1 INFO-VAX	Fri, 12 Apr 2002	Volume 2002 : Issue 201       Contents:  Re: Alphaserver EV7 announcement  Re: Alphaserver EV7 announcement  Re: Alphaserver EV7 announcement  Re: Alphaserver EV7 announcement  Re: Alphaserver EV7 announcement- ANN: CVTLIS updated (converts .LIS to source) 2 Re: Announcing ht://Dig 3.1.6 Release1 for OpenVMS) Re: Best DCL programm you've ever written  Re: Blade architectures  Re: Blade architectures  Re: EWA-bootdriver messages . Re: From comp.sys.dec - deserves exposure here Re: IA64 is not the VAX  Re: Is AMD doing an Intel?P online atricle about sample X11 server architecture and DECwindows architecture?< Re: Samba V2.0.6 Failed to UID error Network map drive error, Re: Time to finally rename VMS back to VMS ? Total cost of ownership # Re: WIreless networking + VMS/Tru64   F ----------------------------------------------------------------------  % Date: Thu, 11 Apr 2002 22:24:13 -0400 ( From: David Froble <davef@tsoft-inc.com>) Subject: Re: Alphaserver EV7 announcement , Message-ID: <3CB6454D.5000908@tsoft-inc.com>   Fred Kleinsorge wrote:    > Bill Todd wrote in message ... >  >>L >>That said, I have no idea how much it would cost to shrink EV6 (though theK >>fact that the core is almost identical to EV7's might allow it to ride on L >>the coattails of the EV7 shrink), and hence whether that would make sense. >> >> > 1 > Remember that EV7 uis built around a EV68 core.   P Wasn't sure which to reply to, so I closed my eyes and stabbed in the dark.  :-)    O Not being a HW guy, I'm not always up on the names of things.  Just what is an  L EV68?  I could see an EV67, if the EV7 was  a smaller size, and the EV6 was % built with the smaller size, but '8'?   L What I think I know.  EV4 was the first released Alpha design, EV5 the next O design (not process shrink), and EV56 was the EV5 design in the process shrink  O used for the EV6, prior to the EV6 coming out, which was a new design.  EV7 is   also a new design.  1 Does EV68 indicate a skipped process shrink size?    Dave   ------------------------------  % Date: Thu, 11 Apr 2002 23:31:24 -0400 2 From: rdeininger@mindspring.com (Robert Deininger)) Subject: Re: Alphaserver EV7 announcement J Message-ID: <rdeininger-1104022331240001@1cust66.tnt2.nashua.nh.da.uu.net>  9 In article <3CB6454D.5000908@tsoft-inc.com>, David Froble  <davef@tsoft-inc.com> wrote:    P >Not being a HW guy, I'm not always up on the names of things.  Just what is an M >EV68?  I could see an EV67, if the EV7 was  a smaller size, and the EV6 was  & >built with the smaller size, but '8'? > M >What I think I know.  EV4 was the first released Alpha design, EV5 the next  P >design (not process shrink), and EV56 was the EV5 design in the process shrink P >used for the EV6, prior to the EV6 coming out, which was a new design.  EV7 is  >also a new design.   B There was also EV45, an EV4 in the EV5 process.  (With a few other tweeks.)  Among other examples.   2 >Does EV68 indicate a skipped process shrink size?  I More or less.  Some kind of significant change in the process, and least.   G There are actually many processor variants in each family.  Most of the G changes are not visible in terms of functionality.  Every little tweek, I fix, or speed change gets a version number, a pass number, or something.  I SDA CLUE CONFIG shows more of a processor's name, but probably not all of I it.  There are internal (visible) registers in the CPU with version info, I and likely internal (invisible to mortals) registers as well.  The EV7 is I chock-full of mystical bits and incantations that are only useful to chip 
 engineers.  F The alpha engineers likely have a secret decoder ring to keep all this? straight, but ordinary folks like us can't get all the secrets.    ------------------------------  # Date: Fri, 12 Apr 2002 04:36:41 GMT * From: "Bill Todd" <billtodd@metrocast.net>) Subject: Re: Alphaserver EV7 announcement ? Message-ID: <tvtt8.2151$043.452732@bin6.nnrp.aus1.giganews.com>   ? "Robert Deininger" <rdeininger@mindspring.com> wrote in message D news:rdeininger-1104022331240001@1cust66.tnt2.nashua.nh.da.uu.net...; > In article <3CB6454D.5000908@tsoft-inc.com>, David Froble  > <davef@tsoft-inc.com> wrote: >  > K > >Not being a HW guy, I'm not always up on the names of things.  Just what  is an J > >EV68?  I could see an EV67, if the EV7 was  a smaller size, and the EV6 was ( > >built with the smaller size, but '8'?  J I'm not sure Robert made it clear below (nor sure that I know exactly whatL the situation is either).  But until someone knowledgeable answers, here's a stab at it:   L The first digit after 'EV' indicates the overall design version of the chip.J The second digit indicates the overall version of the process in which theE chip is fabricated, and really has no relationship to the first digit L (except that both generally increase with time).  Yes, Robert's answer tendsG to contradict that, and he may well be right - this is just my own very  vague understanding.  H So the EV67 (which IIRC did exist) was an EV6 design in a 7th generationI process (either 250 nm or perhaps the half-way shrink to 180 nm that EV68 J completed), while the EV68 was an EV6 design in an 8th generation (180 nm)G process.  The first EV7 to appear will IIRC technically be an EV78 (EV7 K design in that same 8th generation 180 nm process), while the shrink to 130 J nm scheduled for two years from now will, if it occurs, be essentially the, same EV7 design in a 9th generation process.   - bill   > > I > >What I think I know.  EV4 was the first released Alpha design, EV5 the  nextJ > >design (not process shrink), and EV56 was the EV5 design in the process shrinkI > >used for the EV6, prior to the EV6 coming out, which was a new design.  EV7 is > >also a new design.  > D > There was also EV45, an EV4 in the EV5 process.  (With a few other! > tweeks.)  Among other examples.  > 4 > >Does EV68 indicate a skipped process shrink size? > K > More or less.  Some kind of significant change in the process, and least.  > I > There are actually many processor variants in each family.  Most of the I > changes are not visible in terms of functionality.  Every little tweek, J > fix, or speed change gets a version number, a pass number, or something.K > SDA CLUE CONFIG shows more of a processor's name, but probably not all of K > it.  There are internal (visible) registers in the CPU with version info, K > and likely internal (invisible to mortals) registers as well.  The EV7 is K > chock-full of mystical bits and incantations that are only useful to chip  > engineers. > H > The alpha engineers likely have a secret decoder ring to keep all thisA > straight, but ordinary folks like us can't get all the secrets.  >    ------------------------------  # Date: Fri, 12 Apr 2002 05:16:47 GMT * From: "Bill Todd" <billtodd@metrocast.net>) Subject: Re: Alphaserver EV7 announcement ? Message-ID: <35ut8.2912$043.486329@bin6.nnrp.aus1.giganews.com>   : "JF Mezei" <jfmezei.spamnot@videotron.ca> wrote in message& news:3CB6548D.336BE08B@videotron.ca...   ...   K > Now, if the DS10 were upgraded to support EV7, it would be able to use up L > those EV7 chips that are not certified for the highest speed but work fine at > the lower speeds.   I What makes you think they wouldn't be worth much more installed in server * systems that don't need the highest speed?  H A well-controlled process doesn't produce much speed (bin) variation.  IG happened to see prices for AthlonXPs recently:  the 1700+ is the lowest H grade they bother selling, indicating that there aren't enough dice thatI function at all but not at that frequency (whatever it really is, but the F 'quantispeed grade *differences* are linear at 1.5x the real frequencyL differences, so we can use the q-numbers without loss of generality here) toK be worth selling; the 1800+ costs well under 10% more; price goes up around D 20% for each of the next two grades (1900+ and 2000+), and the 2100+L top-of-the-line model costs IIRC a bit over twice what the 1700+ costs (even" though it's less than 25% faster).   > K > The problem with having some Alphas use EV6x and some use EV7x means that  thatJ > you have to have two lines, and the "rejects" from both lines don't have	 much use.   J As noted above, there aren't enough rejects to bother with if your processK is good.  For very large chips like the EV7 it's sometimes possible to make J use of chips that have actual defects (e.g., a cache segment doesn't work)F if you have a mechanism for working around such defects such that theyA aren't fatal, but that's an entirely different issue from that of  speed-grade rejections.    > K > Now, in the post june 25 context, this is probably moot because all EV68s  haveJ > probably already been fabbed, and the DS line of machines probably has aD > finite lifetime already set due to the fixed supply of chips left.  I Only if Compaq wants it that way:  if they want IBM to fab more, I'm sure  IBM will accommodate them.   > K > However, what puzzles me is at the time EV7 was designed, did they make a F > conscious management decision that Alpha was to be relegated to only wildfireK > class machines with no plans to upgrade the DS10/20 stuff (hence allowing L > engineers to put stuff in EV7 that was of no use to DS10 class machines) ?  H DS20s and ES45s are hardly 'wildfire-class' machines, but definitely can benefit from EV7.    > G > Could the Alpha engineers have designed EV7 to be usable in DS10/DS20  class J > machines while keeping all the goodies for the wildfire class machines ?  L EV7 of course *could* be used in a DS10 class machine, but what would be theI market for it if EV6-based DS10s were available too at a noticeably lower  price?  I It sounds as if EV7 *does* have a two-processor configuration planned, so K that presumably covers DS20 and in doing so eliminates any need to have SMP G EV6 boards.  This gives you an inexpensive entry-level Alpha (DS10 with J EV6) - less expensive than it would have been using EV7 - plus the rest ofH the line starting with a dual-processor EV7-based DS20.  While there's aG noticeable gap between those two, it's not clear that, especially given L Alpha's market, it's a significant one:  I'd be happy that the DS10 is being' continued at all (assuming that it is).    - bill   ------------------------------   Date: 12 Apr 2002 00:07 CST ' From: carl@gerg.tamu.edu (Carl Perkins) ) Subject: Re: Alphaserver EV7 announcement - Message-ID: <12APR200200071028@gerg.tamu.edu>   , David Froble <davef@tsoft-inc.com> writes...P }Not being a HW guy, I'm not always up on the names of things.  Just what is an M }EV68?  I could see an EV67, if the EV7 was  a smaller size, and the EV6 was  & }built with the smaller size, but '8'? } M }What I think I know.  EV4 was the first released Alpha design, EV5 the next  P }design (not process shrink), and EV56 was the EV5 design in the process shrink P }used for the EV6, prior to the EV6 coming out, which was a new design.  EV7 is  }also a new design.  } 2 }Does EV68 indicate a skipped process shrink size? }  }Dave    Yes.  F I don't think there will actually be an EV7 as such, I think the firstB version of it is really EV78. The shrunk version will be the EV79.  > The process shrinks have happened faster than the CPU designs.   --- Carl   ------------------------------  # Date: Fri, 12 Apr 2002 05:40:26 GMT - From: goathunter@goatley.com (Hunter Goatley) 6 Subject: ANN: CVTLIS updated (converts .LIS to source)1 Message-ID: <3cb671d3.219360533@news.process.com>   ? CVTLIS V3.5 is now available for download.  CVTLIS converts VMS @ listing files into their original source files.  It supports all< the various languages for which .LIS files appear on the VMS= source-listing CDs.  The program simply strips out all of the ; "listing" output, leaving clean source code, which is often $ easier to read than the .LIS format.  @ This version includes an enhancement from Jon Pinkley to support> B32E files compiled with /SOURCE_LIST=(REQUIRE) (a file on the@ V7.2-2 source-listing CDs was built that way), as well as a new,A optional output file specification as the CVTLIS command's second 
 parameter:  %     $ cvtlis whatever.lis outfile.b32   - You can find CVTLIS using the following URLs:    http://www.process.com/openvms/   6 ftp://ftp.process.com/vms-freeware/fileserv/cvtlis.zip; http://ftp.process.com/ftp/vms-freeware/fileserv/cvtlis.zip 2 ftp://ftp.tmk.com/vms-freeware/fileserv/cvtlis.zip7 http://www.tmk.com/ftp/vms-freeware/fileserv/cvtlis.zip   > As usual, it will appear on the other mirrors within 24 hours.   Hunter ------9 Hunter Goatley, Process Software, http://www.process.com/ 8 goathunter@goatley.com    http://www.goatley.com/hunter/< New Robert R. McCammon site: http://www.RobertRMcCammon.com/   ------------------------------  % Date: Fri, 12 Apr 2002 06:50:57 +0200 2 From: martin@radiogaga.harz.de (Martin Vorlaender); Subject: Re: Announcing ht://Dig 3.1.6 Release1 for OpenVMS ; Message-ID: <3cb667b1.524144494f47414741@radiogaga.harz.de>   * Rob Buxton (rob.buxton@wcc.govt.nz) wrote:D > For those of us playing with some of the VAX based Web Servers any% > chance of a VAX variant of ht//Dig?   H As I don't have a access to a VAX running 7.2 (plus the Compaq C and C++D compilers and MMS/MMK) right now: sorry, no - not from me, at least.0 If anyone wants to give it try, please go ahead.   One general note: @ $PRODUCT EXTRACT RELEASE_NOTES doesn't work (due to a PCSI bug),3 $PRODUCT EXTRACT FILE /SELECT=*.RELEASE_NOTES does.    cu,    Martin --  B                         | Martin Vorlaender | VMS & WNT programmer1  OpenVMS: Where do you  | work: mv@pdv-systeme.de D  want to BE today?      |   http://www.pdv-systeme.de/users/martinv/8                         | home: martin@radiogaga.harz.de   ------------------------------    Date: 11 Apr 2002 21:09:35 -0700& From: chessmaster1010@hotmail.com (JG)2 Subject: Re: Best DCL programm you've ever written< Message-ID: <dd3f0cb7.0204112009.57ee25d@posting.google.com>  Y "Gijs" <Gijs@nospam.com> wrote in message news:<r0%s8.79635$oI.6433580@zwoll1.home.nl>... B > When was it, what does it do for you and whe can it be obtained? >  > Just curious >  > G.  F I use my QUEUE_COMMANDS.COM all the time.  For instance a bit before I shut a node down I do:    $ QC STOP/QUEUE/NEXT /BATCH/NODE  D which allows executing jobs to finish unlike STOP/QUEUE/ON_NODE (for7 some ungodly reason /NEXT is disallowed with /ON_NODE).   % Then in our SYSTARTUP_VMS.COM we use:   E $ QC START/QUEUE/BATCH/NODE/NOON *   !Starts all batch queues on this  node  E !QUEUE_COMMANDS.COM allows you to feed wildcarded queues to any queue  command.E $!This uses the F$GETQUI lexical function to find all matching queues 	 and saves B $!them in a list.  It then executes the specified command for each saved queue.D $!(Note executing queue commands inside a wildcard F$GETQUI loop may
 not work.) $!F $!This supports the following "SHOW QUEUE" selection qualifers for any command:) $! /BATCH               batch queues only , $! /DEVICE              symbiont queues onlyC $! /GENERIC             generic _and_ logical queues (blame DCL for 
 the combo) $!C $!This also supports the following qualifers that even "SHOW QUEUE"  won't take: F $! /NODE[=node_name]    Queues that execute on given node.  "*" and "" allowed.E $! /NOGENERIC           Execution queues (ignores generic and logical  queues).A $! /CONFIRM             Prompts for confirmation before executing  command.C $! /VERIFY              Displays the command it is executing before 	 doing it. E $! /NOON                Continues with next queue if an error occurs.  $!D $!Note there are two special /NODE= cases and useful combinations of
 the above:D $![To keep doc simple we ignore case of a non-generic queues not /ON
 any node.]B $! /GENERIC /NODE=""    !True generic queues only (i.e. no logical queues) , $! /GENERIC /NODE=*     !Logical queues onlyF $! /NODE=*              !execution and logical queues (no true generic queues)  $!F $! P1 is the queue command to execute.  Default is "SHOW QUEUE".  Note thatF $! since "SHOW QUEUE" also supports wildcarded queues if no qualifiers fromB $! the second set above are used we just let DCL do the SHOW QUEUE command.D $! Qualifers may be attached to the end of the command.  If the verb is two? $! words then place P1 in quotes: @QUEUE_COMMANDS "SET QUEUE" *o /CLOSE/NODE  $!= $! The queue name is normally P2 but it can also be after any:
 qualifiers on D $! the command line.  Wildcards are allowed; "*" is default for SHOW QUEUE. $!E $! Use P3 through P8 for additional qualifers.  Qualifiers not in the  aboverB $! lists are used as is, so they must be supported by DCL for that command.  E email me at g o o d m a n @ a c c u w x . c o m .nospam if you want ae
 copy of this.s   ------------------------------    Date: 11 Apr 2002 20:49:54 -0600) From: Joe Pfeiffer <pfeiffer@cs.nmsu.edu>e  Subject: Re: Blade architectures( Message-ID: <1bbscptyul.fsf@cs.nmsu.edu>  @ "Russell P. Holsclaw" <rholsclaw@nospam.xxx.fatline.com> writes: > H > Yet? Actually, we should be preparing for the day when "virtual memory > paging" disappears entirely.   Never.  I > Back in 1988, I went to one of those week-long James Martin seminars in.J > which he prognosticated about the future direction of computing. I don'tE > know if he's still doing these or not, but I haven't heard of them.  > I > Anyway, one of his predictions, based on extrapolations of then-current J > cost/capacity trends, was that semiconductor memory devices would becomeM > cheaper per megabyte than magnetic media (i.e. hard disks). His predictions B > called for the cost-curves to cross sometime in the late 1990's. > M > That hasn't happened yet, because disk capacity has advanced faster than hehJ > predicted, but the trendlines are still moving toward convergence in the > not-too-distant future.   @ I'll grant that it looks likely that at some point semiconductorE memory will get cheaper than disk (no moving parts!).  But that won't * mean we want to get rid of virtual memory.  K > When that happens, there will no longer be any economic justification forwK > virtual memory as such, because the paging devices will be more expensivesM > than the RAM. At that point, the only justification for magnetic media will0 > be non-volatility, not cost. > J > I suspect, however, that people are still so imbued with the paradigm ofN > virtual memory paging that it will continue to exist even after it no longerN > makes economic sense. This is because there are so many people around in the7 > industry who have no memory of the "pre-virtual era".  > M > Over the long haul, however, it will be seen as an idea whose time has come M > ... and gone.   ...and those who continue to cling to the idea will be seen  > as silly old fools.E  C You're confusing paging with paged VM with paging to disk.  Even inoD the limit where the price of memory is 0, we'll still want to have aE distinction between a process's logical space and physical addresses;7E we'll want to be able to start every program at address 0 rather thanoC having ugly relocatable-code kluges, we'll want controllable sharedVC memory between processes, we'll want to be able to protect our codedF from being overwritten by accident...  you're going to have to presentC a better model for all this than paging before paging will go away.o  C And before anybody brings it up, no, segmented address spaces don'te work as well as paging.u --  < Joseph J. Pfeiffer, Jr., Ph.D.       Phone -- (505) 646-1605< Department of Computer Science       FAX   -- (505) 646-1002E New Mexico State University          http://www.cs.nmsu.edu/~pfeifferiM Southwestern NM Regional Science and Engr Fair:  http://www.nmsu.edu/~scifairy   ------------------------------  % Date: Thu, 11 Apr 2002 22:00:50 -0700h" From: Erik Magnuson <erik@cts.com>  Subject: Re: Blade architectures9 Message-ID: <Pine.SOL.4.43.0204112137260.317-100000@erik>e  ) On Fri, 12 Apr 2002, Andrew Reilly wrote:e  9 >On Fri, 12 Apr 2002 04:00:57 +1000, Nick Maclaren wrote:@G >> Neither are likely to help a great deal against errors in privilegedtG >> code.  For that, you need a system like MVS (heaven help us) or, FAR<I >> better but less generally, one based on a proper capability or similar0
 >> design. >.B >Well, in capability space, Amoeba and AS400 are there to be used. >DG >In the protected privileged code space, QNX seems to get a bit of use,uE >controlling nuclear reactors and medical equipment (or so says their6I >ad copy).  You could probably make a pretty large HPC cluster run QNX if  >you felt like it. >eJ >No, I don't use QNX, Amoeba or AS400 myself, and have no affiliation with >etc...c > C QNX is a pretty nifty OS, but it isn't impervious to poorly writteneJ applications - there's one system at work that I have to reboot once everyI few months to clear up some resource allocation issues. With a reasonablyiD well behaved set of applications, QNX should be able to go for years without being touched.*   I The nice thing is that the kernel takes up on the order of 20 KB and mostrE of the OS work is done in user space (albeit as user "root"). OTOH, Io( don't think it would lend itself to HPC.    
 Erik Magnuson   H * One of the electronic engineer rags had an article about using WindozeJ NT for EDA applications - the issue was stability in long simulation runs.G The reviewer pointed out that NT could be very stable if you didn't runa MS Office applications.l   ------------------------------  % Date: Fri, 12 Apr 2002 07:48:44 +0100l* From: "Rainer Giese" <giese@volkswerft.de>$ Subject: Re: EWA-bootdriver messages4 Message-ID: <a95sft$jm5u$1@ID-138444.news.dfncis.de>  & > Is the system's firmware up to date?  I Now it is, and the number of messages have reduced to the half. But stilln more than necessary.G And : how to find someone, who runs through the company to all DPWs ? Ii guess I'll be that.o   ------------------------------  % Date: Fri, 12 Apr 2002 06:51:31 +0200a- From: Didier Morandi <Didier.Morandi@Free.fr>a7 Subject: Re: From comp.sys.dec - deserves exposure heree' Message-ID: <3CB667D2.3CD35CC5@Free.fr>s   :-)   M btw, how is Ken? Last time I tried to access his company web site, the server  did not answer anymore.n   D.   Bill Todd wrote: > E > 4/1/2002  11:15:01 AM  MAYNARD, Mass,   Apr 1, 2002 (APP Online viatJ > JALTEX) -- The fight against the computer industry's biggest merger tookF > a totally unexpected twist today... Ken Olsen, founder of the formerK > Digital Equipment Corporation, Maynard, Ma. stated that he is buying both 9 > Compaq and Hewlett-Packard Co. for $40 billion dollars.s ../..    ------------------------------    Date: 12 Apr 2002 00:05:47 -04000 From: cstacy@theworld.com (Christopher C. Stacy)  Subject: Re: IA64 is not the VAX( Message-ID: <ud6x5a7dw.fsf@theworld.com>  ^ >>>>> On Thu, 11 Apr 2002 23:43:27 +0100, Andrew Harrison SUNUK Consultancy ("Andrew") writes:D  Andrew> Things changed when the Sun 480/410 were released which wasC  Andrew> also the same timeframe for the first HP-PA and MIPS basednD  Andrew> systems, they were apparently faster but no one as far as I*  Andrew> know did any formal benchmarking.  @ Performance analysis of Lisp systems is very complicated, mostlyB because there are lots of things that can be measured (on purpose F and accidently).  Benchmarking Lisp systems is the subject of a numberC of graduate thesis, including the well-known PhD thesis by Gabriel.i  @ While looking around for some numbers for you, I ran across some< benchmarks done in 1991 by David Gadbois and Akira Kurihara.< The program is for ISQRT on very large numbers (thousands of@ digits of precision).  Several different algorithms were tested,> including the built-in ISQRT operator and two different faster> versions (mostly testing Lisp function calling; recursion, in ? one of the algorithsm) and Lisp's infinite precision arithmetic5) (especially the FLOOR and ASH functions).n  F So I have some Lisp benchmarks (for square roots of numbers with 1000 D to 10,000 digits) in front of me showing the Symbolics machine being. at least twice as fast as the SPARCStation 1+.   ------------------------------   Date: 12 Apr 2002 05:20:28 GMT- From: djweath@attglobal.net (Dave Weatherall) # Subject: Re: Is AMD doing an Intel?v5 Message-ID: <DTiotGxQ0bj6-pn2-0KpCpZh2fYyB@localhost>   @ On Mon, 8 Apr 2002 13:59:38 UTC, koehler@encompasserve.org (Bob  Koehler) wrote:h  m > In article <iThr8.1930$fL6.39119@news.cpqcorp.net>, "Fred Kleinsorge" <kleinsorge@star.zko.dec.com> writes:i > > P > > Ultimately, any register used has to be saved on Alpha.  This may or may notO > > be true on IA64, depending on the final design.  You can't really limit thesK > > C complier, you just have to rely on the fact that it will save/restoresN > > registers.  But few drivers are in C, and only one that I know of entirely	 > > in C.i > @ >    In traditional Intel architectures the calling standard hasH >    specified that the calling routine saves any registers it needs and$ >    restores them after the return. > D >    On both VAX and Alpha the architectures specify that the calledH >    routine saves any registers is may modify and restore them prior to >    return. > H >    Although the VAX actually provides the register save and restore inI >    the CALLx/RET instructions, both are valid techniques and both couldr0 >    be used on VAX, Alpha, or Intel processors. > J >    Which is to be used on iVMS?  Sticking with the VAX and Alpha aproachC >    might make porting easier, but may cause headaches other ways.i > G >    I've seen other decisions which would make the Intel approach seemeH >    more consistent (but I can't remember which were public, so I won't >    name them). >   < ISTR having to change some code, after a C compiler upgrade D (Zortech/Symantec), to explicitly save SI, DI when I did some c/X86 F ASM programming years ago. It did feel wrong but not terribly so. All D that PUSH and POP was annoying. Too used to CALLx,RET I suppose.The A 186/286 saved some fuss with PUSHA and POPA but you still had to iE diddle, if you wanted/needed to return something in a register. Then sF again 8086's only had 8(?) registers anyway... A bit different to VAX  (16) and Alpha (32).   -- - Cheers - Dave.   ------------------------------  # Date: Fri, 12 Apr 2002 05:33:59 GMTc$ From: "Upadhyaya" <ups@hotvoice.com>Y Subject: online atricle about sample X11 server architecture and DECwindows architecture?)9 Message-ID: <blut8.34$GN4.177217@cacnews.cac.cpqcorp.net>A  2 Does anyone have pointer to the following article?    : Digital Technical Journal of Digital Equipment Corporation  Volume 2, Number 3, Summer, 1990?                  S. A. McGregor   An overview of the DECwindows J                                   architecture . . . . . . . . . . . . . . 9--15e              S. Angebranndt andlJ                    T. D. Newman   The sample X11 server architecture . . . 16--23             L. P. Treggiari andaJ                   M. D. Collins   Development of the XUI toolkit . . . . . 24--33H                 S. R. Greenwood   The DECwindows user interface language 34--43                 T. M. Spine andaG                    J. L. VanNoy   The evolution of the X user interface?J                                   style  . . . . . . . . . . . . . . . . . 44--51                  R. J. Rost andl             J. D. Friedberg ande<                 P. L. Nishimoto   PEX: a network-transparentJ                                   three-dimensional graphics system  . . . 52--63C                      C. A. Kent   XDPS: a Display PostScript SystemjJ                                   extension for DECwindows . . . . . . . . 64--73                  M. R. Ryan and H                 J. H. VanGilder   The development of DECwindows VMS mail 74--83              D. Mirchandani anda@                       P. Biswas   Ethernet performance of remoteJ                                   DECwindows applications  . . . . . . . . 84--94     Regards,   ------------------------------  # Date: Fri, 12 Apr 2002 03:57:56 GMTv- From: "John E. Malmberg" <wb8tyw@qsl.network> E Subject: Re: Samba V2.0.6 Failed to UID error Network map drive errort* Message-ID: <3CB6782C.7040302@qsl.network>   Paul Richardson wrote:
 > Help me! > H > I trust you good souls may be able to help. I seem to have come so far > and feel that B > I'm so close to getting Samba for VMS v2.0.6 working but for the > following error that > is reported in the SMB.LOG./ > A > [2002/04/11 17:18:07, 0, pid=4680, effective(0, 0), real(0, 0)]r. > CMS_ROOT:[SAMBA.SOURCE.LIB]UTIL_SEC.C;2:(52)A >   Failed to set uid privileges to (-1,8388736) now set to (0,0)eA > [2002/04/11 17:18:07, 0, pid=4680, effective(0, 0), real(0, 0)]i0 > CMS_ROOT:[SAMBA_VMS.SOURCE.LIB]UTIL.C;1:(2456) >   PANIC: failed to set uid5 > ***************************************************l   $x = 8388736 $show sym x 2 $write sys$output f$identifier(x,"NUMBER_TO_NAME")  H For some reason, SMBD is attempting set the user to uic [200,200], that : is the DEFAULT account on OpenVMS and is usually disabled.  C I would recommend checking to make sure that there is a one to one .C relationship between the UICs and the usernames in your SYSUAF and gC RIGHTS Identifiers, and that the USERNAMES and diretories have the W expected UICs on them.  G > I have used SMBPASSWD from the system account to create both the user E > account I want to use and also to join the NT Domain for the domain,I > security flag in smb.conf. I have also checked there is a correspondinga >  VAX account.   H The domain support in SAMBA 2.0.6 is kind of primitive.  I did not have ' a domain controller to test it against.E  F I would recommend first reading the SAMBA-VMS FAQ in the distribution,F and then try the tests in DIAGNOSIS.TXT.  If that file is missing fromE the distribution, then it can be found on the documentation link fromt, your closest mirror of http://www.samba.org/  F See if you can get it working as a standalone server before trying to  join a domain.  H You can also ask on the SAMBA-VMS(at)SAMBA.ORG mailing list, where your G post may stand out more, to see if anyone else has succeeded in getting>" SAMBA on OpenVMS to join a domain.  C The SAMBA 2.0.6 for OpenVMS with few exceptions is bug and feature i" compatible with the UNIX platform.  F So you can also post on comp.protocols.smb and the SAMBA(AT)SAMBA.ORG G mailing lists to see if there are any special tricks to getting domain   membership to work.T   -JohnO wb8tyw@qsl.network Personal Opinion Only    ------------------------------  # Date: Fri, 12 Apr 2002 02:39:48 GMTi+ From: "Kenneth Farmer" <kfarmer@farmer.org>O5 Subject: Re: Time to finally rename VMS back to VMS ?r? Message-ID: <UNrt8.35827$2J2.10283285@typhoon.southeast.rr.com>y  : "JF Mezei" <jfmezei.spamnot@videotron.ca> wrote in message& news:3CB63A72.BFFB813B@videotron.ca...H > Since the engineers are going to be going through all of VMS's modules duringE > the porting effort, wouldn't it be a great time for them to finallys replacedL > all "openvms" back to "vms"  and do away with that "open" stupidity nobody > wants anyways ?i  @ Wait a minute!  For some unknown reason I like it the way it is.  D BTW, can I send you a cash incentive to never mention this again? :)   KF   --   Kenneth Farmer http://www.Tru64.org http://www.OpenVMS.org http://www.LinuxHPTC.com   ------------------------------  % Date: Fri, 12 Apr 2002 01:40:30 -0400a- From: JF Mezei <jfmezei.spamnot@videotron.ca>   Subject: Total cost of ownership, Message-ID: <3CB67335.F73875C7@videotron.ca>  F Some time ago, it was mentioned that VMS had lowest cost of onwership.  # Here is Compaq's take on the issue:t  ? Compaq NonStop Himalaya Server Ranks #1 in Lowest Total Cost of    Ownership Study      4/11/02 10:01am   U   HOUSTON, April 11 /PRNewswire-FirstCall/ -- For the second consecutive time, CompaqbI   Computer Corporation's (NYSE: CPQ) NonStop Himalaya server is rated the  industry leader inL   lowest Total Cost of Ownership (TCO) study of high-end enterprise systems.  M   The Standish Group International, an independent research and advisory firmc focusing on theeL   mission-critical marketplace, has released the results of their latest TCO study entitled, Dollars to"   Cents: TCO in the Trenches 2002.  J   Compaq NonStop systems overall ranked lower than any high-end enterprise
 system in its L   competitive range of price performance, and first among all systems in the study when factoring,   in "application cost" and "downtime cost".  N   Jim Johnson, chairman, The Standish Group, said, "Looking at cost from a one dimensional point J   of view, such as the basic cost of hardware, the NonStop Himalaya server doesn't appear that L   competitive, but adding infrastructure software like a relational database and manpower toeM   maintain the applications and operation -- it starts to be very competitive  against more perceivedH   cheaper alternatives." Johnson continued, "If you then add the cost of downtime into mix -- the*   value of NonStop servers becomes clear."  G   Pauline Nist, vice president and general manager for Compaq's NonStop. Division, said, "The needlK   for continuous availability systems has gone mainstream, leading more and, more vendors to sellL   the concept of continuous availability, without delivering the goods. Most continuous availabilitytM   discussions neglect real-world availability issues by ignoring downtime forb planned maintenanceiB   and upgrades, or by measuring downtime in relationship to system availability rather than   application availability."  L   Nist continued, "The Standish Group's TCO study results reflect a true and complete view of totalN   system cost, providing further evidence that when considering all associated costs, including theG   biggest cost of all -- the cost of downtime -- Compaq NonStop systems  achieve the overall lowest+   total cost of ownership in the industry."h  I   The Standish Group research study compared overall TCO costs of sixteenn popular system typesN   to evaluate the operational and cost environment of the various systems. The study results areaD   based exclusively on customer data input from over 1,200 end users representing over 2,000 L   case studies and more than fifty applications. The study examined critical TCO factors such as N   basic system cost, mission-critical application costs, hardware and software costs, manpower "   costs, and the cost of downtime.  M   For more information on The Standish Group's TCO research study results, gol
 to the Compaqh<   website at: http://nonstop.compaq.com/view.asp?IO=TCORPT .   ------------------------------  % Date: Thu, 11 Apr 2002 23:36:24 -0500 , From: "Rich Jordan" <rjordan@mindspring.com>, Subject: Re: WIreless networking + VMS/Tru642 Message-ID: <a95o5a$j2k$1@slb6.atl.mindspring.net>   Peter,H      I currently do just that with a pair of DEC Roamabout access pointsK (V2.4 firmware) with 915MHz PMCIA radio cards (the older Wavelan cards thatpL topped out at 2 Mbits/sec) in a wireless bridge configuration (where the twoB APs talk to eachother and act as a bridge between the upstairs andL downstairs LAN segments).  It works perfectly and the access points were notK too expensive on Ebay.  In addition, these old access points can reportedly,H be upgraded to a newer firmware release (5.x from Enterasys) and use theG newer Enterasys/Cabletron Roamabout 802.11 high rate cards to get up tor 11Mbits/sec.  J The nice thing is its all done at the LAN level so you don't need to worryK about special drivers or hardware; just plug the APs into hubs at each end, J or directly into your Alpha with a crossover if thats the only node there.K One nice thing about using the old cards I have that offsets the lower data2J rate is that the myriad snoopers driving around with notebooks looking forE bandwidth to steal rarely look at the older frequencies, and unlike afI 'roaming' access point, these point-to-point bridges don't accept outside0J roamers.  They only talk to eachother (unless someone figures out a way to spoof).D  G      I've heard that at least one of the new/cheap home-type APs can bedH configured in a point-to-point bridge configuration also; not sure which@ brand (Linksys?  D-Link?) but that would be another alternative.   Rich Jordan   I Peter Watkinson wrote in message <3cb56477.1983265@news.cable.ntl.com>...bE >On Thu, 11 Apr 2002 06:58:06 +0000 (UTC), mustang@ucc.asn.au.invalid  >wrote:. >. >W? >Quite simple really. I want to connect my Cable modem internet F >connection which is downstairs to a VMS/Tru64 workstation (when I getA >it) which will be upstairs without having to drill a hole in they
 >ceiling. :-)o	 > .......o >cheers, >m >  >Peter Watkinson >peter.watkinson1@ntlworld.com   ------------------------------   End of INFO-VAX 2002.201 ************************