1 INFO-VAX	Thu, 11 May 2000	Volume 2000 : Issue 263       Contents:/ Re: 1 instance galaxy and getting EBA0: online? ' Any Way To Suppress LOGs From UCX$RSH ?  Re: Arne your mail is bouncing Re: big Fortran data file  big Fortran data file  Re: big Fortran data file  Re: big Fortran data file  Re: big Fortran data file  Re: big Fortran data file  Re: big Fortran data file  Re: big Fortran data file  Re: big Fortran data file  CMUIP and FTP problems Compaq and CORBA CORBA on OpenVMS AXP Re: CORBA on OpenVMS AXP Re: CORBA on OpenVMS AXP Re: CORBA on OpenVMS AXP1 Re: DEC C, table mapping options across platforms 1 Re: DEC C, table mapping options across platforms 1 Re: DEC C, table mapping options across platforms 1 Re: DEC C, table mapping options across platforms  DS700 question... * free credit card machine for your business/ Re: ftp.decus.org is back up!  And running VMS! / Re: ftp.decus.org is back up!  And running VMS! / Re: ftp.decus.org is back up!  And running VMS! / Re: ftp.decus.org is back up!  And running VMS! / Re: ftp.decus.org is back up!  And running VMS!  Re: Internal date  Re: Internal date  Re: Internal date  Re: Internal date ! Is "The GNU on VMS Project" dead? % Re: Is "The GNU on VMS Project" dead? ' Re: Kind of portable Alpha, VMS capable % Length of Username - Was RE: Username ) Re: Length of Username - Was RE: Username ) Re: Length of Username - Was RE: Username 
 LMF weirdness  Re: LMF weirdness  Re: LMF weirdness  Re: LMF weirdness " Re: Looking for infor for 4000/200" Re: Looking for infor for 4000/200" Re: Looking for infor for 4000/200" Re: Looking for infor for 4000/200" Re: Looking for infor for 4000/200 Re: Marketing opportunity  Re: Marketing opportunity  Re: Mozilla M15's out... Re: Mozilla M15's out... Re: MPI for OpenVMS?& Re: ODBC Drivers for OpenVMS RMS files	 Re: OPCOM 	 Re: OPCOM ! Re: Pitch of the Margin bell Beep ! Re: Pitch of the Margin bell Beep 1 problem linking to ucx$ipc.olb on AXP/VMX71-UCX40 5 Re: problem linking to ucx$ipc.olb on AXP/VMX71-UCX40 5 Re: problem linking to ucx$ipc.olb on AXP/VMX71-UCX40 5 Re: problem linking to ucx$ipc.olb on AXP/VMX71-UCX40 4 Re: Problem with Seagate disk on VAXstation 4000 VLC Question on VAXstation 2000  Re: Question on VAXstation 2000  Re: tape label in EBCDIC test Thanks Hoff (re SDL),  but... ! Re: Thanks Hoff (re SDL),  but...  Re: the latest billybox virus  Re: the latest billybox virus  Timeout Issues Take II RE: Timeout Issues Take II Re: Which VAX to buy?  Wildfire and the future of VMS" Re: Wildfire and the future of VMS" Re: Wildfire and the future of VMS" Re: Wildfire and the future of VMS  F ----------------------------------------------------------------------  % Date: Thu, 11 May 2000 12:10:56 -0400 * From: Clair Grant <grant@evms.zko.dec.com>8 Subject: Re: 1 instance galaxy and getting EBA0: online?0 Message-ID: <391AA350.38DA4732@evms.zko.dec.com>   Jeff Schreiber wrote:  >    P >     From what I can tell from the Galaxy Guide, I need another 8Mb to put intoP >     shared memory [although I couldn't figure out from the galaxy guide how toC >     configure shared memory, other than the SMCI_ SYSGEN params].  > 9 >     But without that extra 8Mb, It sounds like I'm SOL.  > 3 >                                             -Jeff   C You need to run the GCU. It will ask you for the amount of "shared" E memory you would like defined. (I'm assuming your are on platform not G capable of multiple instances.) Chapter 9 in the Galaxy Guide describes  this.      --   Clair Grant  VMS Exec Group Project Leader  COMPAQ Computer Corporation  110 Spit Brook Rd. Nashua, NH  03062    ------------------------------   Date: 11 May 2000 12:44:27 GMT) From: leslie@clio.rice.edu (Jerry Leslie) 0 Subject: Any Way To Suppress LOGs From UCX$RSH ?' Message-ID: <8fe9vb$kfc$1@joe.rice.edu>  Keywords: vms,ucx,rsh,log   C We have an application of the UCX RSH service that will create many B log files, with the danger of the RSH service failing once version 32767 is created.   @ Is there any way to suppress the creation of the RSH LOG files ?  %   $ ucx help set service /log_options    We've already tried:  *   $ ucx set service /log_options=noall rsh  # Here's the versions of VMS and UCX:      $ ucx show version  H     Digital TCP/IP Services for OpenVMS Alpha Version V4.1 - ECO Level 2>     on a AlphaServer 8400 Model 5/625 running OpenVMS V7.1-1H1   Thanks in advance,  4 --Jerry Leslie     (my opinions are strictly my own)   ------------------------------  % Date: Thu, 11 May 2000 16:38:45 +0200 = From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com> ' Subject: Re: Arne your mail is bouncing ) Message-ID: <391AC5F5.5C2934F9@gtech.com>    Robert Young wrote:  > Arne,  >  > Your mail is bouncing:   It should hopefull work now.  # If not then try my private address:    arne.vajhoej@gtech.com   Arne   ------------------------------   Date: 11 May 2000 12:17:18 GMT* From: helbig@astro.rug.nl (Phillip Helbig)" Subject: Re: big Fortran data file. Message-ID: <8fe8ce$svu$1@info.service.rug.nl>  [ In article <8fe81u$svq$1@info.service.rug.nl>, helbig@astro.rug.nl (Phillip Helbig) writes:   I >I have a program where the bottleneck is computing a certain quantity as F >a function of 5 variables.  However, this quantity can be computed inH >advance at leisure.  I'm thinking of precomputing this and then readingH >the file into the program and using a lookup table/interpolation rather( >than computing the values on the fly.    C There is also the question, of course, of whether to write out the  D values individually (though presumably the OS would cache these and H write them out in bunches) or put everything into a big array and write D it out at the end, or something in-between.  (The problem of memory F arises in the application which uses this data; it only arises in the C application which computes it if I opt for the big-array approach.)    ------------------------------   Date: 11 May 2000 12:11:42 GMT* From: helbig@astro.rug.nl (Phillip Helbig) Subject: big Fortran data file. Message-ID: <8fe81u$svq$1@info.service.rug.nl>  H I have a program where the bottleneck is computing a certain quantity asE a function of 5 variables.  However, this quantity can be computed in G advance at leisure.  I'm thinking of precomputing this and then reading G the file into the program and using a lookup table/interpolation rather G than computing the values on the fly.  Especially useful, perhaps, if I F run the application more than once, but even in this minimal case it's probably worth it.    G The application uses the values in a certain order and I can precompute E them in this order.  Should I make one large file and let the virtual F memory system take care of the fact that it won't all fit into memory,H keeping in mind that the values are accessed in the order they are used,F or should I split it up into smaller files?  Should I used a formattedF or unformatted file?  (Four places are enough precision, so the sizes 1 would be about the same.)  Indexed or sequential?   E This is with Compaq Fortran 95 on ALPHA/VMS.  At the moment, I have a C choice between a 255/233 with 64 MB or a 3000/600 with 192 MB.  I'm E assuming the processor speed, not the memory size, will determine the F speed if things are set up correctly with regard to the large file (orE many smaller files, as the case might be).  The application otherwise > needs very little memory and does very little I/O, just number crunching.    H We're talking about 1.5 GB or so of data, i.e. 360 million values to be 	 computed.   I I asked a similar question a few months ago.  I've decided to ask again,  G since there didn't seem to be a consensus in the answers and since I'm  2 finally getting around to actually doing this now.  G If it turns out to be too slow, I might compute it on a faster machine. + Thus, I'd like to stick to standard code.     F The question of one large or several smaller files is mainly relevant G for the application which will USE the data, i.e. I could in principle  G calculate the stuff in several smaller files (which I would have to do  F if I split things up among several machines) and combine them later.  G Things would be easier, though, if one large file could be used in all   circumstances.     --M Phillip Helbig                       Email .............. helbig@astro.rug.nl M Kapteyn Instituut                    Email ................. helbig@man.ac.uk M Rijksuniversiteit Groningen          Tel. ................... +31 50 363 4067 M Postbus 800                          Fax .................... +31 50 363 6100 M NL-9700 AV Groningen                 Web ... http://www.astro.rug.nl/~helbig/   5 My opinions are not necessarily those of my employer.   N <A HREF=" http://gladia.astro.rug.nl:8000/helbig/hire/hire.html ">HIRE ME!</A>   ------------------------------  % Date: Thu, 11 May 2000 09:10:50 -0400 # From: Jim Agnew <agnew@hsc.vcu.edu> " Subject: Re: big Fortran data file+ Message-ID: <391AB15A.3847CFBC@hsc.vcu.edu>   D I'd use unformatted I/O... it waaaaay faster in my experience on vaxE fortran.. if you're truly reading sequentially, then there is no need  for indexes.  G try to keep array operations in row-major order, so the program pointer C does not jump around.  (again, it could be collum-major, i'm really * forgetting here, been a sysadmin too long)   hhmm.. whatelse..??   7 1.2 GB is too much for any disk cache (unless... ;-)  ) B so, hhmm...  could you set global buffercount on the disk to force@ "lookahead"?  (now, i'm assuming i have the name right on globalE buffercount, it could be global buffers, or some other name entirely, ? but there is a setting on the disk or sysgen that allows bigger ; lookahead reads, i think)  or, i could be in extreme coffee / deprivation.  others will have to step in here.    Phillip Helbig wrote:  > J > I have a program where the bottleneck is computing a certain quantity asG > a function of 5 variables.  However, this quantity can be computed in I > advance at leisure.  I'm thinking of precomputing this and then reading I > the file into the program and using a lookup table/interpolation rather I > than computing the values on the fly.  Especially useful, perhaps, if I H > run the application more than once, but even in this minimal case it's > probably worth it. > I > The application uses the values in a certain order and I can precompute G > them in this order.  Should I make one large file and let the virtual H > memory system take care of the fact that it won't all fit into memory,J > keeping in mind that the values are accessed in the order they are used,H > or should I split it up into smaller files?  Should I used a formattedG > or unformatted file?  (Four places are enough precision, so the sizes 3 > would be about the same.)  Indexed or sequential?  > G > This is with Compaq Fortran 95 on ALPHA/VMS.  At the moment, I have a E > choice between a 255/233 with 64 MB or a 3000/600 with 192 MB.  I'm G > assuming the processor speed, not the memory size, will determine the H > speed if things are set up correctly with regard to the large file (orG > many smaller files, as the case might be).  The application otherwise @ > needs very little memory and does very little I/O, just number > crunching. > I > We're talking about 1.5 GB or so of data, i.e. 360 million values to be  > computed.  > J > I asked a similar question a few months ago.  I've decided to ask again,H > since there didn't seem to be a consensus in the answers and since I'm4 > finally getting around to actually doing this now. > I > If it turns out to be too slow, I might compute it on a faster machine. + > Thus, I'd like to stick to standard code.  > G > The question of one large or several smaller files is mainly relevant H > for the application which will USE the data, i.e. I could in principleH > calculate the stuff in several smaller files (which I would have to doF > if I split things up among several machines) and combine them later.H > Things would be easier, though, if one large file could be used in all > circumstances. >  > --O > Phillip Helbig                       Email .............. helbig@astro.rug.nl O > Kapteyn Instituut                    Email ................. helbig@man.ac.uk O > Rijksuniversiteit Groningen          Tel. ................... +31 50 363 4067 O > Postbus 800                          Fax .................... +31 50 363 6100 O > NL-9700 AV Groningen                 Web ... http://www.astro.rug.nl/~helbig/  > 7 > My opinions are not necessarily those of my employer.  > P > <A HREF=" http://gladia.astro.rug.nl:8000/helbig/hire/hire.html ">HIRE ME!</A>   ------------------------------   Date: 11 May 2000 08:52 CST ' From: carl@gerg.tamu.edu (Carl Perkins) " Subject: Re: big Fortran data file- Message-ID: <11MAY200008520469@gerg.tamu.edu>    helbig@astro.rug.nl writes... I }I have a program where the bottleneck is computing a certain quantity as F }a function of 5 variables.  However, this quantity can be computed inH }advance at leisure.  I'm thinking of precomputing this and then readingH }the file into the program and using a lookup table/interpolation ratherH }than computing the values on the fly.  Especially useful, perhaps, if IG }run the application more than once, but even in this minimal case it's  }probably worth it.  } H }The application uses the values in a certain order and I can precomputeF }them in this order.  Should I make one large file and let the virtualG }memory system take care of the fact that it won't all fit into memory, I }keeping in mind that the values are accessed in the order they are used, G }or should I split it up into smaller files?  Should I used a formatted G }or unformatted file?  (Four places are enough precision, so the sizes  2 }would be about the same.)  Indexed or sequential? } F }This is with Compaq Fortran 95 on ALPHA/VMS.  At the moment, I have aD }choice between a 255/233 with 64 MB or a 3000/600 with 192 MB.  I'mF }assuming the processor speed, not the memory size, will determine theG }speed if things are set up correctly with regard to the large file (or F }many smaller files, as the case might be).  The application otherwise? }needs very little memory and does very little I/O, just number  }crunching.  } I }We're talking about 1.5 GB or so of data, i.e. 360 million values to be  
 }computed. } J }I asked a similar question a few months ago.  I've decided to ask again, H }since there didn't seem to be a consensus in the answers and since I'm 3 }finally getting around to actually doing this now.  } H }If it turns out to be too slow, I might compute it on a faster machine., }Thus, I'd like to stick to standard code.   } G }The question of one large or several smaller files is mainly relevant  H }for the application which will USE the data, i.e. I could in principle H }calculate the stuff in several smaller files (which I would have to do G }if I split things up among several machines) and combine them later.   H }Things would be easier, though, if one large file could be used in all  }circumstances.  }-- N }Phillip Helbig                       Email .............. helbig@astro.rug.nl  D The obvious way is to use a single file with each value as a record.G Since they are used sequentially, you can just read them in as you need K them. If the file is opened with the proper options then multiple processes C (across a cluster even) can all use the same file at the same time. D Since they are numbers, I'd suggest an unformatted file to eliminate% the ASCII to binary conversion stage.   A If you expect to need to be able to start at someplace other than B the beginning, at value number N, in the above case you could justC read and discard the initial N-1 records to start but this could be B rather slow. You might want to use a direct access file instead of> a sequential access file, but in the long run this may be more? overhead than skipping records, since once you get to the first 6 record you need the rest are all sequentialy accessed.  F Additional speed improvements could be gotten by storing more than oneI value per record and using a subroutine to get the value you want (with 4 I byte data I could easily see putting 255 or more of them in each record).0F This could result in less I/O overhead, although it may be possible toF get much the same effect simply by specifying a good set of options inA your OPEN statment, particularly a large "buffercount". You mighta@ also get some benefit from making it a fixed record length file.F Possibly using of all of these things together would be most helpfull.  K Since the values are used sequentially, at the moment the only benefit that M I see from making it multiple files is the ability to put each on a different N disk which could speed up the case where you have multiple runs accessing themH at the same time (and it would be helpful if you don't ahve the requiredH space on a single disk, of course), but complicating the issue by havingG to switch files at various points during a run (which is a fairly minorf	 problem).i  L I would probably use a single file, unformatted, sequential access, multipleM values per fixed length record, and specifying a moderatly large buffercount. . Then I'd see if this worked sufficiently well.   --- Carl   ------------------------------   Date: 11 May 2000 14:23:43 GMT* From: helbig@astro.rug.nl (Phillip Helbig)" Subject: Re: big Fortran data file. Message-ID: <8fefpf$1k3$1@info.service.rug.nl>  F In article <11MAY200008520469@gerg.tamu.edu>, carl@gerg.tamu.edu (Carl Perkins) writes: s  G >Additional speed improvements could be gotten by storing more than one C >value per record and using a subroutine to get the value you want    I Actually, the first four variables are stepped through sequentially, and  E the fifth variable is used in a more-or-less random order.  Thus, it  G would make sense to put all of these values (between 1 and 599) in one  I record.  Once this record is in memory, the computation time is probably oF comparable to the time to read the next record so I/O won't be a real  bottleneck.   G >This could result in less I/O overhead, although it may be possible tonG >get much the same effect simply by specifying a good set of options ino9 >your OPEN statment, particularly a large "buffercount". X  
 Non-standard?e   > You mightnA >also get some benefit from making it a fixed record length file.    Why?   ------------------------------  % Date: Thu, 11 May 2000 10:32:43 -04000$ From: Herman D. Knoble <hdk@psu.edu>" Subject: Re: big Fortran data file8 Message-ID: <68glhsgbhs6v0d3t9n4ribm2uhossuqi41@4ax.com>  ; Phil,  You are a seasoned Fortran programmer and innovator. 9 Thus, you have the key to the best way to do this becausedA you know the data, and you know that I/O is often the bottleneck;eH and what binary format the data is in, or how it's buffered is probably 9 a  moot point when one is dealing with such a large file.   ? One technique that could help reduce the file size is "manual" e? compression (your own algorithm based on knowledge of the data :; itself). Does the data repeat? Are the numbers integers and @ limited to certain ranges?  How accurate do the non-integer realB numbers have to be to affect results? That is, can they be encodedF as integers?  I can recall a case where we helped a professor process E stock market data; we were able to compress the data by three orders u< of magnitude because the REAL data was only to 0.1 cents and@ so several reals could be encoded as one  integer; and the data F repeated a lot, so a repeat factor instead of actual data enabled even; more compression. In your case, even one order of magnitude  obviously be a big help. p  A Also, can you write "manually" compressible data and compute only A that data on the fly that cannot be manually compressed? That is,p9 a combination of I/O and CPU instead of one or the other.-   Just a thought.M  H On 11 May 2000 12:11:42 GMT, helbig@astro.rug.nl (Phillip Helbig) wrote:  J -|I have a program where the bottleneck is computing a certain quantity asG -|a function of 5 variables.  However, this quantity can be computed ineI -|advance at leisure.  I'm thinking of precomputing this and then reading I -|the file into the program and using a lookup table/interpolation rathereI -|than computing the values on the fly.  Especially useful, perhaps, if I H -|run the application more than once, but even in this minimal case it's -|probably worth it. S -|I -|The application uses the values in a certain order and I can precomputelG -|them in this order.  Should I make one large file and let the virtualnH -|memory system take care of the fact that it won't all fit into memory,J -|keeping in mind that the values are accessed in the order they are used,H -|or should I split it up into smaller files?  Should I used a formattedH -|or unformatted file?  (Four places are enough precision, so the sizes 3 -|would be about the same.)  Indexed or sequential?  -|G -|This is with Compaq Fortran 95 on ALPHA/VMS.  At the moment, I have atE -|choice between a 255/233 with 64 MB or a 3000/600 with 192 MB.  I'm-G -|assuming the processor speed, not the memory size, will determine thegH -|speed if things are set up correctly with regard to the large file (orG -|many smaller files, as the case might be).  The application otherwise.@ -|needs very little memory and does very little I/O, just number
 -|crunching. a -|J -|We're talking about 1.5 GB or so of data, i.e. 360 million values to be  -|computed.v -|K -|I asked a similar question a few months ago.  I've decided to ask again, mI -|since there didn't seem to be a consensus in the answers and since I'm  4 -|finally getting around to actually doing this now. -|I -|If it turns out to be too slow, I might compute it on a faster machine.t- -|Thus, I'd like to stick to standard code.  t -|H -|The question of one large or several smaller files is mainly relevant I -|for the application which will USE the data, i.e. I could in principle  I -|calculate the stuff in several smaller files (which I would have to do  H -|if I split things up among several machines) and combine them later.  I -|Things would be easier, though, if one large file could be used in all G -|circumstances.    .    Herman D. (Skip) Knoble, Research Associate    Mailto:hdk@psu.edu8'    Web: http://www.personal.psu.edu/hdke     Center for Academic Computing    Penn State University      214C Computer Buildingb#      University Park, PA 16802-2101o.    Phone:+1 814 865-0818   Fax:+1 814 863-7049   ------------------------------   Date: 11 May 2000 15:10:07 GMT0 From: "Dale A. Dellutri" <ddellutr@enteract.com>" Subject: Re: big Fortran data file, Message-ID: <8feigf$2b5$1@news.enteract.com>  X On 11 May 2000 12:11:42 GMT, in comp.os.vms, Phillip Helbig <helbig@astro.rug.nl> wrote:J > I have a program where the bottleneck is computing a certain quantity asG > a function of 5 variables.  However, this quantity can be computed in I > advance at leisure.  I'm thinking of precomputing this and then readinghI > the file into the program and using a lookup table/interpolation rathert' > than computing the values on the fly.f  F It's hard to make any useful comments without more details.  There areE many approaches, including even the use of global sections, dependingi8 on exactly what is being done here.  So, some questions:  6 1. How often will you re-compute the data?  Just once,! forever?  Once a week? day? hour?s  = 2. How often will you run the application that uses the data?g  @ 3. What proportion of the data will it use when it runs?  All of  it?  Half?  A tiny amount (<5%)?  F 4. If it uses just a small part of the data, will subsequent runs tend4 the use data in the same area?  In a different area?  F We have an application that creates a database about once a month, butE there are many applications that use almost all of the data which runiB thousands of times per day.  We use a system-wide page-file globalA section for this.  I was going to suggest the same for you when In7 realized that I knew none of the details of your usage.i   -- h& Dale Dellutri -- ddellutr@enteract.com   ------------------------------  % Date: Thu, 11 May 2000 10:05:11 -0500 " From: Earl Lakia <lakia@ipact.com>" Subject: Re: big Fortran data file) Message-ID: <391ACC26.E33766CE@ipact.com>f  9 I would opt for a program that would map a global sections< with a backing RMS file.  Calculate all your number into theB global section and update the section to the file.  All subsequent9 users could then create & map the global section with the.@ RMS backing file.  With the introduction of the address variable5 in FORTRAN and C, mapping the global section is easy.e  A I have small C program routine that maps a global section using a ? RMS backing page file.  To create the file, use the  Create/fdl  something like the following:g  A IDENT   "27-JAN-1999 13:11:52   OpenVMS ANALYZE/RMS_FILE Utility"i   SYSTEM'         SOURCE                  OpenVMS    FILE#         ALLOCATION              275b#         BEST_TRY_CONTIGUOUS     yes-!         CLUSTER_SIZE            9e"         CONTIGUOUS              no!         EXTENSION               0 "         FILE_MONITORING         no!         GLOBAL_BUFFER_COUNT     0c@         NAME                    "bf_regions:iastatus_Region.pag"*         ORGANIZATION            sequential%         OWNER                   [1,1]nO         PROTECTION              (system:RWED, owner:RWED, group:RE, world:RWED)g   RECORD#         BLOCK_SPAN              yes /         CARRIAGE_CONTROL        carriage_returnr%         FORMAT                  fixedw#         SIZE                    512r    O /******************************************************************************t  File Name : MAP_IMAGE.C  Author    : M. Huttingerh  H  Purpose   : Opens an RMS file and maps it into the user's address space      Revision History:  =     10-JUL-96   M. Huttinger    Ported over from IQR Software     Arguments:d       name:   filename     type:   descriptor     access: by reference  #     Name of RMS file that is mappedg       name:   regionname     type:   descriptor     access: by reference       Name of the region to mapa       name:   va     type:   longword array     access: by reference  E     Pointer to the start of an eight longword array. Will contain the C     virtual address start/end, FAB and RAB start and end addresses.m       name:   access     type:   longword     access: by value  5     Access for mapping mode: 0=readonly, 1=read/writen       name:   system     type:   longword     access: by value  ?     System or group global section selection: 0=group, 1=systemo  
  Abstract:  B This routine opens an RMS file and maps it into the user's address space.  O *******************************************************************************s. * Header files, externals, macros, definitionsO ******************************************************************************/  #include <descrip.h> #include <rms> #include <starlet> #include <secdef>  #include <stdio>   struct mapping_array {C4   int region_va[2];     /* where region is mapped */.   int fabrab_va[2];     /* fab and rab v.a. */ };  O /******************************************************************************e * Global variablesO ******************************************************************************/r  O /******************************************************************************o * EntryeO ******************************************************************************/   0 int map_image(  struct dsc$descriptor *filename,2                 struct dsc$descriptor *regionname,*                 struct mapping_array *map,                 int access,n                 int system)I { O /******************************************************************************  * Local Variable DefinitionoO ******************************************************************************/s* int alloc_size; /* how much to allocate */: struct rabdef *rab;     /* pointer to RAB of image file */: struct fabdef *fab;     /* pointer to FAB of image file */3 struct namdef *nam;     /* pointer to NAMe block */I int status; 
 int flags; int space[2] = {0,0};i   /*+ * First find some space for the FAB and RABf */  # alloc_size= sizeof(struct rabdef) +c9                 sizeof(struct fabdef); /* size of each */   0                         /* round up to a page */) status=sys$expreg((alloc_size + 511)/512,.%                   &map->fabrab_va[0],i                   0,                   0);.  D if(status%2 == 0)return(status);  /* tell the fellow the bad news */   /*  * Now create the FAB and the RAB */  ( fab= (struct fabdef*) map->fabrab_va[0];8 rab= (struct rabdef*) (int) fab + sizeof(struct fabdef);  = *fab= cc$rms_fab;           /* a lot of runtime code to    */s9 *rab= cc$rms_rab;       /* initialize these structures */-   /* * Define FAB values  */   fab->fab$b_fac = FAB$M_BIO |                  FAB$M_PUT |.                  FAB$M_GET ; /* file access */  C fab->fab$l_fna = filename->dsc$a_pointer;  /* point to file name */sC fab->fab$b_fns = filename->dsc$w_length;   /* length of filename */uC fab->fab$l_fop = FAB$M_UFO;                 /* file open options */eE fab->fab$b_rfm = FAB$C_FIX;                 /* Record format fixed */n fab->fab$b_shr =                  FAB$M_UPI|u                  FAB$M_GET|t#                          FAB$M_PUT|s                  FAB$M_UPD|uC                  FAB$M_DEL;                /* file sharing flags */iE fab->fab$w_mrs=512;                            /* size of a record */h  D rab->rab$l_fab = fab;                      /* point to rab to fab */1 rab->rab$w_usz=512;     /* user buffer address */m3 rab->rab$l_rop= RAB$M_BIO;  /* Mark as block i/o */f   /* * Open the filei */   status = sys$open (fab, 0, 0); if(status%2 == 0)  { *   printf("  MAP_IMAGE> failure to open ");   return (status); }    /*# * Create and map the global sectionp */  M if(map->region_va[0] != 0)      /* User has somewhere to map this already? */q    flags= 0; else    flags= SEC$M_EXPREG;   C if(system != 0)                 /* System global or group global */6&    flags |= SEC$M_SYSGBL | SEC$M_GBL ; else    flags |= SEC$M_GBL;  : if(access != 0)flags |= SEC$M_WRT;      /* Write access */  ( status=sys$crmpsc(map,      /* inaddr */1                   map,              /* retaddr */ 5                   0,                /* access mode */n/                   flags,            /* flags */ 5                   regionname,       /* region name */6/                   0,                /* ident */d2                   0,                /* rel page */1                   fab->fab$l_stv,   /* channel */n4                   0,                /* page count */-                   0,                /* vbn */t.                   0,                /* prot */)               0);               /* pfc */k  
 #ifdef DEBUG_. if(status%2):     printf("\n  MAP_IMAGE> region mapped at: %08X, %08X ",0            map->region_va[0],map->region_va[1]); else/     printf("\n  MAP_IMAGE> failed: %d",status);  #endif   return (status);   }  /* Module end */i     Phillip Helbig wrote:e  J > I have a program where the bottleneck is computing a certain quantity asG > a function of 5 variables.  However, this quantity can be computed inHI > advance at leisure.  I'm thinking of precomputing this and then reading I > the file into the program and using a lookup table/interpolation ratheryI > than computing the values on the fly.  Especially useful, perhaps, if IeH > run the application more than once, but even in this minimal case it's > probably worth it. >sI > The application uses the values in a certain order and I can precomputeuG > them in this order.  Should I make one large file and let the virtualfH > memory system take care of the fact that it won't all fit into memory,J > keeping in mind that the values are accessed in the order they are used,H > or should I split it up into smaller files?  Should I used a formattedG > or unformatted file?  (Four places are enough precision, so the sizes 3 > would be about the same.)  Indexed or sequential?k >iG > This is with Compaq Fortran 95 on ALPHA/VMS.  At the moment, I have aeE > choice between a 255/233 with 64 MB or a 3000/600 with 192 MB.  I'mtG > assuming the processor speed, not the memory size, will determine thegH > speed if things are set up correctly with regard to the large file (orG > many smaller files, as the case might be).  The application otherwise @ > needs very little memory and does very little I/O, just number > crunching. >dI > We're talking about 1.5 GB or so of data, i.e. 360 million values to be  > computed.l >gJ > I asked a similar question a few months ago.  I've decided to ask again,H > since there didn't seem to be a consensus in the answers and since I'm4 > finally getting around to actually doing this now. >aI > If it turns out to be too slow, I might compute it on a faster machine.f+ > Thus, I'd like to stick to standard code.e >yG > The question of one large or several smaller files is mainly relevanteH > for the application which will USE the data, i.e. I could in principleH > calculate the stuff in several smaller files (which I would have to doF > if I split things up among several machines) and combine them later.H > Things would be easier, though, if one large file could be used in all > circumstances. >p > --O > Phillip Helbig                       Email .............. helbig@astro.rug.nl.O > Kapteyn Instituut                    Email ................. helbig@man.ac.uk O > Rijksuniversiteit Groningen          Tel. ................... +31 50 363 4067eO > Postbus 800                          Fax .................... +31 50 363 6100gO > NL-9700 AV Groningen                 Web ... http://www.astro.rug.nl/~helbig/r >m7 > My opinions are not necessarily those of my employer.r >cP > <A HREF=" http://gladia.astro.rug.nl:8000/helbig/hire/hire.html ">HIRE ME!</A>   --
 Earl D. Lakiad0 Senior Staff Engineer         Web: www.ipact.com4 Snail Mail:                   Email: lakia@ipact.com
 IPACT Inc.1 260 S. Campbell St.           Phone: 219-464-7212n Valparaiso, IN 46383   ------------------------------   Date: 11 May 2000 11:51 CSTp' From: carl@gerg.tamu.edu (Carl Perkins)d" Subject: Re: big Fortran data file- Message-ID: <11MAY200011515614@gerg.tamu.edu>a   helbig@astro.rug.nl writes...gG }In article <11MAY200008520469@gerg.tamu.edu>, carl@gerg.tamu.edu (Carly }Perkins) writes:  } H }>Additional speed improvements could be gotten by storing more than oneD }>value per record and using a subroutine to get the value you want  } J }Actually, the first four variables are stepped through sequentially, and F }the fifth variable is used in a more-or-less random order.  Thus, it   D The "random order" is a problem, then. This would be an argument forD breaking it into two files, one sequential access for the 4 that areC always used sequentially, and one direct access file for the other.   H }would make sense to put all of these values (between 1 and 599) in one   E Do you mean that the randomly used value only has up to 599 cases? IfsE so, then reading it in all at once into an array with 599 elements ise the way to go.  J }record.  Once this record is in memory, the computation time is probably G }comparable to the time to read the next record so I/O won't be a real e }bottleneck.  G I thought the idea was to get it to go *faster* than computing it everycF time? If so, you want to get the I/O time to less than the computationH time. If your goal was to free up some processor time for others to use,( then having them the same would be fine.  H }>This could result in less I/O overhead, although it may be possible toH }>get much the same effect simply by specifying a good set of options in: }>your OPEN statment, particularly a large "buffercount".  }  }Non-standard?  C It's a "standard" VAX extension from the F77 era. Since most modernmF F77 (and probably F90 - I don't know about F95) compilers support mostB of the VAX extensions, it is probably a common thing. It's hard toB say if any given implementation would have it. I was assuming thisC would run under OpenVMS (what with this being comp.os.vms and all),mG and both the old VAX F77 and newer Alpha F90 compilers have it (I don't > have F95 to check it, but it certainly has something similar).  C It enables "multibuffered I/O", i.e. the I/O reads multiple records|E in one fell swoop, then passes them to you one at a time as READs aree encountered.   }> You mightB }>also get some benefit from making it a fixed record length file. }  }Why?   H Two reasons, really. First: Less overhead in the I/O. It obviously takesH less processing to read fixed length records than it is to read variableK length records. (Not much, I'm sure - just using the first two bytes of theyG record to determine how much of the chunk of data read from the disk isnH part of the needed record instead of always using the same value. RMS isI also rather heavily optimised so it shouldn't be noticable unless you arefH reading a *lot* of records, which you are.) Second: It also uses 2 bytesJ less disk space per record than a variable record length file. If you wereL using 4 byte data, one data element per record, then the 2 bytes of variableG length data per record would increase the size of the file by 50%. More K sensibly for the above data, even at 16 bytes per record it is 12.5% extra.tG Since the data is naturally going to end up in uniform chunks, it makesiK sense to store it the same way - there is no variation in the record lengthy9 so you don't need to store it as variable length records.n  H (I just checked and notice that I almost never specify this even when itI is appropriate. Oops. Then again, I don't recall often having a data fileiL much above 2 orders of magnitude lower than the size you are talking about.)  G And, of course, there are other ways to do the whole thing like using a G memory mapped file or files, or resorting to asynchronous I/O so it can H be loading more data while you are doing the math with the current data.   --- Carl   ------------------------------  # Date: Thu, 11 May 2000 09:09:12 GMT," From: ejnw@bitmill.fi (Esa Jkl) Subject: CMUIP and FTP problemsr, Message-ID: <391a759f.11999555@news.inet.fi>  E I have been experiencing strange FTP corruptions between CMUIP (usingw= MadGoat FTP) and NT4+SP6a. Every 2000th character is missing!G? I have tried various combinations of RMS attributes but nothingc helps...  = On the VAX side I have: VMS V5.3, CMUIP V6.6-5A, NETLIB V2.2GC and MGFTP V2.6-1.C  F With NT4 + SP3 all works fine. Also, I can download files from another# MGFTP site without any corruptions.e  $ Any suggestions will be appreciated.   Esa-   ------------------------------  % Date: Thu, 11 May 2000 12:01:43 +0100D. From: Roger@natron.demon.co.uk (Roger Barnett) Subject: Compaq and CORBAg, Message-ID: <19974196wnr@natron.demon.co.uk>  0 From a recent OMG meeting sponsored by Compaq...  B "the zero latency enterprise demonstration featured a 111-terabyteD  real-time operational data store containing 100 billion call detail?  records ...  75,000 CORBA transactions per second - five times C  the combined call rate of the world's largest telecomms companies"   = "...using both Java and C++. It also simultaneously processedn=  customer service inquiries at 1000TPS, conducted data miningf=  with 128-way parallel queries, wrote summary data to severali?  data marts on remote platforms, published events and supportedw!  online database reorganisation."u  1 "utilized Compaq DOM and JORB [CORBA] middleware"?    @ Hmm, pretty impressive, so what platforms did they use (you know% where this is going, don't you ?) ...a  : "based upon massively parallel and scalable Compaq NonStop;  Himalaya servers, ProLiant servers, and Tru64UNIX servers, :  all connected by CORBA-conformant Object Request Brokers"    ( Still, VMS supports DCOM, so no worries.    9 Reference:    http://www.omg.org/cgi-bin/doc?omg/99-11-05c    @ Incidentally, the above demo/system was put together by Compaq'sB Business-Critical Server Division. The OMG audience will typicallyD have included technical architects from 100 or so major corporations and enterprise product vendors.n     -- f
 Roger Barnett    ------------------------------  # Date: Thu, 11 May 2000 11:38:44 GMT- From: erber@my-deja.coma Subject: CORBA on OpenVMS AXP0) Message-ID: <8fe643$miv$1@nnrp1.deja.com>u   Hello,  > I'd like to know, which implementations of CORBA are currentlyG supporting VMS. I already know, that BEAs Object Broker is running out.c   best regards   Jakob     & Sent via Deja.com http://www.deja.com/ Before you buy.o   ------------------------------  # Date: Thu, 11 May 2000 12:32:05 GMTs- From: Adam Dorosz <adam.dorosz@softax.com.pl>n! Subject: Re: CORBA on OpenVMS AXP - Message-ID: <391AA4F1.4DCE40CD@softax.com.pl>    erber@my-deja.com wrote:   > Hello, >a@ > I'd like to know, which implementations of CORBA are currently > supporting VMS.    OmniORB   '  http://www.uk.research.att.com/omniORB    regards9 Adam   ------------------------------  % Date: Thu, 11 May 2000 16:29:12 +0100 . From: Roger@natron.demon.co.uk (Roger Barnett)! Subject: Re: CORBA on OpenVMS AXPN- Message-ID: <303370446wnr@natron.demon.co.uk>   D In article: <8fe643$miv$1@nnrp1.deja.com>  erber@my-deja.com writes: > @ > I'd like to know, which implementations of CORBA are currentlyI > supporting VMS. I already know, that BEAs Object Broker is running out.   H BEA may no longer sell ObjectBroker as a separate product - if you look H at their Web site you'll see their business nowadays is in TPMs (Tuxedo)E and EJB servers (WebLogic). Of course ObjectBroker is still in there,A! but largely hidden from the user.r      I I am in the process of porting the current version of the Interbroker(R)  J ORB to Alpha/VMS. Once ready I will be making it available in (C++) sourceH with an unrestricted free user licence - there will also be a for-money ? licence+support package for those who can't use free software.    J Unfortunately the VAX version will have to wait until I can get access to H a machine that can run the C++ compiler at an acceptable speed - initialI experiments have shown that the trusty old VaxStation 3100 isn't quite upa to the job !  I Note that this product requires one of the various TCP/IP products to be nH installed before it can be used on a VMS system. The addition of DecNet  support awaits a sponsor  :)  D ( in terms of messaging protocols CORBA ORBs are only *required* to E   support IIOP, which in turn can only be run over TCP/IP - as far as E   I know there are no currently available ORBs which can be used withd   plain DecNet )   -- t
 Roger Barnett    ------------------------------  % Date: Thu, 11 May 2000 17:35:48 +0200   From: Paul Sture <paul@sture.ch>! Subject: Re: CORBA on OpenVMS AXPu+ Message-ID: <VA.00000040.04c52ac8@sture.ch>s  B In article <303370446wnr@natron.demon.co.uk>, Roger Barnett wrote: > F > In article: <8fe643$miv$1@nnrp1.deja.com>  erber@my-deja.com writes: > > B > > I'd like to know, which implementations of CORBA are currentlyK > > supporting VMS. I already know, that BEAs Object Broker is running out.c > J > BEA may no longer sell ObjectBroker as a separate product - if you look J > at their Web site you'll see their business nowadays is in TPMs (Tuxedo)G > and EJB servers (WebLogic). Of course ObjectBroker is still in there,*# > but largely hidden from the user.l > G I note that BEA are missing from the list of OpenVMS Software Partners * included in the "Dummies" book.* >  >    ___*
 Paul Sture Switzerland*   ------------------------------  % Date: Thu, 11 May 2000 09:50:27 -0400*( From: Ed Vogel <edward.vogel@compaq.com>: Subject: Re: DEC C, table mapping options across platforms* Message-ID: <391ABAA2.9D17BB11@compaq.com>  
     David,  K >Could we please have a single a table in the documentation/on the web site G >giving equivalent compiler switches on ALL the various platforms where*H >DEC/Compaq C (and Fortran) are available.  It would ease porting a bit. >Something like: >*0 >    Linux             Tru64             OpenVMS< >    -O3               -O3               /optimize=(level=2)3 >    -arch host        -arch host        /arch=hostr  E    This is a good suggestion.  I have added this to the wish list forkH    the Compaq C compiler.  I'll also forward this to the C++ and Fortran	    folks.     F >Even better would be completely portable option, something like this: > ; >linux> ccc -O3 -arch host -export_default compiler_options  > ) >and then on other platforms we could do:e >e6 >linux>    ccc -import_default compiler_options file.c6 >tru64>    ccc -import_default compiler_options file.c6 >OpenVMS$  cc/import_default=compiler_options   file.c >(Tandem too!) > ; >and get exactly, or as close as possible, the same result.e > E >Ideally "compiler_options" would be a text file (probably a bunch ofdK >#pragma statements). It would set the default state for the compiler, then 4 >the command line options, if any, would be applied.  =     There are really two parts to this request.  The first is E     to specify a file that is included before any source.  The second D     is for a set of pragmas that provide all the capabilities of the     command line options.n  J     The first almost exists today on Tru64 and Linux using the -FI switch.D     We plan to include similar functionality in our next release forI     OpenVMS.  The qualifier there will be /FIRST_INCLUDE.  I say "almost" E     because any pragmas that change the compiler environment would beB;     applied *after* the command line options are processed.M  D     The second partially exists.  There are pragmas that allow many,C     but certainly not all, options to be specified.  It has been onbC     our wish list for some time to add additional pragmas.  A major G     difficulty with adding the pragmas is deciding which syntax to use:oJ     match the Tru64 option spelling, match the OpenVMS qualifier spelling,J     or (most likely) something in between.  For example, enabling messages+     that detect potential coding errors is:a  (         Tru64:     -msg_enable questcode)         OpenVMS:   /WARN=ENABLE=QUESTCODEe3         pragma     #pragma message enable questcode   C     Adding pragmas to support all command line options will require6@     a lot of effort.  Over time we have added additional pragmasK     to specify command line options.  For example in Compaq C V6.2 we addedsG     a #pragma optimize directive that allows some optimization settingsr8     to be set within a file (and to change them from one     function to another).     D     Thank you for your feedback.  As you can see, we are making someJ     progress in providing some of the features you need.  We hope continue(     to make improvements in these areas.  0                                         Ed Vogel=                                         Compaq C Engineering.    ------------------------------  % Date: Thu, 11 May 2000 10:12:00 -0400p, From: Steve Lionel <Steve.Lionel@compaq.com>: Subject: Re: DEC C, table mapping options across platforms8 Message-ID: <fiflhss1cd4op8njcumksl9r71ms92o2hm@4ax.com>  A On 10 May 2000 20:03:23 GMT, mathog@seqaxp.bio.caltech.edu (David: Mathog) wrote:   >This is for the Compaq folks. >oK >Could we please have a single a table in the documentation/on the web sitelG >giving equivalent compiler switches on ALL the various platforms whereoH >DEC/Compaq C (and Fortran) are available.  It would ease porting a bit. >Something like:   >s0 >    Linux             Tru64             OpenVMS< >    -O3               -O3               /optimize=(level=2)3 >    -arch host        -arch host        /arch=hosti >eJ >This comes up because I've found on Linux that a particular program worksN >fastest when compiled with -O3 -arch host, but the VMS equivalent isn't clearG >since /optimize=(level=5) is the top on VMS, but the Unices only go to  >-O4.   F That's because you have old UNIX versions.  The optimization levels onD our compilers (at least for Fortran - I'm not 100% sure about C) are@ the same on all our platforms and go to 5 on UNIX and Linux (and	 Windows).s  F For Fortran, we try hard to keep the switch names and options the same@ across platforms.  However, where on VMS we have qualifiers withB keyword lists, UNIX doesn't have the concept so there are separateE switches.  Also, some UNIX switches which are cryptic, sucn as -fpe3,1 have meaningful names on VMS.o  E I think the idea of a table like this is good - we'll see what we can  do on the Fortran side.   F >Even better would be completely portable option, something like this: >h; >linux> ccc -O3 -arch host -export_default compiler_options  >y) >and then on other platforms we could do:  >e6 >linux>    ccc -import_default compiler_options file.c6 >tru64>    ccc -import_default compiler_options file.c6 >OpenVMS$  cc/import_default=compiler_options   file.c >(Tandem too!) >s; >and get exactly, or as close as possible, the same result.   A Urgh.  Not too likely, unfortunately.  A complication is that thebC defaults are different on various platforms.  We'd need to add somee7 sort of "translator" step to the import/export process.a  F >Ideally "compiler_options" would be a text file (probably a bunch of K >#pragma statements). It would set the default state for the compiler, thenv5 >the command line options, if any, would be applied. l  E We already have this for Fortran on UNIX and are looking to extend itu to the other platforms.o    - Steve Lionel (mailto:Steve.Lionel@compaq.com). Fortran Engineering & Compaq Computer Corporation, Nashua NH  6 Compaq Fortran web site: http://www.compaq.com/fortran   ------------------------------   Date: 11 May 2000 14:51:34 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman): Subject: Re: DEC C, table mapping options across platforms6 Message-ID: <8fehdm$mua$1@mailint03.im.hou.compaq.com>  a In article <8fcfab$c7k@gap.cco.caltech.edu>, mathog@seqaxp.bio.caltech.edu (David Mathog) writes:lK :Could we please have a single a table in the documentation/on the web sitebG :giving equivalent compiler switches on ALL the various platforms wherenH :DEC/Compaq C (and Fortran) are available.  It would ease porting a bit.  E   One of my pet projects lurking on the back burner is a "Porting to nH   OpenVMS" manual.  (The best I can currently offer in this area is fromE   the Tru64 UNIX porting manual set, and a manual that I hestiate to bH   mention in this context.)  The compiler switches would be just a part (   of this documentation work, of course.  G   The downside of working on the porting manual right now involves its tJ   expected and imminent obsolescence with the shipping of the COE-related #   work currently under development.a  I   Funny you should mention this.  I have expressed a need for this manualeC   to the OpenVMS product management staff as recently as yesterday.   I   Unfortunately, I can't currently get you a date nor any commitment for ,H   the availability of the manual.  (I can certainly accept and stage any9   suggestions for the contents of the manual, of course.)   N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Thu, 11 May 2000 10:17:31 -0700n5 From: "Larry D Bohan, Jr" <LBohan@dbc.spam_less..com>.: Subject: Re: DEC C, table mapping options across platforms2 Message-ID: <2+kaOZCfTiy2Fmsgbutqr2kp2zb9@4ax.com>  @ On 11 May 2000 14:51:34 GMT, hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) wrote:*  H >  The downside of working on the porting manual right now involves its K >  expected and imminent obsolescence with the shipping of the COE-related r$ >  work currently under development.  # COE,  Common Operating Enviroment? c1 ie,  http://diicoe.disa.mil/coe/faq/faq_page.htmle  " or something completely different?   ------------------------------  % Date: Thu, 11 May 2000 12:00:57 -0400t; From: "Carlc Internet Services" <carlc@iname-nospammen.com>a Subject: DS700 question...4 Message-ID: <zVAS4.2535$dQ.45745995@news.axxsys.net>   Hi COMP.OS.VMS,a  I     I'm trying to use a fax software connected to a multitech modem via am LATs connection on ourI vax. When I use the Emulex NETQUE server, it works fine... When I use theu DS700, it doesntF work correctly (I'm getting lines in the faxes). I've checked the port setups, and they look as* identical as an Emulex to Digital can get.  !     The DS700 is at this version:.  5 DECserver 700-08 V1.1C BL46A-13  LAT V5.1  ROM V3.4-9l  L     It is to old or maybe there is a new version that can allow it to handle the 19200 baud correctly?Y       Thanks in advance.-     Carl C. (ccouric at fhp dash mfg dot com)l   ------------------------------  # Date: Thu, 11 May 2000 10:55:40 GMTt From: "" <gsxr7@core.com>e3 Subject: free credit card machine for your business 0 Message-ID: <36657.2880650463@news.corecomm.net>  ? Advantage Banking Services is giving a free credit card machineoB with a printer to all new accounts. Sign up with Advantage Banking@ Services and receive a new credit card machine with a printer to% use for free for as long as you like.b? We are also ofering a Web Link/Virtual Terminal package for alle? web site based businesses. This package enables your businessesiB web site to accept and process credit card sales automatically andH in real time. With this package your web site will be able to run itselfB while serving any number of clients simultaneously. This lets your web site work for you.C All of our equipment is brand new and state of the art. Whether youoC are a web site based business or a retail merchant we have the card-! sale processing solution for you.d9 Do not buy a machine for $400-$1,200 or get tied in to anrA expensive four year lease for your equipment. Use our machine fortF free for as long as you like, it's just that simple. Advantage BankingD Services is changing the credit card machine industry by eliminating! equipment purchasing and leasing.,A The time is right to enable your business to accept credit cards. . Your competition accepts credit cards, do you? We offer excellent rates: & 1.75% retail (card present and swiped)< 2.20% Internet/mail-order/telephone-order (card not present)B Applying is free and easy. To receive your application or to learn> more about what Advantage Banking Services can do to help your5 business grow call our offices today at 440-885-0896. 8 Advantage Banking Services provides banking solutions to8 business because money matters. If your money maters use Advantage Banking Services.c   ------------------------------  % Date: Thu, 11 May 2000 12:56:37 +0200a= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com> 8 Subject: Re: ftp.decus.org is back up!  And running VMS!) Message-ID: <391A91E4.9FCA3D90@gtech.com>u   Dirk Munk wrote:H > Does this mean there are FTP clients and/or servers that will copy anyG > VMS file exactly the way it is? And if so, what happens when both VMSo- > systems have different FTP servers/clients?g  7 Sometimes it is possible to transfer a file between two + VMS system keeping all the file attributes.a  / UCX uses something called +VMS+ while the othery  FTP implementation use STRU VMS.  . If they can agree on a protocol then it works.   Arne   ------------------------------  % Date: Thu, 11 May 2000 08:46:52 -0400t+ From: Tim Shoppa <shoppa@trailing-edge.com>.8 Subject: Re: ftp.decus.org is back up!  And running VMS!1 Message-ID: <391A737C.14FBC745@trailing-edge.com>e   Dirk Munk wrote: >  > Jeff Schreiber wrote:rO > >     Unfortunately their FTP client and Server use a format for transferringnM > >     vms file attributes that is different from everyone else.  Some otherRO > >     stacks understand the TCP/IP services format [+VMS+], but they all seeme# > >     to understand STRU O VMS...t > H > Does this mean there are FTP clients and/or servers that will copy anyG > VMS file exactly the way it is? And if so, what happens when both VMSd- > systems have different FTP servers/clients?   ? FTP.DECUS.ORG is running the MadGoat FTP server, which supports/C STRU VMS for full VMS filetype support.  If your current FTP cliente? doesn't support STRU VMS, you can (in all cases that I know of)DF simply install the freeware MadGoat FTP client and take full advantageE of it.  MadGoat FTP, through the NETLIB layer, runs on all VMS TCP/IPn stacks.f  @ I know I've tried to be a muckraiser in the past with respect to? FTP.DECUS.ORG's functionality, but as far as I'm concerned it'soB now perfect.  I can fetch all the VMS-specific DECUS files from itE without them having been mangled through a Windows NT or Unix system.pB And I think you can too, Dirk.  I'm unwilling to do anything whichA changes it from its now "perfect" state, though what you consideroC as "perfect" may be different (it seems you really like DECNET over1= TCP/IP, which is something I'm not knowledgable of.)  If this.B functionality can be added without screwing up its current "pefectF by me" state, then great, it can be added.  But I'm very happy with it	 as it is.    Tim.   ------------------------------  # Date: Thu, 11 May 2000 14:12:11 GMT * From: "David Cressey" <david@dcressey.com>8 Subject: Re: ftp.decus.org is back up!  And running VMS!5 Message-ID: <%czS4.487$s4.49075@petpeeve.ziplink.net>n  @ I don't know if this is helpful or not,  but I succeeded once inH reconstituting a BACKUP save set,  that had been FTP'd to a Unix system,E stored there for redistribution,  and then FTP'd to another VMS site.c  G The file, as received had RMS records that were fixed length, 512 bytes  long.A  K It was a matter of multiplying 63 by 512,  and using that number to set thep
 RMS recordL length attribute in the received file.  63 is some sort of magic number,  by default, for BACKUP.  K Figuring out how to reset the RMS record length attribute without writing adH program involved some fairly tedious and obscure research in DCL and RMS documentation.  L Once the change was made,  presto!  BACKUP/LIST showed an error free saveset file,  and BACKUPeH was perfectly happy restoring the files.  I did a few trials with  files( that I could compare with the originals,G and in all the tests I did,  not one bit was lost!  Of course,  I can't ' guarantee you that 63 blocks per recordeD is universal among BACKUP savesets.  It isn't.  But it's widespread.   I hope this helps someone.   David Cressey, consultante david@dcressey.com      K Hoff Hoffman wrote in message <8fccq0$dqh$1@mailint03.im.hou.compaq.com>...i >aI >In article <39199AA3.90AFA580@home.nl>, Dirk Munk <munk@home.nl> writes:rH >: This connection is a full decnet connection, and it is not limited to >: just copying. >hL >  In a recent conversation with a customer, I received a report of problemsJ >  connecting to a DFS disk via DECnet-Plus over IP -- this does not work,I >  per the report.  (I passed the problem report along to the engineering:K >  team.)  This DFS connection approach would be one obvious way to provide K >  access to the disk (were it to work, of course), as would configuring anrJ >  NFS server or DECnet-Plus over IP and FAL -- but each of these requiresC >  rather more than just the current FTP-only access to the server.w >hJ >  (And no, I haven't seen the earlier postings in this thread -- I'm sureL >  somebody has already suggested this protocols, so please ignore this. :-) >BD >  (And no, I'm not in a position to get DECnet-Plus nor DFS nor NFS
 configuredJ >  at the Compaq ECO website -- I'm told that configuration is expected to be7 >  unlikely to ever happen.  FTP only, in other words.)  > + > --------------------------- pure personalt# opinion ---------------------------*0 >   Hoff (Stephen) Hoffman   OpenVMS Engineering hoffman#xdelta.zko.dec.com >-   ------------------------------   Date: 11 May 2000 15:17:42 GMT1 From: JONESD@er6.eng.ohio-state.edu (David Jones)c8 Subject: Re: ftp.decus.org is back up!  And running VMS!: Message-ID: <8feium$b3t$1@charm.magnus.acs.ohio-state.edu>  5 In message <%czS4.487$s4.49075@petpeeve.ziplink.net>,n.   "David Cressey" <david@dcressey.com> writes:L >It was a matter of multiplying 63 by 512,  and using that number to set the >RMS recorduM >length attribute in the received file.  63 is some sort of magic number,  by  >default, for BACKUP.3  G 32256 is the default block size for save sets created on ODS-x volumes, M the default is 8192 bytes for save sets created on tapes.  The first block ofhL the save set contains a copy of the command line so a non-default block size0 can be determined by using "dump/block=count:1".   >dL >Figuring out how to reset the RMS record length attribute without writing aI >program involved some fairly tedious and obscure research in DCL and RMSw >documentation.   K Yes, I can never remember whether it's /attrib=(rfm:fix,lrl:32556) (logicaldG record length) or /attrib=(rfm:fix,mrs:32256) (max. record size) on thei "set file" command.m      < David L. Jones               |      Phone:    (614) 292-6929- Ohio State University        |      Internet:aL 140 W. 19th St. Rm. 231a     |               jonesd@er6s1.eng.ohio-state.edu: Columbus, OH 43210           |               vman+@osu.edu  + Disclaimer: Dogs can't tell it's not bacon.l   ------------------------------   Date: 11 May 2000 17:28:07 GMT& From: brown@taiga.gmcl.com (Rob Brown)8 Subject: Re: ftp.decus.org is back up!  And running VMS!, Message-ID: <8feqj7$nhi$1@burn.ab.videon.ca>  ) David Cressey (david@dcressey.com) wrote:u <snip>M : It was a matter of multiplying 63 by 512,  and using that number to set the  : RMS recordN : length attribute in the received file.  63 is some sort of magic number,  by : default, for BACKUP.  ? Yes, it almost always is 32256.  The easy way to find out is to      $ BACKUP/LIST <filename>/SAVEs  J This prints out the save set information, including the block size, before it blows up with INVRECSIZ.t  M : Figuring out how to reset the RMS record length attribute without writing adJ : program involved some fairly tedious and obscure research in DCL and RMS : documentation.  # The command, as you discovered, is:l  7   $ SET FILE/ATTRIBUTES=LRL=<block size>/LOG <filename>k   - Robt     -- --  / Rob Brown                        brown@gmcl.comc6 G. Michaels Consulting Ltd.      (780)438-2101 (voice)4 Edmonton                         (780)437-3367 (FAX)1                                  http://gmcl.com/    ------------------------------  % Date: Thu, 11 May 2000 09:26:02 +0200e% From: "Fred Zwarts" <F.Zwarts@KVI.nl>  Subject: Re: Internal date. Message-ID: <8fdna9$nrn$1@info.service.rug.nl>  8 "David Turner" <d_b_turner@yahoo.com> wrote in message =( news:shjsfdf16fc22@corp.supernews.com...I > What happened on that date((17-Nov-1858 being the OpenVMS system base =i date)n8 > to make Digital start the VMS date then...???!?!?!?!?! >=20 > Davidk >=20 ------------  A  38 Why Is Wednesday November 17, 1858 The Base Time For VAX/VMS?   C COMPONENT:  SYSTEM TIME                   OP/SYS:  VMS, Version 4.n   # LAST TECHNICAL REVIEW:  06-APR-1988   0 SOURCE: Customer Support Center/Colorado Springs      	 QUESTION:-  > Why is Wednesday, November 17, 1858 the base time for VAX/VMS?     ANSWER:l  @ November 17, 1858 is the base of the Modified Julian Day system.  F The original Julian Day (JD) is used by astronomers and expressed in = daysG since noon January 1, 4713 B.C.  This measure of time was introduced byTC Joseph Scaliger in the 16th century.  It is named in honor of his =b father, G Julius Caesar Scaliger (note that this Julian Day is different from theL< Julian calendar named for the Roman Emperor Julius Caesar!).  H Why 4713 BC?  Scaliger traced three time cycles and found that they wereH all in the first year of their cyle in 4713 B.C.  The three cycles are = 15, H 19, and 28 years long.  By multiplying these three numbers (15 * 19 * 28J =3D 7980), he was able to represent any date from 4713 B.C. through 3267 = A.D.E The starting year was before any historical event known to him.  In =n fact,rH the Jewish calendar marks the start of the world as 3761 B.C.  Today hisI numbering scheme is still used by astronomers to avoid the difficulties =v ofF converting the months of different calendars in use during different = eras.a  H So why 1858?  The Julian Day 2,400,000 just happens to be November 17, = 1858.a3 The Modified Julian Day uses the following formula:h      MJD =3D JD - 2,400,000.5v  G The .5 changed when the day starts.  Astronomers had considered it moretJ convenient to have their day start at noon so that nighttime observation = timespG fall in the middle.  But they changed to conform to the commercial day.o  F The Modified Julian Day was adopted by the Smithsonian Astrophysical = Obser-D vatory (SAO) in 1957 for satellite tracking.  SAO started tracking =
 satellitesH with an 8K (non-virtual) 36-bit IBM 704 computer in 1957, when Sputnik = waspD launched.  The Julian day was 2,435,839 on January 1, 1957.  This isG 11,225,377 in octal notation, which was too big to fit into an 18-bit =e fieldcH (half of its standard 36-bit word).  And, with only 8K of memory, no oneH wanted to waste the 14 bits left over by keeping the Julian Day in its = ownfI 36-bit word.  However, they also needed to track hours and minutes, for =  whichmG 18 bits gave enough accuracy.  So, they decided to keep the number of =t days inaF the left 18 bits and the hours and minutes in the right 18 bits of a = word.r  I Eighteen bits would allow the Modified Julian Day (the SAO day) to grow =e asI large as 262,143 ((2 ** 18) - 1).  From Nov. 17, 1858, this allowed for =F sevenoF centuries.  Using only 17 bits, the date could possibly grow only as = large asD 131,071, but this still covers 3 centuries, as well as leaving the = possibility,H of representing negative time.  The year 1858 preceded the oldest star = catalog-I in use at SAO, which also avoided having to use negative time in any of =z thec  satellite tracking calculations.  J This base time of Nov. 17, 1858 has since been used by TOPS-10, TOPS-20, = and0@ VAX/VMS.  Given this base date, the 100 nanosecond granularity = implementedaG within VAX/VMS, and the 63-bit absolute time representation (the sign =e bit must6 be clear), VMS should have no trouble with time until:      31-JUL-31086 02:48:05.47a  F At this time, all clocks and time-keeping operations within VMS will = suddenly( stop, as system time values go negative.  G Note that all time display and manipulation routines within VMS allow =  for H only 4 digits within the 'YEAR' field.  We expect this to be corrected = in: a future release of VAX/VMS sometime prior to 31-DEC-9999.   <NO MORE TEXT>   --   ------------------------------  % Date: Thu, 11 May 2000 12:49:47 +0200n= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>  Subject: Re: Internal date) Message-ID: <391A904B.A2D83070@gtech.com>e   David Turner wrote:dM > What happened on that date((17-Nov-1858 being the OpenVMS system base date)o8 > to make Digital start the VMS date then...???!?!?!?!?!  - It is a base date, that comes from astronomy.c   Arne   ------------------------------  % Date: Thu, 11 May 2000 14:39:14 +0100a* From: "Richard Brodie" <R.Brodie@rl.ac.uk> Subject: Re: Internal date, Message-ID: <8fed63$175s@newton.cc.rl.ac.uk>  [ "Arne Vajhj" <arne.vajhoej@gtech.com> wrote in message news:391A904B.A2D83070@gtech.com...  > David Turner wrote: O > > What happened on that date((17-Nov-1858 being the OpenVMS system base date) : > > to make Digital start the VMS date then...???!?!?!?!?! >m/ > It is a base date, that comes from astronomy.p  I If you fire up the DECwindows calendar and read the help, it is discussed  there at some length.    ------------------------------   Date: 11 May 2000 14:58:34 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) Subject: Re: Internal date6 Message-ID: <8fehqq$mua$2@mailint03.im.hou.compaq.com>  \ In article <shjsfdf16fc22@corp.supernews.com>, "David Turner" <d_b_turner@yahoo.com> writes:L :What happened on that date((17-Nov-1858 being the OpenVMS system base date)7 :to make Digital start the VMS date then...???!?!?!?!?!   )   Time to add another section to the FAQ.e  C   As to the question, where is what DECwindows Calendar has to say:g    "OpenVMS and the Julian Perioda  9    Why OpenVMS regards November 17, 1858 as the beginning 
    of time...t  7    The modified Julian date adopted by SAO (Smithsonian.B    Astrophysical Observatory) for satellite tracking is Julian Day4    2400000, which turns out to be November 17, 1858.  A    SAO started tracking satellites with an 8K (nonvirtual) 36-bitn;    IBM 704 in 1957 when Sputnik went into orbit. The Julian >    day was 2435839 on January 1, 1957. This is 11225377 octal,>    which was too big to fit into an 18-bit field. With only 8K>    of memory, the 14 bits left over by keeping the Julian date;    in its own 36-bit word would have been wasted. They alsolA    needed the fraction of the current day (for which 18 bits gave4<    enough accuracy), so it was decided to keep the number ofB    days in the left 18 bits and the fraction of a day in the right    18 bits of one word.k  >    Eighteen bits allows the truncated Julian day (the SAO day)<    to grow as large as 262143, which from November 17, 1858,A    allowed for 7 centuries. Possibly, the date could only grow asrE    large as 131071 (using 17 bits), but this still covers 3 centuriesr@    and leaves the possibility of representing negative time. The>    1858 date preceded the oldest star catalogue in use at SAO,?    which also avoided having to use negative time in any of thew$    satellite tracking calculations."    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  / Date: Thu, 11 May 2000 08:50:51 +0200 (MET DST)o& From: Rudolf Wingert <win@fom.fgan.de>* Subject: Is "The GNU on VMS Project" dead?/ Message-ID: <200005110552.HAA31428@fom.fgan.de>a   Hello,  H I was on the official "The GNU on VMS Project" WEB page and did see thatJ the last change was at 1997. Three year ago it a long time. Is the project dead?a   TIA and regards Rudolf Wingert   ------------------------------  # Date: Thu, 11 May 2000 12:14:23 GMTc' From: Colin Blake <colin@theblakes.com>l. Subject: Re: Is "The GNU on VMS Project" dead?- Message-ID: <391AA41C.D6D299C2@theblakes.com>   ) Have you seen http://gnv.sourceforge.net/.   ------------------------------  % Date: Thu, 11 May 2000 07:15:30 -0500 ( From: yyyc186.illegaltospam@flashcom.net0 Subject: Re: Kind of portable Alpha, VMS capable9 Message-ID: <391aa4a4$1$lllp186$mr2ice@news.flashcom.com>s  3 In <oMICKujZPHcE@tachxxsoftxxconsult>, on 05/11/00 tG    at 07:15 AM, wayne@tachysoft.xxx.302774.killspam.0140 (Wayne Sewell)  said:0  G >This one is a little big for that type of thing anyway, with a 15-inchcF >screen and what appears to be a full keyboard.  It's definitely not aC >laptop, unless the lap belongs to the fat lady at the circus.  :-)e  I I have a Pentium notebook with 15.1" active matrix and full keyboard withyH numeric keypad on the side.  While it is awesome in the hotel, I miss my3 old  8" 486 running DOS when on a plane or a train.b   Roland   -- f; -----------------------------------------------------------aD yyyc186@flashcom.net              To Respond delete ".illegaltospam"6                             MR/2 Internet Cruiser 1.528                             For a Microsoft free univers; -----------------------------------------------------------a   ------------------------------  % Date: Thu, 11 May 2000 10:57:01 -0500M) From: "John E. Malmberg" <wb8tyw@qsl.net>-. Subject: Length of Username - Was RE: Username. Message-ID: <shlln3bikcp12@corp.supernews.com>  3 Hoff Hoffman <hoffman@xdelta.zko.dec.nospam> wrote:   > :The username field length is 12 characters on both platforms,? : but I would tend to reserve a 32 character field when this is] : feasible.l  H Is there a compile or link time symbol that can be used to determine the length of the username?r  H Some programs, such as SAMBA use Geoff Fitch's SYS_SETUSER to change the@ effective username of a process by writing the desired name into1 JIB$T_USERNAME[PCB$L_JIB(R4)] and CTL$T_USERNAME.,  K Since these programs must be linked to the specific version of VMS they arecH running on, a link time symbol would allow them to take advantage of any future enhancement.e   -JohnG wb8tyw@qsl.network   ------------------------------   Date: 11 May 2000 16:28:42 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)2 Subject: Re: Length of Username - Was RE: Username6 Message-ID: <8fen3q$pbv$3@mailint03.im.hou.compaq.com>  Z In article <shlln3bikcp12@corp.supernews.com>, "John E. Malmberg" <wb8tyw@qsl.net> writes:4 :Hoff Hoffman <hoffman@xdelta.zko.dec.nospam> wrote: : ? ::The username field length is 12 characters on both platforms,h@ :: but I would tend to reserve a 32 character field when this is :: feasible. :bI :Is there a compile or link time symbol that can be used to determine thei :length of the username?  
   Usually.  I :Some programs, such as SAMBA use Geoff Fitch's SYS_SETUSER to change theaA :effective username of a process by writing the desired name into 2 :JIB$T_USERNAME[PCB$L_JIB(R4)] and CTL$T_USERNAME.  F   This is dangerous practice, and -- as I have reported before -- veryE   few of these tools get this "right".  I know, I wrote one of these.-F   (And things get _very_ interesting with this patch-the-kernel schemeJ   and the per-threads personnas that are found on OpenVMS V7.2 and later.)  L :Since these programs must be linked to the specific version of VMS they areI :running on, a link time symbol would allow them to take advantage of anyh :future enhancement.  C   This approach is not always trustworthy.  V7.2 made changes in a i3   closely-related area of the kernel, for instance.u  B   Best approach is to use the personna services whenever feasible.  @   Failing that, see UAF$S_USERNAME, JIB$S_USERNAME, and friends.A   Most of the data structures have size values, though I have haddE   a few cases in a few data structures where I have needed to resort  E   to the "offsetof" approach and some ASSUME statements (to "squawk" C7   if any incompatible data structure changes are made).e  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Thu, 11 May 2000 12:11:37 -0500s) From: "John E. Malmberg" <wb8tyw@qsl.net> 2 Subject: Re: Length of Username - Was RE: Username/ Message-ID: <shlq3080kcp118@corp.supernews.com>a  = Hoff Hoffman <hoffman@xdelta.zko.dec.nospam> wrote in message00 news:8fen3q$pbv$3@mailint03.im.hou.compaq.com... > C > In article <shlln3bikcp12@corp.supernews.com>, "John E. Malmberg"j <wb8tyw@qsl.network> writes: >c= > :Some programs, such as SAMBA use Geoff Fitch's SYS_SETUSER ? > :to change the effective username of a process by writing thet6 > :desired name into JIB$T_USERNAME[PCB$L_JIB(R4)] and > :CTL$T_USERNAME. >nH >   This is dangerous practice, and -- as I have reported before -- veryG >   few of these tools get this "right".  I know, I wrote one of these.iH >   (And things get _very_ interesting with this patch-the-kernel schemeL >   and the per-threads personnas that are found on OpenVMS V7.2 and later.)  J Ok, I will restructure the setuid() emulation on SAMBA 2.0.6 so it can use. the persona services when it installed on 7.2.  L For VMS versions 5.5-2 through 7.1, is there a better supported method for aE server process to temporarily change it's effective username and UIC?e   -Johno wb8tyw@qsl.network   ------------------------------  # Date: Thu, 11 May 2000 14:00:49 GMTd From: Ed.Wilts@merrillcorp.com Subject: LMF weirdness) Message-ID: <8feee6$vu3$1@nnrp1.deja.com>o  < I'm getting stumped by some licensing issues (technical, notB business).  I've got some VMSCLUSTER and VOLSHAD licenses that areD getting loaded on systems that they're not supposed to be loaded on.G Specifically, all of these licenses have a /INCLUDE that is supposed to3G restrict the license to a specific hosts.  However, one host is loadingtD ALL the licenses, overriding the /INCLUDE.  A SHOW LICENSE will showG one host with 4400 units loaded, even though it's only supposed to havetG 1100.  When I try to load my own license on another node, I then end upt@ exceeding my license units (%LICENSE-F-EXCEEDED, attempted usageC exceeds active license limits).  The only way around this that I've F found is to modify each of licenses with a /NO_SHARE, which should not be required.  @ Have I overlooked something obvious, or is this what I should beD doing?  In theory, I should not even need to do the /INCLUDE and the units should magically add up.  D This is on a pair of systems running OpenVMS Alpha 7.1-2 with Update
 V3.0 applied.C   Thanks, 	    .../Edi   mailto:Ed.Wilts@merrillcorp.comh    & Sent via Deja.com http://www.deja.com/ Before you buy.t   ------------------------------  % Date: Thu, 11 May 2000 10:04:55 -0500m) From: "John E. Malmberg" <wb8tyw@qsl.net>| Subject: Re: LMF weirdness7 Message-ID: <198601bfbb5a$46e1ff10$020a0a0a@xile.realm>.  H Use the /EXCLUDE option to prevent SIP license keys from being loaded on nodes that they should not be.  H I do not understand all of the reasons why, but that is the only thing I have found to work.-   -John  wb8tyw@qsl.network   ------------------------------   Date: 11 May 2000 16:19:46 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) Subject: Re: LMF weirdness6 Message-ID: <8femj2$pbv$2@mailint03.im.hou.compaq.com>  J In article <8feee6$vu3$1@nnrp1.deja.com>, Ed.Wilts@merrillcorp.com writes:= :I'm getting stumped by some licensing issues (technical, not C :business).  I've got some VMSCLUSTER and VOLSHAD licenses that areiE :getting loaded on systems that they're not supposed to be loaded on.PH :Specifically, all of these licenses have a /INCLUDE that is supposed to+ :restrict the license to a specific hosts. I  D   If you have more than one license database around, make sure that C   every license PAK is registered (the same way) in every database.u  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Thu, 11 May 2000 17:35:48 +0200s  From: Paul Sture <paul@sture.ch> Subject: Re: LMF weirdness+ Message-ID: <VA.00000041.04c52d47@sture.ch>   1 In article <8feee6$vu3$1@nnrp1.deja.com>,  wrote:-  > From: Ed.Wilts@merrillcorp.com > Newsgroups: comp.os.vmsr > Subject: LMF weirdness% > Date: Thu, 11 May 2000 14:00:49 GMT  > > > I'm getting stumped by some licensing issues (technical, notD > business).  I've got some VMSCLUSTER and VOLSHAD licenses that areF > getting loaded on systems that they're not supposed to be loaded on.I > Specifically, all of these licenses have a /INCLUDE that is supposed to!I > restrict the license to a specific hosts.  However, one host is loading7F > ALL the licenses, overriding the /INCLUDE.  A SHOW LICENSE will showI > one host with 4400 units loaded, even though it's only supposed to have I > 1100.  When I try to load my own license on another node, I then end up B > exceeding my license units (%LICENSE-F-EXCEEDED, attempted usageE > exceeds active license limits).  The only way around this that I've.H > found is to modify each of licenses with a /NO_SHARE, which should not > be required. > B > Have I overlooked something obvious, or is this what I should beF > doing?  In theory, I should not even need to do the /INCLUDE and the  > units should magically add up. > F > This is on a pair of systems running OpenVMS Alpha 7.1-2 with Update > V3.0 applied.C > G Are they same processor power or different models? Different licensing eJ schemes, or different number of users per node? That one host should have E 1100 units and is grabbing 4400 units indicates this may be the case.(  I We had a real problem with licensing a couple of years ago when we tried oE adding a 4100 to a 2 x 8400 cluster. The two 8400s had quite happily wG shared the license units beforehand, but adding that 4100 gave us real d grief.  F I peeked into the LMF today and found that we had tied each and every I affected license to the relevant nodes by a combination of /NO_SHARE and  F /INCLUDE. I didn't check all the licenses out today, but IIRC it also 1 applied to at least some of the layered products.d ___a
 Paul Sture Switzerlandf   ------------------------------  # Date: Thu, 11 May 2000 07:31:58 GMTt0 From: carlini@true.lkg.dec.com (Antonio Carlini)+ Subject: Re: Looking for infor for 4000/200n* Message-ID: <8fdn2m$qbn@usenet.pa.dec.com>  W In article <39198F5F.1755B88@laughnet.net>, Stephen Henry <stephen@laughnet.net> wrote: H >    I have a total of 10 terminals, but only know how to hook up one toG >each machine. I think I'm missing a part or something.  (The terminalso@ >don't have any DB25 ports on the back).  The system came with 4I >DECserver 700s but they all have RJ45 ports on them (16).  Any idea what5  >these would have been used for?  M I'm not clear what you know or don't know here so some of this may not help. dM The DECserver 700 is a terminal server that allows terminals atatched  to it iK to connect to networked systems that support the LAT protocol. Think of it tL like Telnet but not routeable (actually, the 700 also supports Telnet so if 1 you have UCX on your VAX you can telnet into it).e  L Before RJ45 there was DECconnect. This was a system that used a 6 conductor M flat cable and Modified Modular Jack connectors - 6 wires and an offset lug. gK Since your terminals do not have DB25 you need a cable that goes from your oM terminal to the RJ45 on the DECserver. You don't say what the terminals are,  M but if theyare VT420 then they have MMJ connectors, in which case you need a bL converter plug to go from MMJ to RJ45. Unless your VAX 4000-200 came with a O CXY08 or similar card, then a network-connected terminal will be your only way y in (other than the console).  G >    And, the biggest question of all.....The system came with 5 drivesTF >and an r400x cabinet to hold them.  One drive boots openVMS 6.1, goesH >through a bunch of stuff and comes to the login prompt.  Unfortunatly II >don't have an account, but I don't want to wipe the drives either, sincer   Take a look at the OpenVMS FAQ  >         http://www.openvms.digital.com/wizard/openvms_faq.html  K MGMT5 answers this particular question but the whole thing is worth a read..O On the VAX 4000-200, the console connection is the MMJ socket on the CPU panel  N (right most metal panel when looking from the front, below the power switch). G Since you have seen OpenVMS boot, I assume you have this all hooked up.X    F >I don't have openVMS to reinstall.  Two of the other drives also bootH >openVMS 6.1, but they don't go through the routuine the first one does.H >They come to a $ prompt rather quickly, but I can't do much from there.B >I've tried several commands and they all fail.  Any idea what I'm >looking at here?m  N Standalone backup? If so, its purpose is to allow you to make a backup of the J disk(s) to other disks or tape. It would be normal to have OpenVMS on one M drive and standalone backup on all the other drives. This way if you need to eN restore the system disk after a failure you can boot SAB off any of the other L surviving drives and save yourself the time it takes to boot from tape - to B say nothing of the time it takes you to find the bootable tape :-)     Antonio7  I Antonio Carlini                            Mail: carlini@true.lkg.dec.com # DECnet-Plus for OpenVMS Engineeringt6 COMPAQ                                     Reading, UK   ------------------------------  % Date: Thu, 11 May 2000 12:54:38 +0200t= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>y+ Subject: Re: Looking for infor for 4000/200d) Message-ID: <391A916D.1AEC56A4@gtech.com>r   Stephen Henry wrote:F >     First off, I'd like to set up a terminal emulator on my PC.  Any > software you'd recommend?n  + If money is no problem: Reflection from WRQn# If you have little money: Kermit-95e- If you want something free: MS-Kermit for DOSa  + Avoid that crap that comes with Windows 9x.t  H >     And, the biggest question of all.....The system came with 5 drivesG > and an r400x cabinet to hold them.  One drive boots openVMS 6.1, goesWI > through a bunch of stuff and comes to the login prompt.  Unfortunatly I J > don't have an account, but I don't want to wipe the drives either, since$ > I don't have openVMS to reinstall.  B See the VMS FAQ for info about what to do when the SYSTEM password is lost.   (there are a link to the FAQ atn* http://www.levitte.org/~ava/vms_faq.htmlx)  D >     Last question:  Does anybody know of any good SMTP servers and  > mailing list software for VMS?   PMDF.o  ! MX (the old version 4.2 is free).h   Arne   ------------------------------  % Date: Thu, 11 May 2000 09:02:10 -0400 # From: Jim Agnew <agnew@hsc.vcu.edu>d+ Subject: Re: Looking for infor for 4000/200u+ Message-ID: <391AAF52.65328141@hsc.vcu.edu>   H for a terminal emuilator, try kermit-95, or if you're only interested inG ONE emulation, even dos kermit would work.... there's also a web search 0 for EWAN, Emulator Without A Name... free also..   Stephen Henry wrote: > I >     I just purchased a couple VAX 4000/200 machines and I'm on the huntoG > for as much information as I can get.  At this point I don't have any I > manuals, though I have been able to identify most of the equipment; I'meE > just not sure how some of it was used when the system was in place.u > F >     First off, I'd like to set up a terminal emulator on my PC.  AnyE > software you'd recommend?  I'm looking for some sort of turorial on-J > hooking up the VAX to the PC too.  I know I'd obviouse go serial, but isC > there something that I have to configure on the VAX to supprt the-I > terminal?   There are a bunch of serial ports hanging out of one of the * > machines, one of which is labeled modem. > I >     I have a total of 10 terminals, but only know how to hook up one tofH > each machine. I think I'm missing a part or something.  (The terminalsA > don't have any DB25 ports on the back).  The system came with 4tJ > DECserver 700s but they all have RJ45 ports on them (16).  Any idea what! > these would have been used for?e > H >     And, the biggest question of all.....The system came with 5 drivesG > and an r400x cabinet to hold them.  One drive boots openVMS 6.1, goes9I > through a bunch of stuff and comes to the login prompt.  Unfortunatly I-J > don't have an account, but I don't want to wipe the drives either, sinceG > I don't have openVMS to reinstall.  Two of the other drives also bootcI > openVMS 6.1, but they don't go through the routuine the first one does.MI > They come to a $ prompt rather quickly, but I can't do much from there.sC > I've tried several commands and they all fail.  Any idea what I'ma > looking at here? > D >     Last question:  Does anybody know of any good SMTP servers and  > mailing list software for VMS? >  > Thanks in advancea >  > Stevet >  > --I > **NEW** Check Out LaughNet After Dark http://www.laughnet.net/afterdarkx> > Lose Weight Laughing At LaughNet!!   http://www.laughnet.net? > Independent Cartoonist's Homepage http://www.laughnet.net/icho   ------------------------------  % Date: Thu, 11 May 2000 08:55:54 -0500-) From: "John E. Malmberg" <wb8tyw@qsl.net>m+ Subject: Re: Looking for infor for 4000/200t7 Message-ID: <194601bfbb50$a1d56d80$020a0a0a@xile.realm>    > Stephen Henry wrote:H > >     First off, I'd like to set up a terminal emulator on my PC.  Any > > software you'd recommend?-  I If you are considering purchasing a terminal emulator, check out the onesJF that are bundled with X-11 emulators along with just the VT Emulators.  I Of these, the last time I checked the prices, Pathworks-32 was the lowesto cost.dE I have also achieved good results with Reflection-X, and Hummingbird.t  E Most of these vendors will send you a timebombed demo so that you cand% determine which will suit your needs.e  G When purchasing a terminal emulator, consider how much time you will be J using it.  Paying a fair value up front can be worth it if you are a heavy user.n   -Johnf wb8ytw@qsl.network.    ------------------------------  % Date: Thu, 11 May 2000 17:35:47 +0200   From: Paul Sture <paul@sture.ch>+ Subject: Re: Looking for infor for 4000/200 + Message-ID: <VA.0000003f.04c52819@sture.ch>b  B In article <shjsi3gl6fc46@corp.supernews.com>, David Turner wrote: > ' > Use hyperterm if you're using windows-3 > It's built in - it's free (well.... questionable)m >  >  > Works fine > J Not if you've ever tried dialling in to a VMS box. My experience the last 7 time I tried it was that it was unusable (W95 version).n ___c
 Paul Sture Switzerlandp   ------------------------------  % Date: Thu, 11 May 2000 12:59:23 +0200f= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>n" Subject: Re: Marketing opportunity) Message-ID: <391A928B.390503D5@gtech.com>h   "Richard D. Piccard" wrote:fJ > Not having access to Navigator on VMS, I can only speculate:  what about< > an HTML-ized E-mail message, where the HTML includes Java?   But it should only run applets.i  7 And in theory an applet can not do anything harmfull top the system.t  9 There has been found implementation bugs in JVM's, but asd I see it then:  8 SUN did think about the problem and developed a complete7 security model to handle the problem. There may be bugs 9 in some implementations, but they get fixed and the basicr architecture is OK.h  * MS choosed to completely ignore securitty.   Arne   ------------------------------  # Date: Thu, 11 May 2000 16:15:02 GMT " From: Art Rice <arice@ue.itug.org>" Subject: Re: Marketing opportunity8 Message-ID: <2umlhso88225jcolblbrqhp58i4osijgce@4ax.com>  / On Wed, 10 May 2000 11:21:28 +0000, Nigel Arnots$ <sysmgr@maxwell.ph.kcl.ac.uk> wrote:  J >> In a previous article, Nigel Arnot <sysmgr@maxwell.ph.kcl.ac.uk> wrote: >>  2 >> ->... Our NT systems are very much multi-user.  >> e >> A minor nit:  >> _J >> While we know what you mean by "multi-user", I believe the correct term! >> should be "roving profiles".  f >dJ >Actually, no. Roaming profiles are very crude, and involve a bulk copy ofF >the user's desktop at every login from server to client, and at everyI >logout vice versa. And if your client crashes you potentially lose a loto	 >of work.- >-C >What we do is basically what's described in Microsoft's oxymoronic F >"Zero Administration Kit"; the desktop is served across the network, C >not copied. The local client is "locked down" so to a fairly greatnD >extent, unprivileged users can't mutate it. I wouldn't go so far asH >to call them secure, though; there are more holes than a swiss cheese.  >n' >> We don't want to advance the popular$G >> misconception that NT is a "multi-user" system. Multi-tasking - yes.p >> Multi-user - no.  >mH >Actually, NT IS multi-user, but is normally restricted to one user at aL >time,  Services can run under different user-ids to the one user logged in M >to the GUI screen. Conceptually, it's not a lot different to an AlphaStationCL >with a one-user VMS license. Microsoft terminal server and Citrix Winframe K >extend that to allow a syatem to perform work for multiple users at once.   >tI >W95/W98 ARE merely multitasking (and not good at that either ... IMHO nocF >use for anything at all, use them and sooner or later you'll come to  >regret it.) > I >Once again, not a recommendation. Merely recognition that NT is less bad  >than it's stable-mates.  & You refer to NT's mates as "stable"?  4 You've got to learn to choose your words better. :>)   >v >	Yours, >		Nigel Arnot. >		NRA@MAXWELL.PH.KCL.AC.UK                    > 8 >		"In the beginning there was nothing, which exploded."   -- ,
 Art Rice   *#:# Special Data Processing Corporationp& --------------------------------------' All opinions expressed are mine and do e% not reflect the views of my employer.(   ------------------------------  # Date: Thu, 11 May 2000 11:43:08 GMT ( From: Terry Kennedy <terry@gate.tmk.com>! Subject: Re: Mozilla M15's out....' Message-ID: <FuE8Jx.Kq8@spcuna.spc.edu>e  ) Colin Blake <colin@theblakes.com> writes:eD > For details, see http://bugzilla.mozilla.org/show_bug.cgi?id=36682  F   Yup. But is there an underlying Mozilla problem, since the fix says:  H | - DE 6216: Reads of 0 bytes incorrectly returned SS$_BADPARAM when theJ |            connection was still active.  This prevented Mozilla V15 from2 |            displaying pages on Multinet systems.  F   I'm fuzzy on why Mozilla was changed to issue zero-byte reads. Is it- looking for some side-effect, or is it a bug?   - 	Terry Kennedy             http://www.tmk.come5         terry@tmk.com             Jersey City, NJ USA?   ------------------------------  # Date: Thu, 11 May 2000 12:09:13 GMTu' From: Colin Blake <colin@theblakes.com> ! Subject: Re: Mozilla M15's out...1- Message-ID: <391AA2E6.653032DB@theblakes.com>.  7 I believe its making sure the connection is still open.j   ------------------------------  % Date: Thu, 11 May 2000 10:13:42 +0200-C From: Eberhard Heuser-Hofmann <vaxinf@chclu.chemie.uni-konstanz.de>" Subject: Re: MPI for OpenVMS?:@ Message-ID: <009E9EBA.0C402FCD.542@CHCLU.CHEMIE.UNI-KONSTANZ.DE>   Hi,   W >If one could stay close to the standards would be the best way to survive or even make ] >it possible, I'm currently working on trying to get Dan Clark's VAX/UCX PVM-port to work on e[ Alpha OpenVMS 7.2 and "Digital TCP/IP Services 5.0", and, I have a access to other parallelM" >systems to confirm functionality.R >I was at the "OpenVMS Conference" here in Sweden with Rich Marcello where I askeda >"after" if he knew if CPQ have had any thoughts to port PVM or MPI to OpenVMS, none there knew, o` >as none even knew what it was, but they promised to return with an answer, since  there's only _ >a few in the world that have asked... so I understand what the answer could be, guess it's up u0 >to us who like to think and work in parallel...  M If you compare this with the options you have on a OSF/DECUNIX/TRU64 machine:i  X PVM, MPI, parallelized BLAS (for a workstation farm!), OpenMP, High Performance Fortran W extension. And all these things supported and distributed by DEQ. The reason is simple:-Z They could not sell one machine, if they were unable to offer these modules. Under OpenVMS_ the is no "demand". What comes out is that noone will buy OpenVMS, at least if you want to run 0W scientific computational applications (You remember that this was the strength of VMS).-   Eberhard      O ===============================================================================    Dr. Eberhard Heuser-Hofmanns Univ. Konstanz Fakultaet fuer Chemiei Universitaets-Strasse 10 D-78464 Konstanz Germany . Phone: +49-7531-88-2026, FAX: +49-7531-88-3139* email: vaxinf@chclu.chemie.uni-konstanz.de  O ===============================================================================t   ------------------------------  % Date: Thu, 11 May 2000 13:23:36 +0100N. From: "Shaun Doolin" <ShaunDoolin@hotmail.com>/ Subject: Re: ODBC Drivers for OpenVMS RMS filesd1 Message-ID: <kGxS4.12595$sB3.7842@news.indigo.ie>   I or you could try www.nadastech.com they have a product that allows direct 0 access to RMS files without ODBC, no middleware.   Shaunl  , <awrycroft0001@my-deja.com> wrote in message# news:8f90me$u2g$1@nnrp1.deja.com...E > Hi,t >t3 > The Visual Basic documentation makes reference to / > the ODBC drivers to access OpenVMS RMS files.e+ > Does anybody know where I can find them ?d >t > Thanks > Andrew Rycroft >o >l( > Sent via Deja.com http://www.deja.com/ > Before you buy.?   ------------------------------  % Date: Thu, 11 May 2000 17:35:43 +0200u  From: Paul Sture <paul@sture.ch> Subject: Re: OPCOM+ Message-ID: <VA.0000003c.04c51998@sture.ch>   F In article <shbcpsa7qil35@corp.supernews.com>, John E. Malmberg wrote:+ > From: "John E. Malmberg" <wb8tyw@qsl.net>b > Newsgroups: comp.os.vmso > Subject: Re: OPCOM& > Date: Sun, 7 May 2000 13:23:25 -0500 > - > Paul Sture <paul@sture.ch> wrote in messageb' > news:VA.00000035.056d40e9@sture.ch...w > I > > But the cluster had this on last week doesn't have graphics consoles.p: > > Unfortunately it's not a cluster I can reboot at will. > E I've since realised that although this cluster doesn't have graphics w+ consoles, it _does_ have the adapters so...t >h$ > I do not think a reboot is needed. > 9 > Also the WINDOW_SYSTEM parameter appears to be dynamic.o > & > After the required logicals are set: > , > $@sys$system:startup opcom !Restart Opcom. >  Thanks, I'll give that a whirl.- ___-
 Paul Sture Switzerland0   ------------------------------  % Date: Thu, 11 May 2000 17:35:44 +0200:  From: Paul Sture <paul@sture.ch> Subject: Re: OPCOM+ Message-ID: <VA.0000003d.04c51d08@sture.ch>9  J In article <8f6jjb$s9t$3@mailint03.im.hou.compaq.com>, Hoff Hoffman wrote:4 > From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) > Newsgroups: comp.os.vmsn > Subject: Re: OPCOM > Date: 8 May 2000 14:39:39 GMTn >  > P > In article <VA.00000032.1a234ddd@sture.ch>, Paul Sture <paul@sture.ch> writes:L > :This was IMHO a bug introduced in VMS V6.2. A comparison of the V6.1 and O > :V6.2 startup files revealed that under V6.1 the Alpha and VAX versions were yP > :different (IIRC one of the files called by startup.com), and in V6.2 someone L > :had (laudably) "corrected" this, with the result that both VAX and Alpha  > :versions became identical.  > K >   I do not recall off-hand if I munged with SYSTARTUP_VMS.COM and am thusX? >   the culprit referenced above, but it would not surprise me.h > + So I've possibly found the culprit! ;-) ;-)b  Q Seriously, I have dim memories that the code in the newer version ran a .EXE, at rP least on the Alpha, and I assumed that the .EXE was missing the logic contained 8 in V6.1 as DCL. It was 4 years ago so, I could be wrong.  K >   I have been roto-tilling (in my copious spare time) the various startupaK >   and login-related DCL procedures used on OpenVMS VAX and OpenVMS Alpha, J >   with an eye toward neater DCL and toward platform-neutral common code.L >   ost recently, I've had my sticky fingers in SYLOGICALS.COM.  Additional J >   targets include the SYLOGIN.COM and LOGIN.COM procedures -- one of my K >   changes did recently clobber some of the new OpenVMS Alpha work in the l; >   early baselevels of the next big OpenVMS release. :-(  o > M >   If I get into SYLOGICALS.COM again, I will add some comments to point to  F >   the DECwindows console tool, and better describe OPCOM operations. >  Yep, that could be useful.P >  --------------------------- pure personal opinion ---------------------------N >    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com >r   ___l
 Paul Sture Switzerlandi   ------------------------------  % Date: Thu, 11 May 2000 11:40:17 +0200 - From: "Menno van Eck" <m.van.eck@philips.com> * Subject: Re: Pitch of the Margin bell Beep8 Message-ID: <958059881.56684@mmisg1.best.ms.philips.com>  K I have tried the solution on the CDE, all works fine, but I'm sorry to telleI you, we are still using the old style motif desktop. Do you know an equalmK solution for this one. Changing the volume of the keyklick, even the volume 1 of the warning bell does not do any thing for me.o   Greetings, Menno van Eck        > Fred Kleinsorge <kleinsorge@star.zko.dec.com> wrote in message% news:8f94nl$eej$1@lead.zk3.dec.com... K > At some point in time, the fixes to allow pitch/duration to work with theoK > "normal" controls made it out... this may be what you are seeing.  In anyt	 > case...  >dL > Go into the STYLE manager, and select BEEP.  If the volume control changesL > the pitch... then you have the old stuff (in which case the "click" volumeI > under keyboard will change the duration -- there is no key click for PC-I > keyboards).  Otherwise, the volume will only turn it on/off (the system- bell; > has no volume control), and pitch and duration will work.i >r > _Fred/ >d' > Robert Deininger wrote in message ...mF > >On Thu, May 4, 2000 11:20 AM, Menno van Eck <m.van.eck@philips.com> wrote:C > >>Installing VMS 7.1-2 on an AS800/500 I noticed the pitch of theoH > >>margin/warning bell for Motif X-applications on the local system was lowereF > >>than I used to. I know the pitch is influenced by which version of > >>SYS$IKBDRIVER.EXE is used. > >>K > >>Does anyone have a clue how to change (higher) the pitch or maybe lowert > >theL > >>volume of this bell. I get remarks of the users of the systems about it? > > I > >Same thing happened to me last night when I installed 7.1-2.  I didn't  find, > >(or really even look for) a solution yet. > >t > >  > >--------------------------- > >Robert DeiningerS > >rdeininger@mindspring.com > >a > >l > >o >e >Q   ------------------------------  % Date: Thu, 11 May 2000 12:46:18 -0400o5 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com> * Subject: Re: Pitch of the Margin bell Beep+ Message-ID: <8feo3g$msf$1@lead.zk3.dec.com>-  F Hmmm.  I think you may be in a catch-22.  You have the latest drivers,H and server fixes, and CDE now works fine.  But the old volume-only MotifJ "hack" does not.  The only solution that I can think of offhand is to use:   $ @decw$utils:decw$define_utilst $ xset b 1 100 100  K Where the first number is the volume, the second is the pitch and the thirdl/ the duration.  I forget what units they are in.r   _Fred    Menno van Eck wrote in message/ <958059881.56684@mmisg1.best.ms.philips.com>...oL >I have tried the solution on the CDE, all works fine, but I'm sorry to tellJ >you, we are still using the old style motif desktop. Do you know an equalL >solution for this one. Changing the volume of the keyklick, even the volume2 >of the warning bell does not do any thing for me. >  >Greetings, Menno van Eckm >d >I >l >i? >Fred Kleinsorge <kleinsorge@star.zko.dec.com> wrote in messaget& >news:8f94nl$eej$1@lead.zk3.dec.com...L >> At some point in time, the fixes to allow pitch/duration to work with theL >> "normal" controls made it out... this may be what you are seeing.  In any
 >> case... >>E >> Go into the STYLE manager, and select BEEP.  If the volume controly changes F >> the pitch... then you have the old stuff (in which case the "click" volumeJ >> under keyboard will change the duration -- there is no key click for PCJ >> keyboards).  Otherwise, the volume will only turn it on/off (the system >bell < >> has no volume control), and pitch and duration will work. >> >> _Fred >>( >> Robert Deininger wrote in message ...G >> >On Thu, May 4, 2000 11:20 AM, Menno van Eck <m.van.eck@philips.com>  >wrote:fD >> >>Installing VMS 7.1-2 on an AS800/500 I noticed the pitch of theI >> >>margin/warning bell for Motif X-applications on the local system wase >lowerG >> >>than I used to. I know the pitch is influenced by which version ofe >> >>SYS$IKBDRIVER.EXE is used.h >> >> L >> >>Does anyone have a clue how to change (higher) the pitch or maybe lower >> >the L >> >>volume of this bell. I get remarks of the users of the systems about it ?t >> >J >> >Same thing happened to me last night when I installed 7.1-2.  I didn't >findn- >> >(or really even look for) a solution yet.c >> > >> > >> >---------------------------u >> >Robert Deininger >> >rdeininger@mindspring.comr >> > >> > >> > >> >> >  >g   ------------------------------  % Date: Thu, 11 May 2000 13:07:40 +0100/. From: "Shaun Doolin" <ShaunDoolin@hotmail.com>: Subject: problem linking to ucx$ipc.olb on AXP/VMX71-UCX401 Message-ID: <prxS4.12592$sB3.7663@news.indigo.ie>o  E I have a program developed under VAX C on VAX/VMS 5.5-2 UCX 3.3 using@
 DECThreadsK that I am trying to move to an AXP/VMS 7.1 ucx 4.0. I can compile using thee. /STANDARD=VAXC option and it compiles cleanly.  I But when I try to link using the sys$library:ucx$ipc/lib option it cannote* see the thread functions and errors occur. I removed the library option         sys$library:ucx$ipc/lib  and included the share options; sys$share:ucx$ipc_shr/share, sys$share:cma$lib_shr/share, -i  sys$share:cma$open_lib_shr/share  1 linking then gave the following undefined symbolsd+ %LINK-I-UDFSYM,         PTHREAD_ATTR_CREATEi, %LINK-I-UDFSYM,         PTHREAD_ATTR_SETPRIO- %LINK-I-UDFSYM,         PTHREAD_ATTR_SETSCHEDsB %LINK-W-USEUNDEF, undefined symbol PTHREAD_ATTR_SETPRIO referenced)         in psect $LINK$ offset %X00000890 0         in module RWA file USER$A:[AGH]RWA.OBJ;2C %LINK-W-USEUNDEF, undefined symbol PTHREAD_ATTR_SETSCHED referenced:)         in psect $LINK$ offset %X000009D0g0         in module RWA file USER$A:[AGH]RWA.OBJ;2A %LINK-W-USEUNDEF, undefined symbol PTHREAD_ATTR_CREATE referenceds)         in psect $LINK$ offset %X000009E0s0         in module RWA file USER$A:[AGH]RWA.OBJ;2  C Q: is there a library file comparable to ucx$ipc.olb on AXP/VMS 7.1oF         or are all the decthread functions listed in the /share images         listed above?lA Q: have these thread functions been replaced or renamed and if so,8         does anyone know their equivilent function names   Thx,
         Shaung   ------------------------------  % Date: Thu, 11 May 2000 14:53:57 +01003* From: "Richard Brodie" <R.Brodie@rl.ac.uk>> Subject: Re: problem linking to ucx$ipc.olb on AXP/VMX71-UCX40, Message-ID: <8fee1n$1iho@newton.cc.rl.ac.uk>  9 "Shaun Doolin" <ShaunDoolin@hotmail.com> wrote in messagee+ news:prxS4.12592$sB3.7663@news.indigo.ie...yG > I have a program developed under VAX C on VAX/VMS 5.5-2 UCX 3.3 usingu@ > DECThreads that I am trying to move to an AXP/VMS 7.1 ucx 4.0.H > I can compile using the /STANDARD=VAXC option and it compiles cleanly. >aK > But when I try to link using the sys$library:ucx$ipc/lib option it cannott, > see the thread functions and errors occur.  E The thread implementation on VMS 7.1 complies with POSIX 1003.1c-1995 E rather than good old 1003.4a draft 4 (which is obsolete). You can use:? /define=_PTHREAD_USE_D4 for now but in the long run grab a copys< of the DECthreads manual and port your code to the standard.  D You obviously didn't read the small print in the DECthreads manual -B "don't blame us if the IEEE change the sstandard in a non-portable way before it gets issued".t  A The DECthreads manual is online/in the fine manual set. There aren migration details included.h   ------------------------------  % Date: Thu, 11 May 2000 10:16:18 -0500i) From: "John E. Malmberg" <wb8tyw@qsl.net>u> Subject: Re: problem linking to ucx$ipc.olb on AXP/VMX71-UCX40/ Message-ID: <shljasaskcp108@corp.supernews.com>u  0 Shaun Doolin <ShaunDoolin@hotmail.company> wroteG > I have a program developed under VAX C on VAX/VMS 5.5-2 UCX 3.3 usingo > DECThreadsI > that I am trying to move to an AXP/VMS 7.1 ucx 4.0. I can compile using. thel0 > /STANDARD=VAXC option and it compiles cleanly. >eK > But when I try to link using the sys$library:ucx$ipc/lib option it cannott, > see the thread functions and errors occur. > I removed the library option! >         sys$library:ucx$ipc/libm  > and included the share options= > sys$share:ucx$ipc_shr/share, sys$share:cma$lib_shr/share, -A" > sys$share:cma$open_lib_shr/share  H Try removing your option file.  It was only needed for VAX C.  The DEC C? compilation environment is set up to find these images already.u  K This is documented in the DEC C compiler manual.  It is one of the numerouse improvements over VAX C.  5 Consider not using the /STANDARD=VAXC and instead use L /WARN=ENABLE=LEVEL4/PREFIX=ALL.  You may discover some previously undetected bugs.   E IMHO: Fall back to /STANDARD=VAXC as a last resort to get a module toa compile.   -Johny wb8tyw@qsl.network   ------------------------------  % Date: Thu, 11 May 2000 13:10:42 -0400 " From: Dan Sugalski <dan@sidhe.org>> Subject: Re: problem linking to ucx$ipc.olb on AXP/VMX71-UCX408 Message-ID: <4.3.1.0.20000511125422.00eac4f0@24.8.96.48>  . At 01:07 PM 5/11/00 +0100, Shaun Doolin wrote:F >I have a program developed under VAX C on VAX/VMS 5.5-2 UCX 3.3 using >DECThreadseL >that I am trying to move to an AXP/VMS 7.1 ucx 4.0. I can compile using the/ >/STANDARD=VAXC option and it compiles cleanly.   A No, it didn't compile cleanly. /STAND=VAXC masks a whole host of -1 programming problems. It's the equivalent to the eL /CROSS_FINGERS_AND_HOPE_FOR_THE_BEST switch. Take it off and fix the errors > and you'll find your problems go away. (Been there, done that)  G More specifically, threads changed a bunch between 5.5-2 and 7.1. They ?H moved from the d4 draft of the POSIX thread implementation to the final L draft. For reasons sane people can't discern, some functions retained their J name but changed the number or type of the parameters just to make things I more fun. (Which means code that compiles with /STAND=VAXC may well kick nI over dead when run, though taking that switch out will make the compiler s' whine properly about the code problems)n   					Dan  L ----------------------------------------------------------------------------L Dan Sugalski                          General and VMS-specific perl training
 dan@sidhe.org >                                       Mail me for more details   ------------------------------   Date: 11 May 2000 13:07:22 GMTF From: lederman@star.enet.dec.DISABLE-JUNK-EMAIL.com (Bart Z. Lederman)= Subject: Re: Problem with Seagate disk on VAXstation 4000 VLCo, Message-ID: <8febaa$n0o$1@sniff.shr.dec.com>  [ In article <3919EFBA.B1BE5090@bigpond.com>, David B Sneddon <dbsneddon@bigpond.com> writes:> > J >It is not a duplicate ID, I install the disk internally and terminate the, >SCSI - this is the only device on the SCSI. >> >k >show config gives men >	A/1/0V >	A/1/1M >	 ... >	A/1/7e >like it sees 8 devices.  ? This is indeed one of the common symptoms of having two devicesM with the same ID.r  C >The disk was removed from a functioning VAXstation 4000-90 and wash >fine in that box.  A Some VAX systems have the controller at ID 7, while some have theM@ ID at 6, and the VLC is one of them.  If your disk is at ID 6 it? could have worked in the 4000-90 but still give you problems ino? the VLC.  Try changing it's ID to 0 or 1 and see if that helps. > (Also check to see that none of the little jumpers that select; the ID didn't drop off or come lose when you moved the disks from one system to another.)   -- i(  B. Z. Lederman   Personal Opinions Only  8  Posting to a News group does NOT give anyone permission8  to send me advertising by E-mail or put me on a mailing  list of any kind.  5  Please remove the "DISABLE-JUNK-EMAIL" if you have ab5  legitimate reason to E-mail a response to this post.m   ------------------------------  % Date: Thu, 11 May 2000 11:33:48 +0200o& From: Ingo Soetebier <ingo@igd.fhg.de>$ Subject: Question on VAXstation 2000* Message-ID: <391A7E7C.34FDC6A3@igd.fhg.de>   Hallo,  = I got a VAXstation 2000 with a TK50 for free and I'm not very E experienced with VAX computers. Switching on the machine the selftest ' reports (the streamer is not attached):z   KA410-B V2.1  < F...E...D...C...B...A...9...8...7...6...5...4...3_..2_..1?..    ? E 0040 0000.0005   ? 7 0090 7730.7731=  ? 6 00A0 0000.4001= ?? 1 00C0 0000.7004=    ( When I try to boot the computer it says:   -DUA0a   -ESA01   ?Retry  B I think first it tries to boot from the harddisk and then over theH network. It seems there's no OS on the harddisk. How do I install VMS onH a VAXstation 2000? Should I first format the disk with 'test 70'? I have the following tapes:  = AQ-NB26A-BE  VMS V5.1-B BIN TK50 BINARY               ME12522 > AQ-LX08D-BE  VMS V5.1-1 BIN TK50 MAINTENANCE UPDATE   ME124208< AQ-LX08C-BE  VMS V5.1   BIN TK50 MAJOR UPDATE         ME5153  : Can I use them for installing? How do I boot from a tape?    regards, Ingo   ------------------------------  + Date: Thu, 11 May 2000 16:10:43 +0000 (   ) 3 From: Christopher Smith <chriss@Mufasa.pubserv.com>u( Subject: Re: Question on VAXstation 2000I Message-ID: <Pine.LNX.4.05.10005111556400.4375-100000@Mufasa.pubserv.com>   * On Thu, 11 May 2000, Ingo Soetebier wrote:  ? > I got a VAXstation 2000 with a TK50 for free and I'm not very,G > experienced with VAX computers. Switching on the machine the selftest ) > reports (the streamer is not attached):C  ? Congradulations.  The VAXStation 2000 is a nice little machine.s   > KA410-B V2.1 > > > F...E...D...C...B...A...9...8...7...6...5...4...3_..2_..1?.. >  >  ? E 0040 0000.0005e  G The clock isn't right ;) ... depending on how long the machine has beenc powered down, this is normal.n   >  ? 7 0090 7730.7731    A disk controller...(I think)s   >  ? 6 00A0 0000.4001a  # Another disk controller...(I think)t   > ?? 1 00C0 0000.7004    Ethernet isn't terminated.  E If you have a lan, you can hook it in, otherwise just put a T and twoe terminators on it.    * > When I try to boot the computer it says: > -DUA0h > -ESA0l > ?RetryD > I think first it tries to boot from the harddisk and then over theJ > network. It seems there's no OS on the harddisk. How do I install VMS onJ > a VAXstation 2000? Should I first format the disk with 'test 70'? I have > the following tapes:    C You're correct.  DUA0 is the first disk.  ESA0 is the NIC.  MUA0 isH the first tape drive.u  F First you should check to make sure you even have a disk (some of themI didn't....)  Assuming that you have a disk, try test 71.  If your disk iscH actually formatted, this will verify it... (Hint: if it takes an hour or& more, don't bother... just reformat ;)  E If the disk verifys, I'd try booting one of those tapes.  This can be7D accomplished by plugging the tape drive in, powering on, and saying:  	 >>>b mua0C  G If it boots you'll get into "Standalone backup," and be able to restore. VMS to the hard drive.  . To do that, you'll want to say something like:  & backup/init/verify vms051.b/save dua0:  F If you didn't have to re-init the disk, you can try /noinit instead ofF /init, and it might same some information. (That could be good or bad,) depending on what it saves, obviously...).    ? > AQ-NB26A-BE  VMS V5.1-B BIN TK50 BINARY               ME12522 @ > AQ-LX08D-BE  VMS V5.1-1 BIN TK50 MAINTENANCE UPDATE   ME124208> > AQ-LX08C-BE  VMS V5.1   BIN TK50 MAJOR UPDATE         ME5153  < > Can I use them for installing? How do I boot from a tape?   1 I don't know, but i suggest that you find out. :)    Regards,   Chris"  O =============================================================================== @ "My two cents"			(http://rootworks.com/twocentsworth.cgi?128562)= Christopher Smith(chriss@pubserv.com)			Prgramer^W Programmerq Prime Synergy of Champaign, IL.n% ------------------------------------- I "Where a calculator on the ENIAC is equipped with 18,000 vacuum tubes andiH weighs 30 tons, computers in the future may have only 1,000 vacuum tubes; and weigh only 1.5 tons." -- Popular Mechanics, March 1949  O -------------------------------------------------------------------------------o   ------------------------------  % Date: Thu, 11 May 2000 17:35:46 +0200a  From: Paul Sture <paul@sture.ch>! Subject: Re: tape label in EBCDICr+ Message-ID: <VA.0000003e.04c52581@sture.ch>   > In article <rbbEVIySTgfV@eisner.decus.org>, Bob Koehler wrote:  N > In article <8f9rtg$mh$1@news.IAEhv.nl>, "Hans Vlems" <hvlems@iae.nl> writes:@ > > If I remember correctly, ANSI tape labels are always EBCDIC. > I > Certainly not!  ANSI tape labels are in ASCII, the layout happens to begD > almost exactly the same as IBM SL lables, but those are in EBCDIC. > N Agreed. I found it useful in the past to ask the mainframe folks for a sample L tape that they _could_ read, because on more than one occasion I found that 2 they didn't know what their own tapes looked like. ___N
 Paul Sture Switzerlandt   ------------------------------  % Date: Thu, 11 May 2000 13:05:44 +0100-. From: "Shaun Doolin" <ShaunDoolin@hotmail.com>
 Subject: testr1 Message-ID: <ApxS4.12591$sB3.7874@news.indigo.ie>l   ------------------------------  # Date: Thu, 11 May 2000 13:58:15 GMT1/ From: "John Nixon" <jorlnixon@worldnet.att.net>0& Subject: Thanks Hoff (re SDL),  but...G Message-ID: <X%yS4.59511$WF.3320525@bgtnsc04-news.ops.worldnet.att.net>e  L Why do we need to rely on SDL from the FREEWARE disk to make a product which we bought (CMS) work
 correctly?  L I understand that the freeware disk will provide functionality and features,& or improvements to features, which areG not available with pure VMS, but the freeware disk comes with all thosef$ caveats about not being supported by' Compaq and use at your own risk, etc...t  H Isn't there a way to get the SDL functionality from the VMS that we paid' for?  Something that is guaranteed?  IfXJ I wanted a "roll your own" Operating System, I would get Linux,  or one of the other IX variants.  D Can you tell me where the SDL functionality is documented in the VMS documentation?  G But I do thank you for pointing me to the freeware disk for now.  I wask/ completely stuck trying to help the programmer.    ------------------------------   Date: 11 May 2000 16:17:41 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)* Subject: Re: Thanks Hoff (re SDL),  but...6 Message-ID: <8femf5$pbv$1@mailint03.im.hou.compaq.com>  y In article <X%yS4.59511$WF.3320525@bgtnsc04-news.ops.worldnet.att.net>, "John Nixon" <jorlnixon@worldnet.att.net> writes:-M :Why do we need to rely on SDL from the FREEWARE disk to make a product whichE  :we bought (CMS) work correctly?  .   Good question.  I don't have a good answer.   I :Isn't there a way to get the SDL functionality from the VMS that we paid9 :for?   D   I would encourage a formal request to the Customer Support Center.  =   I have informally passed the request along to the CMS team.t  H   Some background: I asked the CMS team to provide the SDL file a while G   back as I was not fond of the (official) CMS-supported and documentediG   technique for resolving the CMS symbols, as this involves explicitly  F   linking against the CMSSHR image to pull in the symbols.  I greatly J   prefer using the more typical include files for the symbol definitions, K   and these are not available with CMS -- which is why I asked for the SDL.u  E :Can you tell me where the SDL functionality is documented in the VMS  :documentation?   C   It is not.  SDL is documented only on the OpenVMS Freeware.  (ThecC   SDL/NOPARSE tool is latent on the OpenVMS kit, but is itself alsor   not documented.)  C   Some background: OpenVMS Engineering is dependent on SDL for mostLC   of the definitions needed to build OpenVMS itself, as well as the 4   creation of the customer-visible definition files.  H :But I do thank you for pointing me to the freeware disk for now.  I was0 :completely stuck trying to help the programmer.  4   That CMS "requires" SDL was also a surprise to me.  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  # Date: Thu, 11 May 2000 13:42:03 GMTe4 From: "Michael D. Ober" <mdo.@.wakeassoc.com.nospam>& Subject: Re: the latest billybox virusE Message-ID: <LMyS4.50795$g4.1406377@newsread2.prod.itd.earthlink.net>.   Marty,  L You have missed a major point in capitalism - Give the buyer what they want,H not what they need.  These two are frequently, but not always, the same.H VMS gives the user what they need, not what they want.  If DEC had givenJ users what they wanted, DEC would probably still be a company instead of aL division of Compaq.  MS has definitely given the users what they wanted, notD necessarily the best software or even the cheapest in all cases, butJ definitely what they wanted.  It simply boils down to marketing and MS canH definitely market while DEC and now Compaq don't even know what the word2 means when it comes to VMS or the Alpha processor.  
 Mike Ober.    7 "Marty Kuhrt" <kuhrt@eisner.decus.org> wrote in message-% news:XLZ8ISDwIPS4@eisner.decus.org... C > I, for one, am definitely tired of your useless posts denegratingbC > VMS in a _VMS_ newsgroup.  You want to pound your chest for M$ don( > it elsewhere.  Welcome to my killfile. >pL > In article <8evm6i$ei5$1@pyrite.mv.net>, "Bill Todd" <billtodd@foo.mv.com> writes: E > > Microsoft pays attention to what most users want:  that's why itss systemsn9 > > are popular, and *one* of the reasons why VMS is not.n >aC > As for the popularity of M$, it is for one reason, and one reasonmC > only.  Their monopoly.  They used strong arm tactics to make sure.C > it was the only viable OS for the PC market.  Had the competition'C > been fair we might have seen products that make the current spateoD > of M$ crud more laughable than it already is.  If they truly caredD > for what the users wanted they'd write bug free products (assumingD > they have that capability) or at least allow for true competition. >h >    ------------------------------  # Date: Thu, 11 May 2000 16:16:40 GMTn" From: Art Rice <arice@ue.itug.org>& Subject: Re: the latest billybox virus8 Message-ID: <03nlhssdgrc082cs7nv1m7ah87sb830ve2@4ax.com>  A On 10 May 2000 21:50:21 GMT, helbig@astro.rug.nl (Phillip Helbig). wrote:  C >In article <XLZ8ISDwIPS4@eisner.decus.org>, kuhrt@eisner.decus.orgi >(Marty Kuhrt) writes: h >oC >>I, for one, am definitely tired of your useless posts denegrating D >>VMS in a _VMS_ newsgroup.  You want to pound your chest for M$ do ( >>it elsewhere.  Welcome to my killfile. >>T >>In article <8evm6i$ei5$1@pyrite.mv.net>, "Bill Todd" <billtodd@foo.mv.com> writes:M >>> Microsoft pays attention to what most users want:  that's why its systems ; >>> are popular, and *one* of the reasons why VMS is not.  d >>D >>As for the popularity of M$, it is for one reason, and one reason D >>only.  Their monopoly.  They used strong arm tactics to make sure C >>it was the only viable OS for the PC market.  Had the competition D >>been fair we might have seen products that make the current spate D >>of M$ crud more laughable than it already is.  If they truly caredE >>for what the users wanted they'd write bug free products (assuming ID >>they have that capability) or at least allow for true competition. >e > F >"Windows NT 5.0 is an evolutionary, not revolutionary, release of theH >Windows NT operating system.  While there are important new features inI >this release, version 5.0 will build on a proven system architecture and > >incorporate tens of thousands of bug fixes from version 4.0."C >            ^^^^^^^^^^^^^^^^^                                     oJ >                                             ---Bill Gates (my emphasis)   @ Question is:  How many thousands of new bugs will be introduced?   >s > N >I had a reference to this at ---http://www.microsoft.com/ntserver/community/  >but I now get l >aP >                                                       Directory Listing Denied >M4 >                           Directory Listing Denied >E@ >   This Virtual Directory does not allow contents to be listed.   -- a
 Art Rice   *#u# Special Data Processing Corporationr& --------------------------------------' All opinions expressed are mine and do n% not reflect the views of my employer.c   ------------------------------  # Date: Thu, 11 May 2000 14:38:52 GMT-) From: "J.G. Peters" <jpeters@sctcorp.com>  Subject: Timeout Issues Take IIk0 Message-ID: <01bfbb56$acaa80a0$32341895@jpeters>  3 As yet, I have been unable to solve this "problem."h. I've tried spawning a subprocess and doing all- manner of nulls, spaces, beeps, DCL commands ,/ - all with the same result - the parent processD. remains unaffected and times out. I have *not*/ tried RUNning a process that executes an image. ) Can anyone send along something like Dave - suggests below? I looked through the Freewarer2 disc. There are clock displays but they all looked- to be written for use with DecWindows, not an + emulation package like TNVTPlus. Did I miss 
 something?  1 I must be getting senile, or something, because I.2 can't seem to figure this out. Time is passing and6 I find myself behind the proverbial "eight ball" here.2 Any other suggestions? Any and all help is greatly appreciated. Thanks.   Joei   "Joe Peters" wrote:uC >Our company is planning to implement a VPN (SecuRemote/SecurID viakG >Earthlink) setup for remote logins. In testing, telnet sessions (OnNett HostC >Suite 5.0 TNVTPlus and MS Telnet) to the AlphaServer 2100 (OpenVMSkH >7.1/Multinet 4.2) are timing out after 15 minutes of inactivity. The 15G >minute setting is set at the firewall. The "network folks" do not want> thisK >setting changed for security reasons. They have been unable to resolve thegD >issue and have now, uh, suggested that the sys admins. come up withI >something to keep the telnet sessions alive. (Our users frequently loginKG >under more than one account, minimizing/restoring sessions to/from thegG >taskbar as needed.) Setting up a simple loop that would require manualmG >execution seems ludicrous. Does anyone have an idea/method for keepingtI >alive a telnet session? There are many "idle process killers" out there.nE >How about a program/job that does just the opposite? By the way, the D >RS/6000 (AIX 4.3.2) admin. has come up with having each login run aJ >background shell script that loops every ten minutes, sending a space andF >thereby keeping the process alive. It works, but certainly seems, uh,H >"Mickey Mouse" (no offense) to me. Any/all help is greatly appreciated. >m >Thanks, Joe  ( "Earl D. Lakia" <lakia@ipact.com> wrote:< > Could you spawn a keep alive sub process in the backgroundK > that writes a Bell every five minutes or a Null?  The subprocess would don> > a breakthrough write to sys$output with no carriage control.; > You could test this with the following (note, the Bell is > > a special insert using the editor).  However, the problem is: > that a <CR><LF> is also output with the bell as well and: > the output screen will advance one line.  Therefore, you6 > might want to create a real program that doesn't add! > these extra control characters.s >  u > $type earl.com > $on controlY then goto 100 > $bell = "<07>" > $100:o > $write sys$output bell > $ wait 0 00:10:00h > $goto 100  >  a: > When user logs in and you determine that he has a telnetG > session (I would do  devnam = f$extract(1,3,f$getdvi("tt:","devnam"))o? > and make that determination, I get NTA as a device when I useA+ > TELNET), spawn the following sub-process.h >  t > $spawn/nowait @earl.como   "Dave Greenwood" wrote:.C >If you try this, be sure to add /input=nla0: to the spawn command.oB >Otherwise a ^C in the top level process will kill the subprocess. > B >Personally, I use a Fortran program that puts up a display on theB >screen and updates the time every minute.  A password is required> >to exit the program (or it'll exit on its own after 4 hours).   >Davei   ------------------------------  % Date: Thu, 11 May 2000 13:18:02 -0400t# From: John Vottero <John@MVPSI.com>R# Subject: RE: Timeout Issues Take II D Message-ID: <C15945A9D9EFCF11BA8B08002BBF1CCC0CD6E8@berry.mvpsi.com>  I You have to find out what the firewall is looking for to reset its timer.fL If you are sending stuff (spaces, beeps etc.) to the terminal and it's stillJ timing out then it appears that the firewall is looking for input FROM theJ terminal to reset its timer.  If that is the case, then you have to see if6 you can write a script for TNVTPlus (rather than VMS).   > -----Original Message-----0 > From: J.G. Peters [mailto:jpeters@sctcorp.com]' > Sent: Thursday, May 11, 2000 10:39 AMm > To: Info-VAX@Mvb.Saic.Com ! > Subject: Timeout Issues Take IIe >  > 5 > As yet, I have been unable to solve this "problem."-0 > I've tried spawning a subprocess and doing all/ > manner of nulls, spaces, beeps, DCL commands  1 > - all with the same result - the parent processh0 > remains unaffected and times out. I have *not*1 > tried RUNning a process that executes an image.i+ > Can anyone send along something like Dave-/ > suggests below? I looked through the Freewarea4 > disc. There are clock displays but they all looked/ > to be written for use with DecWindows, not ane- > emulation package like TNVTPlus. Did I miss  > something? > 3 > I must be getting senile, or something, because It4 > can't seem to figure this out. Time is passing and8 > I find myself behind the proverbial "eight ball" here.4 > Any other suggestions? Any and all help is greatly > appreciated. Thanks. >  > Joey >  > "Joe Peters" wrote:sE > >Our company is planning to implement a VPN (SecuRemote/SecurID viak: > >Earthlink) setup for remote logins. In testing, telnet  > sessions (OnNet_ > HostE > >Suite 5.0 TNVTPlus and MS Telnet) to the AlphaServer 2100 (OpenVMS 8 > >7.1/Multinet 4.2) are timing out after 15 minutes of  > inactivity. The 15> > >minute setting is set at the firewall. The "network folks" 
 > do not wantW > this? > >setting changed for security reasons. They have been unable A > to resolve theF > >issue and have now, uh, suggested that the sys admins. come up with; > >something to keep the telnet sessions alive. (Our users b > frequently login> > >under more than one account, minimizing/restoring sessions 
 > to/from theW; > >taskbar as needed.) Setting up a simple loop that would m > require manual> > >execution seems ludicrous. Does anyone have an idea/method 
 > for keeping/8 > >alive a telnet session? There are many "idle process  > killers" out there.dG > >How about a program/job that does just the opposite? By the way, thenF > >RS/6000 (AIX 4.3.2) admin. has come up with having each login run a9 > >background shell script that loops every ten minutes,   > sending a space andiH > >thereby keeping the process alive. It works, but certainly seems, uh,> > >"Mickey Mouse" (no offense) to me. Any/all help is greatly  > appreciated. > >e > >Thanks, Joe > * > "Earl D. Lakia" <lakia@ipact.com> wrote:> > > Could you spawn a keep alive sub process in the background: > > that writes a Bell every five minutes or a Null?  The  > subprocess would dov@ > > a breakthrough write to sys$output with no carriage control.= > > You could test this with the following (note, the Bell ish@ > > a special insert using the editor).  However, the problem is< > > that a <CR><LF> is also output with the bell as well and< > > the output screen will advance one line.  Therefore, you8 > > might want to create a real program that doesn't add# > > these extra control characters.2 > >    > > $type earl.com > > $on controlY then goto 100 > > $bell = "<07>"	 > > $100:r > > $write sys$output bell > > $ wait 0 00:10:00 
 > > $goto 100g > >  t< > > When user logs in and you determine that he has a telnet" > > session (I would do  devnam = ) > f$extract(1,3,f$getdvi("tt:","devnam"))	A > > and make that determination, I get NTA as a device when I use(- > > TELNET), spawn the following sub-process.r > >  e > > $spawn/nowait @earl.com- >  > "Dave Greenwood" wrote: E > >If you try this, be sure to add /input=nla0: to the spawn command. D > >Otherwise a ^C in the top level process will kill the subprocess. > >bD > >Personally, I use a Fortran program that puts up a display on theD > >screen and updates the time every minute.  A password is required@ > >to exit the program (or it'll exit on its own after 4 hours). >  > >Dave0 >    ------------------------------  % Date: Thu, 11 May 2000 07:19:09 -0500g( From: yyyc186.illegaltospam@flashcom.net Subject: Re: Which VAX to buy?9 Message-ID: <391aa5a6$2$lllp186$mr2ice@news.flashcom.com>@  4 In <3917E220.FCB75242@bornholm-gym.dk>, on 05/11/00 C    at 07:19 AM, Jonas Nielsen <Jonas-Nielsen@bornholm-gym.dk> said:   I I would recommend the 4000-60 if it has any disks.  The 3100 will be slowlI enough to make you go postal when running DEC Windows.  Besides, the SCSId= will be proprietary DEC SCSI of old, not SCSI-II or SCSI-III.a  D I got my 3500 for carrying it away.  Then I spent $5k putting biggerD disks, 15 user DEC BASIC license on it and a fresh open VMS license.   Roland  	 >Hi therel  B >Im a young UNIX-freak, who has become a bit interested in the VMSI >operating system :-). So, i want a VAX and i have been offered some used  >VAXen:D  " >Digital VAXstation 3100/M76 - SPX >Digital VAXstation 4000 - 60-3 >Digital VAXserver 3100 (Scsi disk Ext. SCSI udtag)J  6 >and a Digital Storage Expansion (99 MB Disk + CD-ROM)  D >Basically, i just want to learn OpenVMS, DECwindows a little bit ofH >system administration, and perhaps some programming. I thought it wouldC >be a good idea to buy the Storage Expansion as well, so that i can > >install programs and newer releases of OpenVMS on the system.  I >So i ask you guys, (since i know nothing about VAX), which configurationM- >would you suggest? The VS4000 or VS3100-M76? > >And finally and very important :-), what should i pay for it?  ' >Any help would be greatly appreciated!t   >Kind Regards, >Jonas Nielsen >Denmark   -- ?; -----------------------------------------------------------wD yyyc186@flashcom.net              To Respond delete ".illegaltospam"6                             MR/2 Internet Cruiser 1.528                             For a Microsoft free univers; -----------------------------------------------------------e   ------------------------------  % Date: Thu, 11 May 2000 11:53:45 -0400- From: jlahman@LTVSteel.com' Subject: Wildfire and the future of VMSa8 Message-ID: <852568DC.005765FF.00@notesnta.LTVSteel.com>  L I just received an invitation to attend the Wildfire announcement next week.L However, as I read the articles linked to the invitation, I became concernedP since these articles only talk about how Wildfire will enhance Compaq's positionI in the Unix market.  There was no mention of VMS (even though VMS runs ono
 wildfire).  G So, what is the future of VMS on wildfire?  When I attend this wildfirem: announcement next week, will I only hear about Tru64 Unix?  > Again, I am greatly disappointed in the way Compaq treats VMS.   ------------------------------  % Date: Thu, 11 May 2000 12:13:12 -0400e* From: Clair Grant <grant@evms.zko.dec.com>+ Subject: Re: Wildfire and the future of VMSS0 Message-ID: <391AA3D8.794DBC83@evms.zko.dec.com>   jlahman@LTVSteel.com wrote:v > N > I just received an invitation to attend the Wildfire announcement next week.N > However, as I read the articles linked to the invitation, I became concernedR > since these articles only talk about how Wildfire will enhance Compaq's positionK > in the Unix market.  There was no mention of VMS (even though VMS runs onm > wildfire). > I > So, what is the future of VMS on wildfire?  When I attend this wildfireA< > announcement next week, will I only hear about Tru64 Unix? > @ > Again, I am greatly disappointed in the way Compaq treats VMS.  A I have seen most of the official announcement material and VMS isl prominently mentioned.   -- o Clair Grantr VMS Exec Group Project LeaderB COMPAQ Computer Corporationg 110 Spit Brook Rd. Nashua, NH  03062s   ------------------------------  % Date: Thu, 11 May 2000 13:19:18 -0400  From: jlahman@LTVSteel.com+ Subject: Re: Wildfire and the future of VMSn8 Message-ID: <852568DC.005F3B0E.00@notesnta.LTVSteel.com>  C > I have seen most of the official announcement material and VMS is- > prominently mentioned.    K Thanks for your response.  Am I to assume that VMS will be mentioned at then wildfire announcement?   ------------------------------  % Date: Thu, 11 May 2000 17:35:49 +0200i  From: Paul Sture <paul@sture.ch>+ Subject: Re: Wildfire and the future of VMSi+ Message-ID: <VA.00000042.04c53099@sture.ch>e  C In article <391AA3D8.794DBC83@evms.zko.dec.com>, Clair Grant wrote: , > From: Clair Grant <grant@evms.zko.dec.com> > Newsgroups: comp.os.vmse- > Subject: Re: Wildfire and the future of VMS ' > Date: Thu, 11 May 2000 12:13:12 -0400  >  > jlahman@LTVSteel.com wrote:y > > P > > I just received an invitation to attend the Wildfire announcement next week.P > > However, as I read the articles linked to the invitation, I became concernedT > > since these articles only talk about how Wildfire will enhance Compaq's positionM > > in the Unix market.  There was no mention of VMS (even though VMS runs onu > > wildfire). > > K > > So, what is the future of VMS on wildfire?  When I attend this wildfire > > > announcement next week, will I only hear about Tru64 Unix? > > B > > Again, I am greatly disappointed in the way Compaq treats VMS. > C > I have seen most of the official announcement material and VMS isl > prominently mentioned. > 2 Do you have any comments on the following article Q http://www.theregister.co.uk/000511-000008.html "Compaq prevented from using the t Wildfire name"?t ___p
 Paul Sture Switzerlando   ------------------------------   End of INFO-VAX 2000.263 ************************