, such as when restoring a volume-set from magtape.   F > Of course, pre-allocation is still a good idea in certain scenarios.   Agreed!e   -- o David J. Dachterat dba DJE Systemsi http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/i   ------------------------------   Date: 5 Jul 2000 19:58:16 GMTm2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)0 Subject: Re: Multilingual System Dictionary(ies)6 Message-ID: <8k040o$ge9$1@mailint03.im.hou.compaq.com>  o In article <000301bfe68a$1ed89910$0457c80a@pc2860.lanprd.nbb>, Nichele Claudio <claudio.nichele@nbb.be> writes:DG :I would like to use on all our production systems the french and dutch @ :System Dictionaries in conjunction with the standard dictionary4 :available SYS$LIBRARY:VMS$PASSWORD_DICTIONARY.DATA.C :But i am unaware of how to do it and the official documentation is  :poor in informations.  E   I do not believe that this modification is documented or supported.eD   (eg: if you try it, your changes might get clobbered during an ECOD   or OpenVMS upgrade.)  I don't believe it's particularly difficult.  D :Are System Dictionaries for other languages than english available?     Not that I am aware of.D  A   As for details on how to add in your own password policies intoa$   OpenVMS, please see the following:  5     http://www.openvms.digital.com:8000/72final/5841/f-       5841pro_067.html#create_share_image_seco  D   The VMS$PASSWORD_POLICY module hooks are documented and supported.A   You can use these hooks to check the password for permitted (oruE   excluded) characters, words, or phrases -- adding a simple password8E   plaintext filter based on soundex could be a really slick approach.SC   (eg: contains numerics or dollar signs, or has no good match whenH%   run through the soundex algorithm.)   B   There are examples of the password policy module on OpenVMS VAX,D   in the directory SYS$EXAMPLES:.  (Offhand, I do not know why these>   example modules did not get ported across to OpenVMS Alpha.)  E :The worst approach is to merge all dictionaries in an unique and big ! :.DATA file with all words mixed.i  G   Having a collection of words from a variety of languages incorporated 9   into one monolithic file would be (bluntly) unworkable.   J   Having a second file (using a simple format) containing a site-specific I   list of prohibited passwords is a reasonable request.  (That, or adding J   support for adding to the prohibited words list in the existing password(   dictionary, on a site-specific basis.)  ( :We are using OpenVMS V7.1 and V7.1-1H2.  H   Please move to OpenVMS Alpha V7.1-2 or V7.2-1, or to OpenVMS VAX V7.1 
   or V7.2.  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------   Date: 5 Jul 2000 17:27:21 -0500m9 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen)m0 Subject: Re: Multilingual System Dictionary(ies)+ Message-ID: <k9oJ2yIOKzMm@eisner.decus.org>n  k In article <8k040o$ge9$1@mailint03.im.hou.compaq.com>, hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) writes:h  D >   There are examples of the password policy module on OpenVMS VAX,F >   in the directory SYS$EXAMPLES:.  (Offhand, I do not know why these@ >   example modules did not get ported across to OpenVMS Alpha.)  F I know someone who complained about that, and she was told the absence> of the examples from Alpha would be fixed in a future release.   ------------------------------  % Date: Wed, 05 Jul 2000 16:40:27 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca>J0 Subject: Re: Multilingual System Dictionary(ies), Message-ID: <39639D2B.B224B2F2@videotron.ca>   Hoff Hoffman wrote: G >   I do not believe that this modification is documented or supported.nF >   (eg: if you try it, your changes might get clobbered during an ECOF >   or OpenVMS upgrade.)  I don't believe it's particularly difficult.  M since the file is essentially a fixed length sequential file, isn't it just a@M matter of adding your records (padded with nulls, I beleive) and then sortingg
 the file ?  K Would it not be possible for the VMS group to support this method, with theeJ caveat that customers must keep a separate copy of their additions and add0 them back after every new version installation ?  I It would be an easy way for a system manager to add company name, name of F president, name of popular secretarioes etc etc to the dictionary (eg: localised stuff).a   ------------------------------  # Date: Wed, 05 Jul 2000 18:51:09 GMTi' From: Colin Blake <colin@theblakes.com> 1 Subject: Re: Netscape Fasttrack for OpenVMS 7.2-1v- Message-ID: <3963839B.EF5BD671@theblakes.com>e  J You might want to try enabling and applying "CGI File Type", just in case.   ------------------------------   Date: 5 Jul 2000 15:06:52 -0500m9 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen)e; Subject: Re: OpenVMS loses big, was:  RE: Compaq advertises + Message-ID: <ac3EBx40XvkJ@eisner.decus.org>h  Q In article <8jvs8j$g4$1@pyrite.mv.net>, "Bill Todd" <billtodd@foo.mv.com> writes:a  K > This observation may be reasonable with respect to wording of the post to-L > which it responded, but should not be taken as any indication that defaultM > performance using RMS does not have similar problems - since it does indeedgF > have very similar problems.  They can be alleviated significantly byL > explicitly taking advantage of optional RMS features, but the result stillK > often falls considerably short of the performance one can achieve on Unix K > just using default settings (and without sacrifice in integrity, since inaM > both environments judicious use of $FLUSH or fsync is required at any pointtL > where a guarantee that written data is actually on the disk is important).  G Certainly "default settings" cannot serve all.  But sometimes I programnG for my 10 MB MicroVAX II, and sometimes I program for customer machinestF of unknown memory size, so resource consumption must always be part ofJ the design for me.  I have heard of someone who ordered a Wildfire becauseI they had money in the budget (when perhaps they could have used a smallerJ	 machine).   E Believe it or not, some programs are actually written where the humans, rather than the disk is the limiting factor.  F While I am sure there are cases where default performance on some UnixG is better than tuned performance on VMS, I am also sure there are casesnH (not necessarily disk-related) where the reverse is true.  So DEQ shouldI make improvements to VMS limitations, but changing defaults is a cosmetic G change that should not be used to cover up a need for real improvement.n   ------------------------------  % Date: Wed, 05 Jul 2000 22:33:55 -0400 ) From: yyyc186.illegaltospam_@flashcom.nete; Subject: Re: OpenVMS loses big, was:  RE: Compaq advertisesF9 Message-ID: <3963f11b$2$lllp186$mr2ice@news.flashcom.com>s  0 In <8jvmhh$b7@gap.cco.caltech.edu>, on 07/05/00 B    at 10:33 PM, mathog@seqaxp.bio.caltech.edu (David Mathog) said:    H >For a UNIT text file operation, going RAMDISK to RAMDISK on OpenVMS, orJ >file cache to file cache on Linux, on identical DS10s, the result is thatI >the Linux system is 2.5-6.5 times faster.  This is for an operation like I >"read text record from input, write text record to output" - pure IO. It B >doesn't matter if you do this in 1 file or in 1000 you're already@ >starting out with the Unix systems "lighter" text file handlingG >mechanisms 3X faster than those on OpenVMS.  And it goes downhill frommC >there, rapidly, because the lack of file effective file caching on   F Sure, for accademia speed will always win out over integrity.  You areH overlooking the stability FILES 11 gives you over the unix disk stream. D In the real world, where integrity is king, we want to know the dataC actually made it to the disk (hopefully into a nice record orientedgI indexed file and not some fprintf() stream) when the call returns.  WritedG caching may seem like a wonderfull thing, but it makes check pointing a 
 nightmare.   Roland     -- e; ------------------------------------------------------------D yyyc186@flashcom.net              To Respond delete ".illegaltospam"6                             MR/2 Internet Cruiser 1.528                             For a Microsoft free univers; -----------------------------------------------------------    ------------------------------  % Date: Wed, 05 Jul 2000 22:39:46 -0400f) From: yyyc186.illegaltospam_@flashcom.nete; Subject: Re: OpenVMS loses big, was:  RE: Compaq advertisese9 Message-ID: <3963f18e$3$lllp186$mr2ice@news.flashcom.com>t  0 In <LoKsvv2fycO9@eisner.decus.org>, on 07/05/00 I    at 10:39 PM, Kilgallen@eisner.decus.org.nospam (Larry Kilgallen) said:s  I >Try some tests in Ada, Pascal or Fortran, or else make it quite clear in 6 >your strident posts that yours is a C-only viewpoint.  F Actually on VMS  BASIC seems to be the most closely mated to RMS (sans Macro of course).o   Roland   -- 1; -----------------------------------------------------------mD yyyc186@flashcom.net              To Respond delete ".illegaltospam"6                             MR/2 Internet Cruiser 1.528                             For a Microsoft free univers; -----------------------------------------------------------    ------------------------------  % Date: Wed, 05 Jul 2000 21:59:08 -0700 ! From: Shane.F.Smith@Healthnet.coma: Subject: Re: OpenVMS loses big, was: RE: Compaq advertisesC Message-ID: <OFD316202B.029F77DD-ON88256914.0015CA60@HEALTHNET.COM>N  H Sadly, this is not usually how things happen. In the real world speed is1 often king because PHMs don't get the advantages:k  K PHM: Somebody just told me Unix's file system is faster than VMS's. Is that  true??I Tech:     Well, yes, it's about x% faster, but VMS's filesystem is safer.o= PHM: So if we migrate to Unix, everything will run x% faster?mK Tech:     That depends on how much IO we're doing, but it might corrupt our. files.? PHM: Everybody else seems happy with it, how much does it cost?-J Tech:     Well, depending on which one, it's somewhere between free and z% cheaper than VMS, but....DF PHM: Will it run on that lovely, cheap, fast Intel chip I keep hearing about, the Puntiam?a4 Tech:     Pentium. Well, yes but the Alpha's faster. PHM: What's it's clock speed?o  Tech:     Somewhere in the 800'sI PHM: That's not as fast as the Pantyloom, I've heard that goes at 1000mhziK Tech:     /Pentium/! Alpha's a different architecture type, called RISC, so  the mhz doesn't....oE PHM: Alpha's a risk? I want a plan on my desk by tomorrow morning foro migrating to Unix on Pinto.c  F The trouble is, their eyes glaze over when you try to explain anythingC vaguely technical, and all VMS's advantages are technical. It's notaH particularly pretty, it's not instinctive to someone used to Billyboxes,G it's not advertised and raved about all over management magazines or on J television, and when they have heard the name VMS anytime in the last fiveE years it's been immediately prefixed with either "legacy" or "migratedJ from". It's going to take a lot of time, effort and money to turn all that around.e   Shanee          = yyyc186.illegaltospam_@flashcom.net on 07/05/2000 07:33:55 PMi   To:   Info-VAX@Mvb.Saic.Com  cc:o  ; Subject:  Re: OpenVMS loses big, was: RE: Compaq advertisesi    / In <8jvmhh$b7@gap.cco.caltech.edu>, on 07/05/00hB    at 10:33 PM, mathog@seqaxp.bio.caltech.edu (David Mathog) said:    H >For a UNIT text file operation, going RAMDISK to RAMDISK on OpenVMS, orJ >file cache to file cache on Linux, on identical DS10s, the result is thatI >the Linux system is 2.5-6.5 times faster.  This is for an operation likeeI >"read text record from input, write text record to output" - pure IO. ItiB >doesn't matter if you do this in 1 file or in 1000 you're already@ >starting out with the Unix systems "lighter" text file handlingG >mechanisms 3X faster than those on OpenVMS.  And it goes downhill fromrC >there, rapidly, because the lack of file effective file caching on:  F Sure, for accademia speed will always win out over integrity.  You areG overlooking the stability FILES 11 gives you over the unix disk stream.@D In the real world, where integrity is king, we want to know the dataC actually made it to the disk (hopefully into a nice record orientedtI indexed file and not some fprintf() stream) when the call returns.  WritehG caching may seem like a wonderfull thing, but it makes check pointing ah
 nightmare.   Roland     --; -----------------------------------------------------------rD yyyc186@flashcom.net              To Respond delete ".illegaltospam"6                             MR/2 Internet Cruiser 1.528                             For a Microsoft free univers; -----------------------------------------------------------C   ------------------------------  $ Date: Wed, 5 Jul 2000 14:02:59 -0400' From: "Bill Todd" <billtodd@foo.mv.com> . Subject: Re: RMS indexed file instatly growing( Message-ID: <8jvt5l$12f$1@pyrite.mv.net>  8 JF Mezei <jfmezei.spamnot@videotron.ca> wrote in message& news:39636EC6.BD154BCE@videotron.ca... > Randy Park wrote:u3 > > Several things you should understand about RMS:nD > > 1.  You will have less bucket splitting, and thus a smaller fileB > > size, if your records are added in primary key sequence order.G > > 2.  Even after you delete a record, the RFA (Record's File Address)l9 > > or the retrieval pointer for the record still exists.nC > > 3.  The CONVERT/RECLAIM command will recover space in the file.m >IG > The way it was explained to me is that if you delete all records in aa bucketJ > before it needs to be split, then the bucket becomes available again and  > convert/reclaim is not needed.  J Then you must have missed my post explaining that there is indeed detritusK left over after records are deleted - definitely deleted record RRVs in PV1 L and PV2 files, and at least used-up record IDs (I'm not sure about the RRVs)G in PV3 files which eventually leave a bucket incapable of accepting newS2 records (because no IDs are left to use for them).  I I'm not even sure whether RMS can dynamically remove such 'empty' buckets5J from their current location and use them elsewhere in its b-tree structureA (I know it was planned at one point, but can't remember if it wasOL implemented), or whether this is what CONVERT/RECLAIM does while the file is	 off-line.   +  But if you don't, then unused space in thecH > bucket (deleted records) is not made available and you're stuck with a
 growing file.n  J Actually, unused space left by deleted records has always been reusable ifJ new insertions happened to target a bucket with such free space in it.  InH fact, if you populate even a PV1 file, then delete all the records, thenK re-populate it with the same records (in any order), its space may often be-J reused fairly effectively, since (save for bucket splits caused by deletedK record RRV detritus taking up additional space) the records will target theDI same buckets the first set filled (since the index structure remains from4 that first population).a   - bill   >cI > You should note that convert/reclaim is callable from inside a program.f Just> > ensure the file is closed and call the conv$reclaim routine: > & > $DESCRIPTOR(myfile,"MYBIFILE.DATA"); > struct { long number;r > long data_scanned ;  > long data_reclaimed ;o > long index_reclaimed ; > long total_reclaimed ; > } myvector ; >d > myvector.number = 4C+ > status = conv$reclaim(&myfile,&myvector);- >-> > and voila. This also drops statistics in the myvector thing.   ------------------------------  $ Date: Wed, 5 Jul 2000 19:32:38 -0500) From: "John E. Malmberg" <wb8tyw@qsl.net>s. Subject: Re: SCSI hard drive 4 VAX 4700 needed/ Message-ID: <sm7kgq9qgas134@corp.supernews.com>d  / Al Gill <al@spie.organization> wrote in messaget( news:sm6sgi3fgas97@corp.supernews.com...I > I'm looking for a 5 to 10GB size SCSI drive to attach to our VAX 4700A.a MyE > RF slots are full and I do not want to add another controller, etc.u  % What do you have to connect it up to?   J The VAX 4000-700A as I remember it, does not have a SCSI controller in it.B You need to have either a Q-BUS SCSI controller, or a DSSI to SCSI
 converter.  F The Digital KZQSA SCSI controller is not supported for magnetic disks.   -Johnd wb8tyw@qsl.network   ------------------------------  % Date: Wed, 05 Jul 2000 13:51:33 -0400d* From: David A Froble <davef@tsoft-inc.com>: Subject: Re: Selling software to corporation AND hobbyists- Message-ID: <396375A5.3A8A72DC@tsoft-inc.com>   L Talk to Randy Park. (rjpark@mindspring.com)  He uses a licensing product forM XENTIS and possibly other products.  It's not free, just as your product will0= not be free, but I've not heard him complain about the price.8  G I think Terry gave you some good ideas/advice for your other questions.I   Dave     JF Mezei wrote:  > J > I intend to eventually productize some software. I would like to make itM > available for purchase/download on the web, with a license key emailed once.A > application is processed (and invoice emailed or paper-mailed).i > M > I think that initially, I will only issue the equivalent of temporary paks,eH > with a permanent one issued when payment is received. This allows easy  > management of free-trials etc. > T > (note, that none of this uses LMF since it is not accessible to use lowly humans). > F > However, I would also like to make the product available for free toL > hobbyists, I think that there is value there since they could also provide: > debugging information, suggestions for improvements etc. > J > However, I am a bit at a loss on how to implement this. This is a pagingK > product (for pagers and cell phones), so it is pretty hard to cut down onkO > capabilities without making product unusable, and I don't want to add text toaO > every page sent "this was sent with a hobbyist version" since such text takesrR > up valuable space on a cell phone. (and wouldn't go on a numeric pager anyways). > J > Are there any suggestions on how I could go about this to provide a goodM > product, yet find a way that corporations will want to pay for it, a way tohO > distinguish/define what a hobbyist is, and somehow enforce that distinction ?V > I > In other words, what could the software do in "hobbyist" mode do that a I > corporate environemnt woudl balk at while it would not be a problem forf > hobbysts ?   -- u4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com6 T-Soft, Inc.  170 Grimplin Road  Vanderbilt, PA  15486   ------------------------------   Date: 5 Jul 2000 19:33:55 GMTt2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) Subject: Re: Trapping a STOP/IDa6 Message-ID: <8k02j3$g7t$1@mailint03.im.hou.compaq.com>  Q In article <39630843.A6559095@csc.com>, Adrian Birkett <abirkett@csc.com> writes:sD :I am currently writing a program that requires an event to be firedG :should the program be terminated for any reason. I can trap a CTRL-Y/C-I :and a normal process exit using normal system services but all traps areo" :evaded if someone does a STOP/ID.  E   As they (typically) should be.  $delrpc and STOP/ID are "hard kill"nF   commands. and are deliberately not easy to prevent.  When an OpenVMSA   user is authorized GROUP or WORLD privilege, then that user is  B   authorized to (potentially) damage your application environment.B   And when a user issues $delprc or STOP/ID, the intent is to stop.   all processing in the target process (hard).  <   "Soft kill" commands -- such as $forcex -- can be blocked.  I :Is there any way of trapping this or am I wasting my time? Any responses 	 :welcome.   G   There are brute-force ways, such as inner-mode run-down handlers and r)   such.  I don't recommend this approach.J  @   You can also potentially place the processing in an inner-modeD   module -- execlet or driver -- and entirely avoid process context.@   (And with no process context, there is no target for $delprc.)C   The down-side involves the limitations of inner-mode programming.rB   (Particularly cases where programming errors can lead to system    crashes.)e  C   It is usually far better (and easier) to "protect" these criticalrE   applications via the use of detached processes, and privileges and g	   such.  ?  B   Put the critical portions of the processing in a detached serverE   process, and use a control program (GUI, command line, CGI script) dE   elsewhere to communicate with it.  Thus the only failures that you wD   will typically see in the server are due to use-of-privileges, or E   due to system failures, or (of course) due to application failures.:  C   And please note that no matter what you do, a hard-stop is still  C   possible as a result of a system (eg: power) failure -- you will jF   likely want to survive this case, so you thus have the same problem B   of handling (recovering from) $delprc.  Thus the correct way to D   "block" a $delprc is to simply remove the privilege(s) of anybody G   that is unauthorized to or that proves otherwise unable to correctly  ;   use it.  (Administrative retribution, in other words. :-)   A   And a parallel would be RMS journaling or database recovery, asn@   these can deal with the required all-or-nothing nature of fileD   updates.  Either all parts of the transaction make it to the disk,C   or no parts.  And both RMS journaling and databases can roll-backeD   partial or otherwise failed updates -- whether from $delprc "hits"!   or from a system power failure.i  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Wed, 05 Jul 2000 15:53:56 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca>i Subject: Re: Trapping a STOP/IDo, Message-ID: <39639247.B1ACC324@videotron.ca>   Hoff Hoffman wrote:oB >   You can also potentially place the processing in an inner-modeF >   module -- execlet or driver -- and entirely avoid process context.  N In a previous life, I supported a SWIFT application on VAX and the programmersJ in belgium leanred how to make an ACP and decide it was good and should beM implemented to prevent customers from stopping their software by brute force.u  J As a result, if that process ever got stuck, the only way out of it was toN reboot the system. Yeach. Yeach Yeach from a systems management point of view.K I just hate software that does not allow the system manager to take control 
 when needed.    E So, what exactly is an ACP, and how does it make itself undeletable ?,   ------------------------------  $ Date: Wed, 5 Jul 2000 16:17:06 -0400. From: "Kenneth Randell" <kenr@datametrics.com> Subject: Re: Trapping a STOP/IDi+ Message-ID: <8k04vf$emi$1@bob.news.rcn.net>d  @ If you are really SURE of your application, you could always set; PCB$M_NODELET in the PCB.  This won't trap the STOP/ID, but + it would prevent it from having any effect.   @ I would only use this as a last gasp prevention method, however.   Ken Randell$  > Adrian Birkett wrote in message <39630843.A6559095@csc.com>... >All,> >nD >I am currently writing a program that requires an event to be firedG >should the program be terminated for any reason. I can trap a CTRL-Y/CnI >and a normal process exit using normal system services but all traps aree" >evaded if someone does a STOP/ID. >6I >Is there any way of trapping this or am I wasting my time? Any responsesn	 >welcome.t >  >Thanks, >p >Ade >e >.0 >===== My opinion, not that of my employer ===== >1   ------------------------------  # Date: Wed, 05 Jul 2000 20:33:33 GMTl= From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-)t Subject: Re: Trapping a STOP/IDi0 Message-ID: <009ECA27.397855D6@SendSpamHere.ORG>  \ In article <8k04vf$emi$1@bob.news.rcn.net>, "Kenneth Randell" <kenr@datametrics.com> writes:A >If you are really SURE of your application, you could always setr< >PCB$M_NODELET in the PCB.  This won't trap the STOP/ID, but, >it would prevent it from having any effect. >cA >I would only use this as a last gasp prevention method, however.n  I I caution against this.  There are a number of issues with doing this andmI you are likely to get a call some day from a user bitten by the bit being-J set.  I've encountered a number of apps which do this and let me tell you,I I've had my share of walking users through fixing the ilk this causes.  AeI process which incurs an "unhandled" exception will attempt to run itself -J down (with subsequent process deletion).  Such processes get into a catch-K 22 situation attempting to commit suicide when they have been immortalized.eI The results of this usually manifest themselves in a process which clocksaJ away CPU time and remains COMpute with a priority of 0.  Short of diddlingI the PCB$L_STS bits to zap this PCB$V_NODELTe bit with DELTA, there is go-i8 ing to be a very necessary system reboot in your future.  J One option is to set the PCB$V_DELPND bit in the PCB$L_STS.  This will notJ keep the process from terminating itself if it needs to and it *will* keepK folks from deleting the process via $DELPRC (STOP/ID).  Of course, any user-K suitably privied will be able to zap this bit away with DELTA and issue thes STOP/ID.  K One interesting side effect of setting PCB$V_DELPND in the PCB$L_STS is any2K attempt to SHOW PROCESS will report the process as SUSPended.  Keeps pryinge eyes away from your process.   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMu  H        http://www.OpenVMS.digital.com:8000/72final/9996/9996pro_063.htmlH        http://www.OpenVMS.digital.com:8000/72final/9996/9996pro_072.html   ------------------------------   Date: 5 Jul 2000 18:19:32 -0500h2 From: malmberg@eisner.decus.org (John E. Malmberg) Subject: Re: Trapping a STOP/IDd+ Message-ID: <f317k4E$VBJ2@eisner.decus.org>'  ' In article <39630843.A6559095@csc.com>, .  Adrian Birkett <abirkett@csc.company> writes: > 9 > I am currently writing a program that requires an eventU> > to be fired should the program be terminated for any reason.> > I can trap a CTRL-Y/C and a normal process exit using normal< > system services but all traps are evaded if someone does a
 > STOP/ID. > < > Is there any way of trapping this or am I wasting my time? > Any responses welcome.  7     If the event does not have to be fired by the dyinge8     program or process, then the $ENQ system service can     be of use to you.s  9     You can have a watcher program anywhere on the systemt,     or on another node in a VMScluster then.  9     When the program that requires an event starts up, itA8     aquires an exclusive lock on a resource representing     the program.  <     If the program terminates for any reason, that lock will     be released.    =     The watcher program when it starts up, attempts to aquirey;     an exclusive lock on the resource with an AST.  If thata:     AST ever completes sucessfully, then the watcher knows:     that the monitored program has terminated, and such it     can fire off the event.   :     To avoid a startup/restart race condition, the watcher?     program should also queue up a blocking AST to release it'sm>     exclusive lock on the resource when the watched program is     restarted.  <     Using termination mailboxes for process termination does<     not work if the process does not have suffient resources.     available to send the termination message.   -John  wb8tyw@qsl.network   ------------------------------   Date: 5 Jul 2000 19:42:44 CDT.= From: wayne@tachysoft.xxx.000880.killspam.00bd (Wayne Sewell)a Subject: Re: Trapping a STOP/ID . Message-ID: <u3O0Qtq9uygg@tachxxsoftxxconsult>  \ In article <39639247.B1ACC324@videotron.ca>, JF Mezei <jfmezei.spamnot@videotron.ca> writes: > Hoff Hoffman wrote:eC >>   You can also potentially place the processing in an inner-modexG >>   module -- execlet or driver -- and entirely avoid process context.t > P > In a previous life, I supported a SWIFT application on VAX and the programmersL > in belgium leanred how to make an ACP and decide it was good and should beO > implemented to prevent customers from stopping their software by brute force.e  K This is not the purpose of an ACP.  If you just want to prevent the processtG from being deleted, *any* process can do this with a kernel mode hack. nC However, you want to provide a mechanism for shutting down cleanly.   L > As a result, if that process ever got stuck, the only way out of it was toP > reboot the system. Yeach. Yeach Yeach from a systems management point of view.M > I just hate software that does not allow the system manager to take controlh > when needed.   >  > So, what exactly is an ACP,   G Ancillary Control Process.  Back in the old days, all higher level diskoO operations were performed by the disk ACP.  When you issued a file open or some2J such, the high-level request was passed to the ACP, which broke it up intoM multiple lower level QIOs.   The high-level qio had to wait for the low-level N ones to complete, and this was easier to do with a separate process.  That wasN how disk i/o was originally implemented.   The problem was that *all* disk I/OL for the system was handled by a single process (or there might have been oneO per disk, I don't remember, as this was a *long* time ago).  This functionalityoM is now performed by the Extended QIO Processor, or XQP, which operates in theu1 context of the process doing the disk operations.e  * >and how does it make itself undeletable ?  3 There is a bit in status field of the PCB for this.e  N One of the reasons it is hard to stop a traditional ACP arbitrarily is that itJ is handling I/O operations.  It has to wait for them to complete before itK terminates.   It can cancel or abort all of the  operations pending, but iteO still must wait for them to finish one way or another.  Otherwise you have lostsG i/o request packets, hung devices, and similar chaos in the i/o system.s  G About the only way to stop an acp cleanly would be to send some sort ofrF termination signal to the acp via logical or $enq.  It would then stop1 accepting new i/o request packets and would startaH cancelling/aborting/waiting-for those in progress.  Once *all* irps were4 complete, only then could the process actually exit.  M If you simply stop the process, it can do *none* of this cleanup and you have . hosed the i/o system.  Hence the nodelete bit.  D Of course, this is specific to an ACP.  As you can see, an ACP and aK non-deletable process are not the same thing, though one is a subset of thee other.     Wayneo -- tO ===============================================================================uM Wayne Sewell, Tachyon Software Consulting  (281)812-0738  wayne@tachysoft.xxxr: http://www.tachysoft.xxx/www/tachyon.html and wayne.html  K change .xxx to .com in addresses above, assuming you are not a spambot  :-)PO ===============================================================================(O Otter, on dining with Bluto:"It's perfectly safe if you keep your arms and legsr 			away from his mouth."   ------------------------------   Date: 5 Jul 2000 18:17:21 GMT 2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) Subject: Re: Tutorialu6 Message-ID: <8jvu3h$e18$3@mailint03.im.hou.compaq.com>  e In article <39619ad4@newsserver1.picknowl.com.au>, "Dave Cole" <daveandrosie@picknowl.com.au> writes:sM :Hi, I am new to the world of VMS but have to learn it.  I would like to knowDM :if anyone knows of any tutorials that are available that would be useful for  :a new user.  G   The OpenVMS Frequently Asked Questions (FAQ) section entitled "DOC11.mF   Where can new users find tutorial information about OpenVMS?" might D   be of interest.  The FAQ is accessable via www.openvms.compaq.com.  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Wed, 05 Jul 2000 21:07:30 -0500e) From: Mike Drabicky <drabicky#dallas.net>n: Subject: Re: Upgrade to VMS 7.2 fails on Help Message File8 Message-ID: <vkm7mskrte2g2gt7plkcc6k893vdos4emk@4ax.com>  1 On Wed, 5 Jul 2000 11:40:29 -0400, "Peter Weaver"d <peter.weaver@stelco.ca> wrote:   A >I am upgrading my Hobbyist Charon-VAX from V6.2 to V7.2 and have D >placed the Help message on a different disk. After answering all ofE >the questions and rebooting the node I see the following (the systemf9 >drive is DKA0, the drive for the help messages is DKA1);- >-$ >Any idea on what is happening here?  B So move your help file back to the common part of SYS$HELP for theC install then copy it back to your normal location afterwards. Maybe ? I'm missing something here but that certainly seems the easiest@ solution to this issue.a  
 Mike Drabickyt   ------------------------------   Date: 5 Jul 2000 18:05:21 GMTo2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)! Subject: Re: Very Large RMS Filesh6 Message-ID: <8jvtd1$e18$1@mailint03.im.hou.compaq.com>  I In article <8jr4qg$8qt$1@nnrp1.deja.com>, msi_bruce@earthlink.net writes:uI :Does anyone have any tips or suggestions for dealing with very large RMSlI :files.  We are working with RMS indexed files in the 2 - 4 gigabyte sizedI :range.  These files are very difficult to tune or recover due to run thei : large amount of time...t  9   What order of magnitude is the "large amount of time"? e  5   Some general details of the disk storage subsystem?i  :   Some general details of the OpenVMS VAX or Alpha system?  >   Some general details of the application uptime requirements?  @ :it takes to run the ANALYZE and CONVERT utilities against them.  H   I will assume you are aware of the EDIT/FDL indexed file tuning stuff.  E   The biggest single win I've seen in this area -- when you are usinglF   CONVERT or otherwise writing a whole bunch of data -- is to get the E   RMS extent sizes set appropriately, and to have reasonable process tD   quotas and RMS buffer count settings.  Depending on the details ofB   the conversion, the CONVERT/FAST tool can be of interest, too...  >   Having more indexed keys around makes for slower, of course.   H :Are there any products or tools that would allow us to dynamically tune4 :the files without having to actually recreate them?  D   I'm not immediately aware of such tools -- though they may well beF   around somewhere.  Most RMS indexed file tuning that I have seen is 0   performed statically; it is performed offline.  D   If you are creating and (more importantly) deleting records in an D   RMS indexed file, there are some steps you can take to reduce the C   build-up of dead records in the file, but I have always expected xC   to have to perform some occasional (offline) RMS file tuning for tC   the reclamation.  I have generally also designed the applications-B   to typically only create new records and to only delete records C   from an indexed file when absolutely necessary -- I end up either E   updating the records in placem or marking the records as "deleted".nB   (Or I assume that I will have to go off-line for a while whilst D   rebuilding the RMS indexed file -- though depending on the design E   of the file, going off-line typically only becomes required once a rD   year or less often.  In the last two big RMS application projects C   I was involved with, indexed file "reclamation" conversions were  ,   required only once every couple of years.)  A   If I need read-write capabilities and/or high(er) uptime and/or @   unassisted on-line BACKUP (and I don't want to incorporate the@   BACKUP into the application), then I will generally move to a 0   design that is based on a relational database.  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------   Date: 5 Jul 2000 19:11:46 -0500 * From: kuhrt@eisner.decus.org (Marty Kuhrt)! Subject: Re: Very Large RMS Filess+ Message-ID: <gPW8bYCD86Wf@eisner.decus.org>a  I In article <8jr4qg$8qt$1@nnrp1.deja.com>, msi_bruce@earthlink.net writes:o > Hi,  > J > Does anyone have any tips or suggestions for dealing with very large RMSJ > files.  We are working with RMS indexed files in the 2 - 4 gigabyte sizeJ > range.  These files are very difficult to tune or recover due to run theI >  large amount of time it takes to run the ANALYZE and CONVERT utilitiesn > against them.t > I > Are there any products or tools that would allow us to dynamically tunev5 > the files without having to actually recreate them?  > 	 > Thanks,n > 
 > Bruce C.  > You might want to check out RMS Expert.  The SPD is located at1 http://www.rdperf.com/RDDOC/PRODUCT_RMSX_SPD.HTMLT  2 I have no affilitation with R&D Performance Group.   ------------------------------  # Date: Wed, 05 Jul 2000 17:45:20 GMTs% From: Bill Hobbs <hobbsb@my-deja.com>l Subject: VMS employmentt) Message-ID: <8jvs7c$ig1$1@nnrp1.deja.com>s  H from "Re: OpenVMS commentaries (was Re: Gartner commentary on Wildfire)" post by Hoffman on 2000 Jun 5:F > :The tide of OpenVMS is dead has not turned yet.  I was talking to a@ > :representative of a recruitment company earlier this week who
 believes thathG > :OpenVMS is dead or is on the way.  She was very surprised that therer
 is so muchH > :good work being done at Compaq to reverse this trend.  I take this as anD > :indicator that the companies which she deals with are moving away	 from VMS.  >lF >   Or she's not been contacted for folks looking for OpenVMS -- we'reC >   currently hiring, for instance.  The OpenVMS VP, Rich Marcello,oD >   recently mentioned that one thing under serious consideration isA >   the inclusion of a recruitment or jobs website for folks withp2 >   OpenVMS skills via the Compaq OpenVMS website.  G Hoff, I was wondering if this website is in imminent danger of becomingp
 reality?  :-)n  G > :Perhaps the new marketing materials should be send out in paper copye a littleD > :more widely (not just to existing customers who are still heavily	 committed:C > :to VMS) but to companies with only one or two small systems in a  bigger0 > :environment (the small fish in the big pond). >pC >   I can't speak for the marketing plans for OpenVMS and certainlytB >   not for the contents of the press.  I can say that there are aD >   number of OpenVMS sites looking for OpenVMS-trained folks.  (eg: >   us.)  E I would like to examine your list of sites to see if I've missed any.fF Hmm, would your list be based on calls to customer support looking forH VMS-trained folk?  Which probably means that you can't show it to us ... :-(e  H My current contract ends this month.  I would prefer to continue working- on VMS.  Anyone in Southern California need atF programmer/analyst/DBA/general-purpose-computer-geek?  With an expired Top Secret clearance?   0 <http://home.earthlink.net/~vogonhi/resume.html>/ <http://home.earthlink.net/~vogonhi/resume.pdf>n      & Sent via Deja.com http://www.deja.com/ Before you buy.D   ------------------------------  % Date: Wed, 05 Jul 2000 22:28:41 -0400a) From: yyyc186.illegaltospam_@flashcom.netn Subject: Re: VMS employmente9 Message-ID: <3963ef78$1$lllp186$mr2ice@news.flashcom.com>   . In <8jvs7c$ig1$1@nnrp1.deja.com>, on 07/05/00 5    at 10:28 PM, Bill Hobbs <hobbsb@my-deja.com> said:o  D If you wish to know the lowest you will ever be paid, then check outJ www.dice.com.  Posts there are generally for less than half the going rateI for any geographic area.  Go there first to find out the absolute minimum E you could make...judging from what I see there it appears most of theeI posters are used to pricing people with residency status problem or parol 
 board issues.n   Roland  I >from "Re: OpenVMS commentaries (was Re: Gartner commentary on Wildfire)"l >post by Hoffman on 2000 Jun 5:-G >> :The tide of OpenVMS is dead has not turned yet.  I was talking to aSA >> :representative of a recruitment company earlier this week who  >believes thatH >> :OpenVMS is dead or is on the way.  She was very surprised that there >is so mucheI >> :good work being done at Compaq to reverse this trend.  I take this asR >anyE >> :indicator that the companies which she deals with are moving aways
 >from VMS. >>G >>   Or she's not been contacted for folks looking for OpenVMS -- we'reoD >>   currently hiring, for instance.  The OpenVMS VP, Rich Marcello,E >>   recently mentioned that one thing under serious consideration isrB >>   the inclusion of a recruitment or jobs website for folks with3 >>   OpenVMS skills via the Compaq OpenVMS website.i  H >Hoff, I was wondering if this website is in imminent danger of becoming >reality?  :-)  H >> :Perhaps the new marketing materials should be send out in paper copy	 >a littlenE >> :more widely (not just to existing customers who are still heavily 
 >committedD >> :to VMS) but to companies with only one or two small systems in a >biggerf1 >> :environment (the small fish in the big pond).h >>D >>   I can't speak for the marketing plans for OpenVMS and certainlyC >>   not for the contents of the press.  I can say that there are aTE >>   number of OpenVMS sites looking for OpenVMS-trained folks.  (eg:r	 >>   us.)p  F >I would like to examine your list of sites to see if I've missed any.G >Hmm, would your list be based on calls to customer support looking foroI >VMS-trained folk?  Which probably means that you can't show it to us ...  >:-(  I >My current contract ends this month.  I would prefer to continue workingI. >on VMS.  Anyone in Southern California need aG >programmer/analyst/DBA/general-purpose-computer-geek?  With an expiredh >Top Secret clearance?  1 ><http://home.earthlink.net/~vogonhi/resume.html>A0 ><http://home.earthlink.net/~vogonhi/resume.pdf>      ' >Sent via Deja.com http://www.deja.com/w >Before you buy. -- 0; -----------------------------------------------------------iD yyyc186@flashcom.net              To Respond delete ".illegaltospam"6                             MR/2 Internet Cruiser 1.528                             For a Microsoft free univers; -----------------------------------------------------------'   ------------------------------  $ Date: Wed, 5 Jul 2000 13:25:28 -0500- From: "Jon Power" <jon.power@sector7XXXX.com>g Subject: We need VMS peoplev6 Message-ID: <q4L85.12050$W7.243938@news4.giganews.com>  L Sector 7 (www.sector7.com 800 VMS UNIX) has a need for high grade VMS people (is there any other kind).  I We need people for non migration VMS projects and also VMS to UNIX and NTgI migration projects. Without wishing to create a mass of hate mail in thistJ group, there are VMS users out there that have decided for whatever reasonB to migrate and Sector7 provides this service. The project are VERYK interesting, USA and Worldwide. The great part of this business is that youtL get to see all of the cool technology used on systems ranging from Star Wars4 to Banking and move on to the next exciting project.  J We have immediate needs for VMS people in the UK (FORTRAN) and also PDP-11) BASIC/MACRO-11 (yes its still out there).g  K In the USA we have many projects with ACMS FORTRAN. COBOL. MACRO Assembler.c  J We also need VMS application assessment folks. The aim of this group is toB assess the time/cost/schedule for a migration from VMS to UNIX/NT.  , We need project managers, engineers etc etc.  K If there are people that would like international travel PLEASE CONTACT US.b  
 As an examplel --------------K We have just migrated a misson critical system (HIGHLY VISABLE) from PDP-11 I MACRO to VMS MACRO-32 (they had written their own operating system on theiL PDP-11 directly using the page mapping registers in the APPLICATION), now weL are starting on the next part which is MACRO-32 to C. For VMS we wrote a QIOE driver to manipulate the PFN tables without incurring the overhead ofk. $MGBLSC) The project is in the NY area. ......  " Rdb to Oracle skills always needed  / We are interested in FULL TIME and Contractors.A  F Also: We do a large amount of UNIX to UNIX porting and also we are now seeing NT to UNIX.  	 Jon Powera   ------------------------------  % Date: Wed, 05 Jul 2000 19:34:58 -0500l7 From: "David J. Dachtera" <djesys.nospam@earthlink.net>  Subject: Re: We need VMS people - Message-ID: <3963D432.5792CDB4@earthlink.net>    Jon Power wrote: > N > Sector 7 (www.sector7.com 800 VMS UNIX) has a need for high grade VMS people > (is there any other kind). > 0 > We need people for non migration VMS projects   H Probably _THE_ area where you're likely to get the best response in this
 newsgroup.   [snip]H > Also: We do a large amount of UNIX to UNIX porting and also we are now > seeing NT to UNIX.  @ Don't be surprised if you start seeing NT -> VMS and UN*X -> VMSC migrations as disenchantment with NT and other unsupported softwarem' (Linux, *BSD, etc.) continues to mount.o   -- t David J. Dachterat dba DJE Systems  http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/a   ------------------------------  " Date: Thu, 6 Jul 2000 03:10:37 GMT0 From: "Terry C. Shannon" <shannon@world.std.com> Subject: Re: We need VMS peoplee& Message-ID: <Fx9A3A.MMn@world.std.com>  B "David J. Dachtera" <djesys.nospam@earthlink.net> wrote in message' news:3963D432.5792CDB4@earthlink.net...d > Jon Power wrote: > >.I > > Sector 7 (www.sector7.com 800 VMS UNIX) has a need for high grade VMS  people > > (is there any other kind). > > 1 > > We need people for non migration VMS projects  >kJ > Probably _THE_ area where you're likely to get the best response in this > newsgroup. >  > [snip]J > > Also: We do a large amount of UNIX to UNIX porting and also we are now > > seeing NT to UNIX. > B > Don't be surprised if you start seeing NT -> VMS and UN*X -> VMSE > migrations as disenchantment with NT and other unsupported software ) > (Linux, *BSD, etc.) continues to mount.s  D I know of at least one instance in which a Capital District US Govt.G customer has migrated from NT to OpenVMS (never realized that the Govt.t possessed such wisdom).    ------------------------------  % Date: Wed, 05 Jul 2000 23:19:35 -0400a+ From: "Main, Kerry" <Kerry.Main@compaq.com>  Subject: RE: We need VMS peopleyJ Message-ID: <910612C07BCAD1119AF40000F86AF0D8052844EF@kaoexc4.kao.dec.com>   David,  D >>> Don't be surprised if you start seeing NT -> VMS and UN*X -> VMSK migrations as disenchantment with NT and other unsupported software (Linux,e# *BSD, etc.) continues to mount. <<<   F Yep. Already have some Customers doing NT and UNIX TO VMS migrations.   F Current Customer I am working with is considering an Oracle on UNIX toL Oracle on VMS database migration as well. They are centralizing their OracleL on OpenVMS applications and are planning a disaster tolerant, load balanced, multi-site SAN config. n  L Single site solutions are very hard to justify when doing large scale serverK consolidation projects. Regional, district types typically say "centralize?aL But if we lose that datacenter, then I am out of nbusiness, so forget it !".B By implementing a multi-site, load balanced config, you remove the< "yeahbuts" from those Regional and district folks arguments.  I In this particular Customer case, if approved, they would still leave the 6 application servers on UNIX or NT as the case may be.   K My perspective on this, is that Compaq is a multi-platform company and eachaL OS has its strengths in specific area's. Compaq will assist its Customers toB migrate to and/or from any of its supported OS's. Of course from a+ competitive platform goes without saying ..n   :-)p   Regards,  
 Kerry Main Senior Consultant,
 Compaq Canadau Professional Servicesw Voice : 613-592-4660 FAX   : 819-772-7036 Email : kerry.main@compaq.coma       -----Original Message-----< From: David J. Dachtera [mailto:djesys.nospam@earthlink.net]& Sent: Wednesday, July 05, 2000 8:35 PM To: Info-VAX@Mvb.Saic.Comb Subject: Re: We need VMS peoplee     Jon Power wrote: > G > Sector 7 (www.sector7.com 800 VMS UNIX) has a need for high grade VMSn people > (is there any other kind). > 0 > We need people for non migration VMS projects   H Probably _THE_ area where you're likely to get the best response in this
 newsgroup.   [snip]H > Also: We do a large amount of UNIX to UNIX porting and also we are now > seeing NT to UNIX.  @ Don't be surprised if you start seeing NT -> VMS and UN*X -> VMSC migrations as disenchantment with NT and other unsupported software@' (Linux, *BSD, etc.) continues to mount.H   --   David J. Dachtera2 dba DJE Systems  http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/    ------------------------------   Date: 5 Jul 2000 20:59:20 GMTa2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)9 Subject: Re: What is an ACP? (was Re: Trapping a STOP/ID)Y6 Message-ID: <8k07j8$h16$1@mailint03.im.hou.compaq.com>  \ In article <39639247.B1ACC324@videotron.ca>, JF Mezei <jfmezei.spamnot@videotron.ca> writes: :Hoff Hoffman wrote:C :>   You can also potentially place the processing in an inner-mode G :>   module -- execlet or driver -- and entirely avoid process context.u :cO :In a previous life, I supported a SWIFT application on VAX and the programmerso) :in belgium leanred how to make an ACP...oF :So, what exactly is an ACP, and how does it make itself undeletable ?  N   An Ancillary Control Process (ACP) is effectively a process-level extension M   of an OpenVMS device driver, and permits a device driver access to various tL   tasks -- such as reading or writing a disk file -- that are not otherwise    available to the driver.  F   In the OpenVMS context, ACPs (and a decendent, the XQP) provide the K   bulk of the processing of the file system support -- the driver performs lK   the reads and the writes, caches what it needs frequently, and then lets Y$   the ACP handle the "fancy" stuff.   G   The driver literally queues an I/O request to the ACP for processing,.F   and wakes it up.  The ACP then churns away with whatever it needs toG   do, and then (hopefully) passes the original I/O request back to the  H   driver for eventual completion.  The ACP code executes in kernel mode,!   and in outer modes as required.n  J   Back around V4.0, the Files-11 ODS-2 ACP was moved into process address J   space for various reasons, and renamed the XQP.  The XQP design avoided K   the need for the context switch inherent with ACPs, among other benefits.aI   (This removal of the context switch was a big enhancement for something H   as busy as the file system ACP, but may or may not be a particular winH   for an ACP that is not as busy nor as central as the file system ACP.)  K   I used an ACP to handle the time-consuming voice-related processing with nI   the DECvoice modules, including loading and unloading device microcode cJ   from the microcode files -- it was easier to debug the ACP than to work G   in driver context, and far less impact on the system as a whole if I  C   needed to spend a whole lot of time doing the related processing.a  J   An ACP is not inherently undeletable but, via the PCB$x_NODELET bit, it J   can avoid most deletion attempts.  Given that deleting an ACP is a very H   good way to "loose" an I/O request packet (IRP) somewhere (and thus toM   seriously "stuff up" the kernel-mode context of a process), the prevention eK   of such errant process deletions is usually a good idea.  Unfortunately, fI   an ACP can also potentially fail on accord of its own bugs, leading to hE   the same lost IRP problems.  Symptom of a lost I/O: an undeleteabledE   process "spinning" in the run-down code...  (I saw a _lot_ of thesel   processes during debug. :-)d  I   I was looking at providing a skeleton of an ACP and a pseudo-driver, asUK   the construct is very handy -- and completely undocumented -- and as thisIL   question comes up regularly.  (Yes, I have a copy of Hanrahan and Leahy's L   long-since out-of-print "VMS Advanced Device Driver Techniques" book.  It L   was a good introduction to the topic, but it did have a few holes.  I was M   really annoyed when I first received a copy of the book, having just spent  G   the previous two months learning about, designing, implementing, and  (   debugging a moderately-complex ACP.:-)  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------   End of INFO-VAX 2000.374 ************************