1 INFO-VAX	Sat, 03 Mar 2001	Volume 2001 : Issue 124       Contents:. Re: A Compaq Question a VMSer Would Never Ask.. Re: A Compaq Qwestion a VMSer Would Never Ask.. Re: A Compaq Qwestion a VMSer Would Never Ask.. Re: A Compaq Qwestion a VMSer Would Never Ask.= Re: Alpha: game over.  Was Future outlook for OpenVMS admins? = Re: Alpha: game over.  Was Future outlook for OpenVMS admins? 9 Re: Any VMS Database product compatible w/ Microsoft SQL? 6 Re: Compaq to transfer clustering technology to Oracle6 Re: Compaq to transfer clustering technology to Oracle6 Re: Compaq to transfer clustering technology to Oracle6 Re: Compaq to transfer clustering technology to Oracle6 Re: Compaq to transfer clustering technology to Oracle Re: DCL content (long) Re: Decserver 90M software' Re: Drop C/C++ better sooner than later ' Re: Drop C/C++ better sooner than later ' Re: Drop C/C++ better sooner than later ' Re: Drop C/C++ better sooner than later ' Re: Drop C/C++ better sooner than later ' Re: Drop C/C++ better sooner than later ' Re: Drop C/C++ better sooner than later ' Re: Drop C/C++ better sooner than later ' Re: Drop C/C++ better sooner than later ' Re: Drop C/C++ better sooner than later 3 Re: Drop C/C++ better sooner than later - continued 0 How to licence boot strapping a hobbyist system?4 Re: How to licence boot strapping a hobbyist system?4 Re: How to licence boot strapping a hobbyist system?/ Re: HP DAT drive now functional but JAVA broken  Just Two Days Left to Vote- Low Cost Alpha Price Drop - Now $850 Complete  Re: Low cost workstations  Re: Low cost workstations  Re: Low cost workstations  Re: Low cost workstations  Re: more "disk" IO comparisons Re: New Sun Blade  Re: OpenVMS goodies  Re: OpenVMS goodies # Re: SAMBA 2.0.6 seems to hate me...  Re: Satellite MV2000 won't boot C Re: Sun Blade 100 with a VMS monitor ? (was: Low cost workstations)  Re: Umbrella in Germany 7 Re: VMS V7.2+ bug - directory modification date changes * Re: [INFO] wrong answer in Mgmt agents FAQ  F ----------------------------------------------------------------------  % Date: Sat, 03 Mar 2001 10:25:50 +0000 ) From: Christof Brass <brass@infopuls.com> 7 Subject: Re: A Compaq Question a VMSer Would Never Ask. , Message-ID: <3AA0C6AE.B94932BA@infopuls.com>   "Terry C. Shannon" wrote:  > 9 > http://web14.compaq.com/falco/detail.asp?FAQnum=FAQ2859  >  > -- > Terry C. Shannon > Consultant and Publisher > Shannon Knows Compaq" > email: terryshannon@mediaone.net& > Web (info on SKC):  www.acersoft.com   Surprisingly the answers to this question are always wrong which is especially harmful as the people asking this somewhat silly question have no clue to understand that the answers are wrong. :-)    ------------------------------  % Date: Sat, 03 Mar 2001 02:33:47 -0500 ' From: Jim Becker <jbecker@ui.urban.org> 7 Subject: Re: A Compaq Qwestion a VMSer Would Never Ask. , Message-ID: <3AA09E5B.184703DA@ui.urban.org>  6 This "any key" story was related to me some years ago.  D A couple of programmers had a rough prototype of a program they wereE working on. They wanted to test the interface on someone without much ? computer expertise, so they asked the department secretary. The C secretary reached a point where the program said, "Press any key to 
 continue."  D The secretary said, "Any key... Well, my name begins with J, so I'll hit J."   A "No, not that one!" was the instant reaction of both programmers.   8 "But it says to press any key," protested the secretary.  F Programmer #1 said, "When it says 'any key' it really means the Return key."   4 Programmer #2: "Huh? 'Any key' means the space bar!"   --
 Jim Becker+ The Urban Institute (http://www.urban.org/) ' Encompass (http://www.encompassus.org/) 5 ESILUG (http://eisner.encompasserve.org/lugs/esilug/)    ------------------------------  # Date: Sat, 03 Mar 2001 11:03:14 GMT 7 From: "Lyle W. West" <lyle.west@childrenshc.org.nospam> 7 Subject: Re: A Compaq Qwestion a VMSer Would Never Ask. 6 Message-ID: <3AA07B12.6B698726@childrenshc.org.nospam>   Then there is the:  8 	Press any key to continue,  press any other key to quit   --  0 My opinions seldom reflect those of my employer.   Lyle W. West   ------------------------------    Date: 03 Mar 2001 21:03:41 +0800, From: Paul Repacholi <prep@prep.synonet.com>7 Subject: Re: A Compaq Qwestion a VMSer Would Never Ask. - Message-ID: <87snkv6m36.fsf@prep.synonet.com>   ) Jim Becker <jbecker@ui.urban.org> writes:   8 > This "any key" story was related to me some years ago.  F > A couple of programmers had a rough prototype of a program they wereB > working on. They wanted to test the interface on someone withoutF > much computer expertise, so they asked the department secretary. TheE > secretary reached a point where the program said, "Press any key to  > continue."  F A couple of us had some keys labeled "ANY" YEARS ago. Yeah, they where square, and red!  0 Ten min after installing the first one, 'Bleep'!   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  % Date: Sat, 03 Mar 2001 13:14:44 +0100   From: Paul Sture <paul@sture.ch>F Subject: Re: Alpha: game over.  Was Future outlook for OpenVMS admins?+ Message-ID: <VA.000002d6.9eaea7d5@sture.ch>   M In article <6IUn6.2230$TL4.867333@typhoon.ne.mediaone.net>, Terry C. Shannon   wrote:6 > From: "Terry C. Shannon" <terryshannon@mediaone.net> > Newsgroups: comp.os.vms L > Subject: Re: Alpha: game over.  Was RE: Future outlook for OpenVMS admins?% > Date: Fri, 02 Mar 2001 21:58:26 GMT  >  > J > "andrew harrison" <andrew.nospam@uk.sun.com> factually stated in message& > news:3A9BEF25.B52DED8F@uk.sun.com... > > Carl Perkins wrote:  > > >  > > > K > > > It is also the case that they *have* done some work to solve the cost K > > > problem. You can buy a fully functiounal DS10 system running VMS in a L > > > decent configuration for under $8000. Compare this to the cheapest VMSF > > > based system you could buy 5 years ago and I think you'll find a > reduction J > > > in the entry level system cost of nearly a half. It would be good if > theyE > > > could cut it in half again, but they may not think that this is  > necessary. > > >  > > 7 > > This sounds great but isn't because the market that 5 > > OpenVMS operates in has moved much further than a 6 > > halving in cost for an entry level system over the > > same five year period. >  > Point well taken!  >  > > 7 > > Just to put the OpenVMS DS10 pricing in perspective 5 > > you can get a SunBlade 100 with Solaris 8 for 995  > > dollars. > I > Not quite apples to apples, but it *is* 500MHz system to 500MHz system.   M Definitely not apples to apples. I haven't looked at the specs myself, but a  L colleague who has, tells me that the guts are cheap - IDE disks, slow SCSI, K small cache, probably little room to upgrade etc. The processor is a Sparc  C IIe, which has him wondering what the "e" means in practical terms.   M Didn't I read here that VMS runs like a dog on IDE disks? Dunno about Tru64,  J but that could be a real showstopper in terms of Compaq producing a rival  system.  > > 7 > > Alternatively if you really want to spend 8,000 you & > > could buy a 750 Mhz SunBlade 1000. > M > If and when any UltraSparc III parts show up. Haven't heard of any presence L > in the channels, and the USIII-based systems remain "constrained" on Sun's0 > Web site. Not for much longer, one would hope. >    ___ 
 Paul Sture Switzerland    ------------------------------    Date: 04 Mar 2001 00:30:00 +0800, From: Paul Repacholi <prep@prep.synonet.com>F Subject: Re: Alpha: game over.  Was Future outlook for OpenVMS admins?- Message-ID: <87bsri6cjb.fsf@prep.synonet.com>   " Paul Sture <paul@sture.ch> writes:  @ > Definitely not apples to apples. I haven't looked at the specsE > myself, but a colleague who has, tells me that the guts are cheap - D > IDE disks, slow SCSI, small cache, probably little room to upgradeE > etc. The processor is a Sparc IIe, which has him wondering what the  > "e" means in practical terms.     ^2 Larger on chip cache was the comment in comp.arch.  A > Didn't I read here that VMS runs like a dog on IDE disks? Dunno F > about Tru64, but that could be a real showstopper in terms of Compaq > producing a rival system.   C Given the miniscule delta for a SCSI chip, vs and IDE, It is stupid + to go ideot. Drive cost is a pain though :(    --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  % Date: Sat, 03 Mar 2001 15:06:56 +0100   From: Paul Sture <paul@sture.ch>B Subject: Re: Any VMS Database product compatible w/ Microsoft SQL?+ Message-ID: <VA.000002da.9f156142@sture.ch>   8 In article <3A9D9221.10801@fsi.net>, Mark E. Levy wrote:% > From: "Mark E. Levy" <mark@fsi.net>  > Newsgroups: comp.os.vms @ > Subject: Any VMS Database product compatible w/ Microsoft SQL?' > Date: Wed, 28 Feb 2001 18:04:49 -0600  > H > I have an application that requires Microsoft SQL server. For reasons G > obvious to readers of this NG, I'd prefer to have it live on my OVMS  D > box. Pathworks does a great job of making lowly PCs think they're K > talking to an NT server. Are there any OVMS based database products that  6 > "look" like Microsoft SQL to an unsuspecting PC app? > G If I've read it correctly, the Attunity stuff does that sort of thing.  F Apparently it can even do joins across different database products on  different platforms.   ___ 
 Paul Sture Switzerland    ------------------------------  # Date: Sat, 03 Mar 2001 08:45:08 GMT  From: Dirk Munk <munk@home.nl>? Subject: Re: Compaq to transfer clustering technology to Oracle ' Message-ID: <3AA0AF12.24CB7816@home.nl>   Q With this "Oracle cluster on NT" debate everybody seems to forget that Oracle has Q its own operating system on Intel boxes. I don't know very much about it (haven't G seen it), but it seems to be a kind of bootable Oracle database engine. N Incorporating cluster technology in this OS would make far more sence from theP Oracle point of view. Suppose you want to run a Oracle database on a cheap IntelO box, does it realy matter if that database runs on Windooz, or on a Oracle OS ? O The choice will be a lot easier if Oracle can say "but you can cluster with our 6 OS". And it will most likely be a lot more stable too.   ------------------------------  % Date: Sat, 03 Mar 2001 12:18:13 +0000 ) From: Christof Brass <brass@infopuls.com> ? Subject: Re: Compaq to transfer clustering technology to Oracle , Message-ID: <3AA0E105.C50E28D5@infopuls.com>   Dirk Munk wrote: > S > With this "Oracle cluster on NT" debate everybody seems to forget that Oracle has S > its own operating system on Intel boxes. I don't know very much about it (haven't I > seen it), but it seems to be a kind of bootable Oracle database engine. P > Incorporating cluster technology in this OS would make far more sence from theR > Oracle point of view. Suppose you want to run a Oracle database on a cheap IntelQ > box, does it realy matter if that database runs on Windooz, or on a Oracle OS ? Q > The choice will be a lot easier if Oracle can say "but you can cluster with our 8 > OS". And it will most likely be a lot more stable too.   Are you sure about the Intel box? I thought that "raw iron" was/is targeted to SPARC and is in fact a downsized Solaris (eventually mainly the kernel) and Oracle8i. Do you have any references?   ------------------------------    Date: 03 Mar 2001 22:17:52 +0800, From: Paul Repacholi <prep@prep.synonet.com>? Subject: Re: Compaq to transfer clustering technology to Oracle - Message-ID: <87g0gv6inj.fsf@prep.synonet.com>   - "Main, Kerry" <Kerry.Main@compaq.com> writes:     E > [rathole alert - we have been over this many times before, but what  > the hey..]  E > A shared everything cluster like OpenVMS is not the answer to every  > situation.  @ But Kerry, what does a VMS cluster 'share'? Not disks, queues orD anything except active locks on multiple node. Or active on one, andF mastered on another. There is almost no sharing of resources, there is sharing of *USE*.   A ( Excluding direct attached served storage... etc And cluges like C shared SCSI cruftishness. Mind you, look at the gatespew clustering  for that one. )   D Now IF you can scale the DLM performance, then you can get an almostB arbitarily high reliability and resiliance to failure. You can run? parellel updates, if the first fails, the other will finish the C update, even if this 'failure' is a slow reponce due to a transient C load on the 'primary'. You co-ordinate the useage, and 'allocation' B will/can flow from that. Even worked for real time, but it sure as+ hell isn't 'standard' or POSUX or the like!    --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  # Date: Sat, 03 Mar 2001 15:11:56 GMT  From: Dirk Munk <munk@home.nl>? Subject: Re: Compaq to transfer clustering technology to Oracle ' Message-ID: <3AA109B9.D31FAA04@home.nl>   Y I am under the impression it is Intel, but I will try to find out on the Oracle web site.  The only problem is that one needs a 1 year academic study to just understand the Oracle product portfolio, and how it is all suppose to interconnect.   Christof Brass wrote:    > Dirk Munk wrote: > > U > > With this "Oracle cluster on NT" debate everybody seems to forget that Oracle has U > > its own operating system on Intel boxes. I don't know very much about it (haven't K > > seen it), but it seems to be a kind of bootable Oracle database engine. R > > Incorporating cluster technology in this OS would make far more sence from theT > > Oracle point of view. Suppose you want to run a Oracle database on a cheap IntelS > > box, does it realy matter if that database runs on Windooz, or on a Oracle OS ? S > > The choice will be a lot easier if Oracle can say "but you can cluster with our : > > OS". And it will most likely be a lot more stable too. >  > Are you sure about the Intel box? I thought that "raw iron" was/is targeted to SPARC and is in fact a downsized Solaris (eventually mainly the kernel) and Oracle8i. Do you have any references?   ------------------------------   Date: 3 Mar 2001 12:59:41 -0500 + From: young_r@encompasserve.org (Rob Young) ? Subject: Re: Compaq to transfer clustering technology to Oracle 3 Message-ID: <TT7Z6HEnsT1K@eisner.encompasserve.org>   9 	I got this one wrong.  One in the know pointed out that:     @ 	Folks think MS is being given TruCluster lock stock and barrel. 	That is not the case. 	 ; 	Oracle will license some of the TruCluster components (not 9 	including the cluster filesystem and single-system-image : 	technology as they are too deeply imbedded into Tru64) to+ 	help the deployment of parallel databases.   9 	The new integrated stuff will first appear on Tru64, for ? 	obvious reasons, Tru64 will become a development and reference @ 	platform for the technology.  Later, other first tier platforms$ 	get it, Solaris, HP/UX, AIX and NT.   ---   : 	My apologies to one and all for further seeding confusion 	in this matter.   				Rob    ------------------------------  % Date: Sat, 03 Mar 2001 13:14:45 +0100   From: Paul Sture <paul@sture.ch> Subject: Re: DCL content (long) + Message-ID: <VA.000002d8.9eaeabb5@sture.ch>   F In article <3A9F1576.98CE07D3@earthlink.net>, David J. Dachtera wrote:9 > From: "David J. Dachtera" <djesys.nospam@earthlink.net>  > Newsgroups: comp.os.vms ! > Subject: Re: DCL content (long) ' > Date: Thu, 01 Mar 2001 21:37:26 -0600  >  > Syltrem wrote: > > L > > > Another "wish list" item for me would be to have the ARCH_NAME keywordI > > > work between cluster nodes. For now, HW_NAME is all you can get for   > > > other than the local node. > >  > > ???  > > & > > $ say f$gets("arch_name","kronos")	 > > Alpha0& > > $ say f$gets("arch_name","pallas") > > VAXt >  > Thanx! I'll look into that.  > A > Now - do you happen to know how to get ALLOCLASS between nodes? 3 > (Don't ask why - you don't wanna know (do you?).)  >e This works on 7.2:   $! SHCLU.COM $! --------- $!C $ IF F$GETSYI("CLUSTER_MEMBER") .EQS. "FALSE" THEN GOTO NOT_CLUSTERM $ say := WRITE SYS$OUTPUTa $ CONTEXT = "" $START:/ $ id = F$CSID (CONTEXT)p $ IF id .EQS. "" THEN EXIT& $ nodename = F$GETSYI ("NODENAME",,id)' $ arch_name = F$GETSYI("ARCH_NAME",,id)o1 $ alloclass = f$string(F$GETSYI("ALLOCLASS",,id))e) $ soft_type = F$GETSYI("NODE_SWTYPE",,id)t) $ soft_vers = F$GETSYI("NODE_SWVERS",,id) % $ hdwe_name = F$GETSYI("HW_NAME",,id)l4 $ say f$fao("!9AS !AS - !AS !6AS !AS $!AS$ (!AS)", -; nodename + ":", "''id'", soft_type, arch_name, soft_vers, -  alloclass, hdwe_name)  $ GOTO START
 $NOT_CLUSTER:C/ $ WRITE SYS$OUTPUT "Not a member of a cluster."t $ EXIT   $ @shclu? SPEEDY:   00010002 - VMS  VAX    V7.2 $2$ (VAXstation 3100/SPX)eI ALPHA1:   00010003 - VMS  Alpha  V7.2 $1$ (Digital Personal WorkStation )C   ___e
 Paul Sture Switzerland    ------------------------------  $ Date: Sat, 3 Mar 2001 11:30:27 -0500. From: "Ted Medenblik" <ted.medenblik@duke.edu># Subject: Re: Decserver 90M softwaree( Message-ID: <97r670$e08$1@news.duke.edu>  K Decserver 90M softwareDigital Networks bought the stuff from Cabletron that / bought the stuff from DEC.  http://www.dnpg.com    Later, Tede9 "Olson, Ingemar" <IOlson@dairyworld.com> wrote in message G news:763C579A82F7D3118EE400D0B74723D10166B881@exchsrv.dairyworld.com...iG We have a number of DS90M's, some older than others, but all with flashd memory. I The software version of the older ones is usually V1.5, whereas the newer0 ones are V2.2.K I would like to upgrade the V1.5's to V2.2, but the MNENG2.SYS file we havelD isn't the latest. In fact it says "... SW V1.4" somewhere inside it.L Presumably this means that all the existing ones are booting from flash, and( so never need to access the MNENG2 file.L I know I should be able to reload the flash memory from the load server, but7 obviously I would need a more current MNENG2.SYS first.v/ Does anyone know where I can find such a beast?.
 Ingemar OlsonC   ------------------------------   Date: 3 Mar 2001 08:24:57 GMTF- From: djweath@attglobal.net (Dave Weatherall) 0 Subject: Re: Drop C/C++ better sooner than later5 Message-ID: <DTiotGxQ0bj6-pn2-b9XBNjo5mdn7@localhost>:  C On Fri, 2 Mar 2001 19:59:45, paddy.o'brien@zzz.tg.nsw.gov.au wrote:a   > Dave,i > F > >A bit of analysis with FOR /LIS /MAC allowed me to restructure (or I > >correctly structure :-)) the code to enable the compiler to eliminate - > >the waste. No more overruns.  > + > Yep, I've done that on VAX, but Alpha...?e  ? Me too! I still do my debugging on VAX for this very reason :-)r  H > >I have to confess that an alogorithm change to use POINTERS helped a I > >lot in two areas. It enabled me to move the IF/ELSEIF/ELSE/ENDIF from rH > >the run-time code to the initialisation block. However, the pointers G > >point to static elements so no rubbish pointers are (theoretically) t > >possible. (Ha Ha!!) > L > No comprendo.  Pointers are optimisation inhibitors.  Do you somehow mean K > it gave you a clue in this area?  Is your "initialisation block" a BLOCK  L > DATA routine, or a run-time which does A(:N) = 0, for example (this being , > more efficient in executable size/layout). > L > I maintain many large Fortran applications and am genuinely interested in H > understanding your enhancements, or the point (pun :-) you are making K > here.  Would you be prepared to mail off-line examples of the pre-/post- n > code?.  B Sorry Paddy, I don't think I was 100% wake when I wrote the above.  D To explain, in our real-time test system we have vatious data types D which can be recorded. Each item to be recorded is held in array of F record which contains type, scaling factor etc as well as indeces intoE the big arrays that hold the data at run time. So at runtime we had :t   	if (ac_serial) then) 	   call get_ac_serial_word (parameters) e   	else if (milbus)e% 	   call get_Milbus_word (parameters)-   	else if (...) 	endif  F VAX Fortran V6.5, which we use for our KAV-30/ELN system, understably 2 makes heavy weather of  processing multiple depth F 'record.element(i).item(j) style' constructs and the Parameters above E are just such references. By adding a 'pointer to data word' item in  C the control structure I could move the above to the initialisation o7 routines (block was a stupid word to use) such that :- e   	if (ac_serial) then/ 	   data_ptr = get_ac_serial_word (parameters) p   	else if (milbus)e, 	   data_ptr =  get_Milbus_word (parameters)   	else if (...) 	endif  & 	record.element(i).data_ptr = data_ptr   and at run time I simply haveh   	pointer (data_ptr,data_word)  ..& 	data_ptr = record.element(i).data_ptr   	recorded_word = data_word  F The items being pointed to are static i.e. big structures/records so IF (should) never have invalid pointers. The self-mocking laugh from lastF time was because the latter presumes I set the table up correctly and D on at least once occasion I hadn't. What was that about writing bad  stuff in any language...  F We later used the same technique in the program we use to analyse the F recorded data. The performance increase on VAX was enormous. Alpha gotA much quicker too but the percentage improvement (IIRC) wasn't as  E great. It just shows that Alpha's extra registers and register width r3 allow the optimiser to do a lot of the work itself.    Cheers - Dave.   ------------------------------  # Date: Sat, 03 Mar 2001 10:18:42 GMT + From: "Nikita V. Belenki" <public@kits.net>>0 Subject: Re: Drop C/C++ better sooner than later8 Message-ID: <6y3o6.526$Up.29452@sea-read.news.verio.net>  6 "Christof Brass" <brass@infopuls.com> wrote in message& news:3A9EDDEC.845CA702@infopuls.com...  J > > > > When working with dynamically allocated memory structures, pointer* > > > > arithmetic and handling is a must. > > > *Only* in C/C++.J > > *Everywhere* on the current hardware. If *you* are writing the libraryE > > that does memory management and hides pointer arithmetic from theg > > other parties.K > *Only* in crap languages. BTW this has nothing to do with the HW. Insteadr5 > with the language, namely the compiler/interpreter.s& > Ever heard of a type system or a GC?  G Ever heard of a GC *implementation* written *without* pointer handling?s   Kit.   ------------------------------  % Date: Sat, 03 Mar 2001 10:40:32 +0000 ) From: Christof Brass <brass@infopuls.com>s0 Subject: Re: Drop C/C++ better sooner than later, Message-ID: <3AA0CA20.951D7B8A@infopuls.com>   "Nikita V. Belenki" wrote: > 8 > "Christof Brass" <brass@infopuls.com> wrote in message( > news:3A9EDDEC.845CA702@infopuls.com... > L > > > > > When working with dynamically allocated memory structures, pointer, > > > > > arithmetic and handling is a must. > > > > *Only* in C/C++.L > > > *Everywhere* on the current hardware. If *you* are writing the libraryG > > > that does memory management and hides pointer arithmetic from theb > > > other parties.M > > *Only* in crap languages. BTW this has nothing to do with the HW. Instead 7 > > with the language, namely the compiler/interpreter.o( > > Ever heard of a type system or a GC? > I > Ever heard of a GC *implementation* written *without* pointer handling?e >  > Kit.  < Ever heard of a GC *implementation* supplied in a library???   ------------------------------  # Date: Sat, 03 Mar 2001 10:44:19 GMTl= From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) 0 Subject: Re: Drop C/C++ better sooner than later0 Message-ID: <009F872D.74C58B92@SendSpamHere.ORG>  e In article <DTiotGxQ0bj6-pn2-b9XBNjo5mdn7@localhost>, djweath@attglobal.net (Dave Weatherall) writes:eD >On Fri, 2 Mar 2001 19:59:45, paddy.o'brien@zzz.tg.nsw.gov.au wrote: >l >> Dave, >> eG >> >A bit of analysis with FOR /LIS /MAC allowed me to restructure (or  J >> >correctly structure :-)) the code to enable the compiler to eliminate   >> >the waste. No more overruns. >> h, >> Yep, I've done that on VAX, but Alpha...? >c@ >Me too! I still do my debugging on VAX for this very reason :-)  D Why?  Why do folks here find such an elegant and simple architectureD and instruction set to be so difficult to fathom?  Alpha is pushing E the decade mark and people still haven't figured it out?  What is it,IC a mental block from reading VAX instruction code for so long?  I amh just curious?      --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COM              O city, n., 1. a place where trees are cut down and streets are named after them.6   ------------------------------  % Date: Sat, 03 Mar 2001 13:14:44 +0100-  From: Paul Sture <paul@sture.ch>0 Subject: Re: Drop C/C++ better sooner than later+ Message-ID: <VA.000002d7.9eaea9f2@sture.ch>b  F In article <OF2253DC41.DB6B5919-ON88256A03.00083164@foundation.com>,   wrote:J > I've always said that programmers should be forced to write in assemblerC > for a year before being let loose in the real world. The more youdE > understand, the better you do regardless of the tools you're using.0 > 8 I was never allowed to learn assembler in early days :-(  H BUT. I had to fit an ever increasing workload into a finite system, and I that taught me how to accomplish dramatic performance improvements using nJ the HLL tools available. While the assembler guys would be arguing (often F vociferously) the merits of one technique or another, usually talking K about CPU efficiency, I'd be quietly redesigning and rewriting stuff in my pG HLL to cut out whole processes at a time. Of course, I benefitted from f+ those assembler guys' arguments as well :-)l  F One point I haven't seen raised yet in this discussion is that of the K hardware available to junior programmers. It's all very well enticing them iJ into your employ with promises of the latest and greatest systems, but it K would be a useful exercise for them to at least test all their programs on   old slow stuff.   K A case in point, and I'm sure this is far from unique. Back in DOS and 286  B days, one UK software house did all their development on low spec J Amstrads, with the end result that it would run satisfactorily on all PCs E out there. It was hugely successful. In contrast their first Windows IK version (of which they were immensely proud) ran like a dog even on a well c3 spec'd 486. A lot of folks were unhappy about that.a   ___b
 Paul Sture Switzerlandu   ------------------------------  % Date: Sat, 03 Mar 2001 12:36:37 +0000-) From: Christof Brass <brass@infopuls.com>10 Subject: Re: Drop C/C++ better sooner than later, Message-ID: <3AA0E555.3A212AD9@infopuls.com>   Paul Sture wrote:O > F > In article <OF2253DC41.DB6B5919-ON88256A03.00083164@foundation.com>, > wrote:L > > I've always said that programmers should be forced to write in assemblerE > > for a year before being let loose in the real world. The more you G > > understand, the better you do regardless of the tools you're using.t > > : > I was never allowed to learn assembler in early days :-( > I > BUT. I had to fit an ever increasing workload into a finite system, and J > that taught me how to accomplish dramatic performance improvements usingK > the HLL tools available. While the assembler guys would be arguing (often G > vociferously) the merits of one technique or another, usually talking L > about CPU efficiency, I'd be quietly redesigning and rewriting stuff in myH > HLL to cut out whole processes at a time. Of course, I benefitted from- > those assembler guys' arguments as well :-). > G > One point I haven't seen raised yet in this discussion is that of therL > hardware available to junior programmers. It's all very well enticing themK > into your employ with promises of the latest and greatest systems, but itnL > would be a useful exercise for them to at least test all their programs on > old slow stuff.- > L > A case in point, and I'm sure this is far from unique. Back in DOS and 286C > days, one UK software house did all their development on low specoK > Amstrads, with the end result that it would run satisfactorily on all PCsiF > out there. It was hugely successful. In contrast their first WindowsL > version (of which they were immensely proud) ran like a dog even on a well5 > spec'd 486. A lot of folks were unhappy about that.t >  > ___s > Paul Sture
 > Switzerlande  Happened with WNT. The first public version has been developed on very powerful WSs equipped with lots of RAM. The customers weren't happy because at that time most of them tried install it on the machines they already had which had not much RAM and little processing power.   ------------------------------    Date: 03 Mar 2001 15:54:16 +0100G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de>f0 Subject: Re: Drop C/C++ better sooner than laterH Message-ID: <y4g0gux5rb.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>  ( "Rick Cadruvi" <rick@rdperf.com> writes:  8 > I'm not sure what the point is with size of compilers.? > Most of the size of the code for a compiler isn't necessarily0B > related to the complexity of the language.  Compilers use tablesA > for parsing/lexical analysis.  The tricky part of compilers andk3 > what tends to add size is the code optimizations.   sK All granted. But the claim was "Ada has orders of magnitude more complexitymM than C", and I was looking for zeroth order effects. The data do not bear out  the allegation at all, IMNSHO.   	Jan   ------------------------------    Date: 03 Mar 2001 23:18:31 +0800, From: Paul Repacholi <prep@prep.synonet.com>0 Subject: Re: Drop C/C++ better sooner than later- Message-ID: <87ofvi6fug.fsf@prep.synonet.com>b  ? system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) writes:n  7 > In article <DTiotGxQ0bj6-pn2-b9XBNjo5mdn7@localhost>,a1 > djweath@attglobal.net (Dave Weatherall) writes:b  B > >Me too! I still do my debugging on VAX for this very reason :-)  F > Why?  Why do folks here find such an elegant and simple architectureE > and instruction set to be so difficult to fathom?  Alpha is pushingeC > the decade mark and people still haven't figured it out?  What is F > it, a mental block from reading VAX instruction code for so long?  I > am just curious?  E It is not ( directly ) the machine, it is the compiler. Ever tried tou, guess the code from a snippet of GEM output?   -- /< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.f@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------   Date: 3 Mar 2001 16:41:07 GMTo1 From: JONESD@er6.eng.ohio-state.edu (David Jones)s0 Subject: Re: Drop C/C++ better sooner than later: Message-ID: <97r6r3$403$1@charm.magnus.acs.ohio-state.edu>  0 In message <009F872D.74C58B92@SendSpamHere.ORG>,B    system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) writes:E >Why?  Why do folks here find such an elegant and simple architecturel2 >and instruction set to be so difficult to fathom?  E It's not the instruction set itself, but the way the instructions are H used to build a procedure.  Instead of a 16-bit entry mask that tell youF what registers get saved and restored on RET, you have to sort out theJ functions prologue instructions.  References to the argument list are moreH obvious on the VAX, a reference to R16..R22 on Alpha only means a callerK argument if the procedure hasn't yet called another function.  It is easier0I to comprehend a CALLS, CALLG, JSR, etc, on VAX than the multi-instructione- equivalent on Alpha (the old 'semantic gap').n  M I think it would be helpful if the disasemblers (debugger, SDA) had an optionvH to substitute pseudo-instructions for instructions that use R31.  It theH effect of an instruction is to clear a register or invert its bits, then say CLEAR or COMPLEMENT.   >Alpha is pushing F >the decade mark and people still haven't figured it out?  What is it,D >a mental block from reading VAX instruction code for so long?  I am >just curious?  F I think DEC discouraged direct assembly language programming on Alpha,D one of the principles of RISC being that sophisticated compilers are3 required to make effective use of the architecture.p    < David L. Jones               |      Phone:    (614) 292-6929- Ohio State University        |      Internet:wL 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.    ------------------------------  # Date: Sat, 03 Mar 2001 17:04:02 GMT,= From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-)$0 Subject: Re: Drop C/C++ better sooner than later0 Message-ID: <009F8762.8071D88E@SendSpamHere.ORG>  n In article <97r6r3$403$1@charm.magnus.acs.ohio-state.edu>, JONESD@er6.eng.ohio-state.edu (David Jones) writes:1 >In message <009F872D.74C58B92@SendSpamHere.ORG>,sC >   system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) writes: F >>Why?  Why do folks here find such an elegant and simple architecture3 >>and instruction set to be so difficult to fathom?0 >XF >It's not the instruction set itself, but the way the instructions areI >used to build a procedure.  Instead of a 16-bit entry mask that tell you G >what registers get saved and restored on RET, you have to sort out thelK >functions prologue instructions.  References to the argument list are more@I >obvious on the VAX, a reference to R16..R22 on Alpha only means a callerrL >argument if the procedure hasn't yet called another function.  It is easierJ >to comprehend a CALLS, CALLG, JSR, etc, on VAX than the multi-instruction. >equivalent on Alpha (the old 'semantic gap').  J But I still see VAX bias.  What is so difficult to comprehend in JSR? BSR?  K Finding a prologue is simple and it's easy to see which registers are being ; preserved.  You've got to decode a mask on VAX to see this.h  % FYI, it's R16 through R21, not R22.       N >I think it would be helpful if the disasemblers (debugger, SDA) had an optionI >to substitute pseudo-instructions for instructions that use R31.  It theoI >effect of an instruction is to clear a register or invert its bits, thenh >say CLEAR or COMPLEMENT.d >r >>Alpha is pushingG >>the decade mark and people still haven't figured it out?  What is it,cE >>a mental block from reading VAX instruction code for so long?  I am- >>just curious?0 >0G >I think DEC discouraged direct assembly language programming on Alpha,:E >one of the principles of RISC being that sophisticated compilers areB4 >required to make effective use of the architecture.  G "I think DEC discouraged direct assembly language programming on Alpha"s" so it could sell you compilers. ;)   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMe            sO city, n., 1. a place where trees are cut down and streets are named after them.o   ------------------------------  % Date: Sat, 03 Mar 2001 12:38:06 +0000c) From: Christof Brass <brass@infopuls.com>N< Subject: Re: Drop C/C++ better sooner than later - continued, Message-ID: <3AA0E5AE.E2C17F8B@infopuls.com>   Then Micro$oft decided to put low spec'd machines on the developer's desktops which significantly reduced the resource demand of the next WNT version.   ------------------------------  % Date: Sat, 03 Mar 2001 13:10:26 +0000w) From: Christof Brass <brass@infopuls.com> 9 Subject: How to licence boot strapping a hobbyist system?r, Message-ID: <3AA0ED42.6E356224@infopuls.com>  p I have a problem with my new hobbyist machine which came with VMS 7.2 (I assume) installed but without licences. Booting to get the VMS prompt and logging in was no problem. Then I entered the OPENVMS-ALPHA PAK which I had received from the hobbyist site. Rebooting brought up the well known X11 login prompt. I entered username/password and received a message from the LFM complaining about a missing licence. After confirming I'm put back to the X11 login prompt. I didn't find a way out so I switched the system of, re-booted and broke the startup process before it fired up the X11 login procedure. I was able to login ashSYSTEM and then decided to configure DECnet because I thought this didn't need any additional licence. And I have a commercially licenced VMS system DECnet installed around which I wanted to connect to be able to transfer the layered product licence PAKs in electronical form instead of having to put it in manually. So I fired up @NET$CONFIGURE[.COM] and I'm  now stuck at putting in the machine's domain name after chosing LOCAL from [Domain,DECdns,LOCAL]. Stuck means that the keyboard  doesn't seem to have any effect. While I was reading the explanation text the background colour changed from the well  C known blue to black and from that moment on I couldn't do anything.   D How do I proceed? What could be the reason for this process to hang?  Basically what is the intended/best procedure to put in all the hobbyist licences? I needed about five attempts for putting in the OPENVMS-ALPHA licence PAK from scratch each time as I couldn't start an editor because the graphics console (Trio video card) isn't recognised as terminal (when I tried to startup edit/tpu it complains about unknown type of sys$input) and the multiline command line editing is not a strength of the DCL CLI on VMS (on an unknown terminal especially).n   I have terminal software on various remote systems which are all connected by a TCP/IP ethernet and I would be able to copy/paste the licence PAKs from a working remote terminal session.  xAny suggestions are highly appreciated. And, yes, I checked the FAQ. BTW I read in the FAQ that commercial developers might get discounts for SW and HW from Compaq. Are there any experiences around with that? And what about open source free software developers? I think it would be fair to offer them similar conditions as they will increase the value of the platform as well.   ------------------------------  % Date: Sat, 03 Mar 2001 15:44:11 +0100m  From: Paul Sture <paul@sture.ch>= Subject: Re: How to licence boot strapping a hobbyist system? + Message-ID: <VA.000002e0.9f377c4a@sture.ch>c  * Christof Brass <brass@infopuls.com> wrote:  : > I have a problem with my new hobbyist machine which came: > with VMS 7.2 (I assume) installed but  without licences. > > > Booting to get the VMS prompt and logging in was no problem.< > Then I entered the OPENVMS-ALPHA  PAK which I had received= > from the hobbyist site. Rebooting brought up the well knowna= > X11 login  prompt. I entered username/password and receivedi> > a message from the LFM complaining about a  missing licence.  7 You need the DW-MOTIF license from the Layered Licenses  pack.   ; The easy way to suppress Motif is to set the WINDOWS_SYSTEMe SYSGEN parameter to 0.  : > After confirming I'm put back to the X11 login prompt. I5 > didn't find a way out so I  switched the system of,m< > re-booted and broke the startup process before it fired up= > the X11 login  procedure. I was able to login as SYSTEM andr9 > then decided to configure DECnet because I thought thiss3 > didn't need any additional  licence. And I have ah: > commercially licenced VMS system DECnet installed around7 > which I wanted to  connect to be able to transfer thef> > layered product licence PAKs in electronical form instead of. > having to  put it in manually. So I fired up; > @NET$CONFIGURE[.COM] and I'm  now stuck at putting in the 1 > machine's domain name after chosing LOCAL from a7 > [Domain,DECdns,LOCAL]. Stuck means that the keyboard d; > doesn't seem to have any effect. While I  was reading thep? > explanation text the background colour changed from the well a; > known blue to black and from that moment on I couldn't doB > anything.b >e= > How do I proceed? What could be the reason for this process 
 > to hang? >9  < Could well be a resource problem which an AUTOGEN would fix.  = > Basically what is the intended/best procedure to put in all : > the hobbyist licences? I needed about five  attempts for< > putting in the OPENVMS-ALPHA licence PAK from scratch each: > time as I couldn't start an  editor because the graphics> > console (Trio video card) isn't recognised as terminal (when> > I tried to  startup edit/tpu it complains about unknown type> > of sys$input) and the multiline command line editing is  not: > a strength of the DCL CLI on VMS (on an unknown terminal > especially).  3 On my system, I did a SET TERM/VT52 at the console,o8 although I believe that only works on later Alphas. That5 allowed me to use EDIT/EDT in full screen mode. Slow, 
 effective.  9 If that fails, you can use EDIT/EDT in command line mode.t Not as pretty, but it works.   > > > I have terminal software on various remote systems which are< > all connected by a TCP/IP ethernet and I  would be able to< > copy/paste the licence PAKs from a working remote terminal
 > session.   Other methods: 1  ; o - use a terminal emulation program via a serial port link:< o - grab a copy of PCX from Hunter Goatley's site to read PC     floppies: o - use another VMS machine to create a VMS format floppy.  6 When you finally get it working, you may find that the; startup which fires up Motif the first time does an Autogenb< on your behalf. Unfortunately it doesn't do it properly, and7 goes into an Autogen / reboot loop. IIRC I did a manualu: halt, minimum startup, edited larger values than suggested- into modparams.dat and Autogen with a reboot.g   > = > Any suggestions are highly appreciated. And, yes, I checkedh< > the FAQ. BTW I read in the FAQ that  commercial developers> > might get discounts for SW and HW from Compaq. Are there any; > experiences  around with that? And what about open sourcen> > free software developers? I think it would be fair to offer < > them similar conditions as they will increase the value of > the platform as well.m     ___ 
 Paul Sture Switzerland    ------------------------------    Date: 04 Mar 2001 00:58:17 +0800, From: Paul Repacholi <prep@prep.synonet.com>= Subject: Re: How to licence boot strapping a hobbyist system?l- Message-ID: <877l266b86.fsf@prep.synonet.com>t  + Christof Brass <brass@infopuls.com> writes:e  A > Any suggestions are highly appreciated. And, yes, I checked theeA > FAQ. BTW I read in the FAQ that commercial developers might get6@ > discounts for SW and HW from Compaq. Are there any experiences< > around with that? And what about open source free software< > developers? I think it would be fair to offer them similarE > conditions as they will increase the value of the platform as well.u  C CSA? Well worth it IMO. For a modest sum, you get a years unlimiteda> VMS and layered products. But best of all, you can use them onC machines that are going out to the customer while you are doing the = development. So you can drop them into a cluster, whatever...e  : There is also a sizable discount for HW, up to a $$ limit.  E The web site sucks, unless you have IE4... Well, it would suck if you5
 could get in.o  G Run, don't walk and sign up. That at least, the Q has got pretty right.o   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.y@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  % Date: Sat, 03 Mar 2001 15:06:59 +0100   From: Paul Sture <paul@sture.ch>8 Subject: Re: HP DAT drive now functional but JAVA broken+ Message-ID: <VA.000002dd.9f156c5e@sture.ch>e  J In article <ohgv9tkhm77i2u1ot4okk5pf1bo0du4b8h@4ax.com>, Alan Greig wrote:' > From: Alan Greig <a.greig@virgin.net> 6 > Newsgroups: vmsnet.sdk.openvms.fieldtest,comp.os.vms: > Subject: Re: HP DAT drive now functional but JAVA broken' > Date: Fri, 02 Mar 2001 16:05:07 +0000  > G > On 2 Mar 2001 10:27:47 -0500, koehler@encompasserve.org (Bob Koehler)  > wrote: > c > >In article <u40v9tgp0q3t03eimk19a1mr9utm0e8r2d@4ax.com>, Alan Greig <a.greig@virgin.net> writes:  > >> sE > >> Question: Java 1.2.x is broken on VMS 7.3 as it requires displayi > >> postscript. > > 9 > >What part of Java requires display Postscript?  SWING?e > >oB > >(I don't have the 7.3 release notes.)  Am I safe in assuming my > >non-graphic apps will be OK?  > G > Don't have the release notes to hand but I believe it said plain text F > apps will work fine. Also graphics apps work with 1.1.x but not withD > 1.2.x. Why on earth 1.2 needs it but not 1.1 I have no idea. I didG > attempt to run the JAVA citrx client under 1.2 which worked fine with D > VMS 7.2-1 and it died with a huge splurge of error messages. StillB > haven't been able to make the JAVA acrobat client work under anyH > combination of JRE and VMS though even if I follow the instructions to@ > the letter, The installation procedure hangs half way through. >  From the 1.2.2 release notes  L    * Some graphics operations are slower in the J2SDK v 1.2.2 than they were      in the 1.1.n JDKs.   D      With Java 2, Sun has changed the underlying architecture of theK      graphics subsystem. In JDK 1.1.n, more of the graphics operations werelL      done using native code. For example, the drawLine() method would resultL      in a call to XDrawLine. With Java 2, the rendering of graphics has beenJ      moved to Java code, which computes the image and sends the pixels outI      to the display. As a result, the performance of graphical operationssH      can be slower with J2SDK v 1.2.2, particularly when displaying to aI      remote machine. The slow performance that you see with some graphicsaL      operations is not specific to the Compaq Java 2 SDK, but is inherent in-      the architecture of Java 2 at this time.l  J      Sun has received a number of bug reports on this performance problem.K      For example, see the following list of bug reports in the Bug Database/J      at the Java Developer Connection web site: 4204845, 4185726, 4217446,A      and 4210230. You can access the Java Developer Connection atnJ      http://developer.java.sun.com/developer. Once you have logged in (youG      have to register, but it is free), just follow the link to the Bugg      Database.    .    .    .    hI    * The HotJava browser provided by Sun is not designed to work with then      Java 2 SDK.   Yuk. ___t
 Paul Sture Switzerlandh   ------------------------------  # Date: Sat, 03 Mar 2001 15:23:56 GMT 4 From: "Terry C. Shannon" <terryshannon@mediaone.net># Subject: Just Two Days Left to Vote?< Message-ID: <g08o6.7091$3F2.2074053@typhoon.ne.mediaone.net>  J Yes, the chad-free electronic poll at www.compaqworkinggroup.org closes onF March 5. If you've got a hankering for affordable VMS, a suggestion toJ improve marketing, or even a suggestion on how to ramp up iPAQ production,G visit the site and speak your piece. Many of the issues you can vote on.* contain text boxes for free-form comments.  H Compaq's response to the most popular issues will be revealed to all andI sundry early in May. A new issue submission cycle starts shortly as well.$   -- Terry C. Shannon Consultant and Publisher Shannon Knows Compaq  email: terryshannon@mediaone.net$ Web (info on SKC):  www.acersoft.com   ------------------------------  $ Date: Sat, 3 Mar 2001 12:42:53 -0500- From: "Island Computers" <sales@islandco.com>l6 Subject: Low Cost Alpha Price Drop - Now $850 Complete/ Message-ID: <ta2b65gtvum368@corp.supernews.com>r  ) http://www.islandco.com/lowcost_alpha.htm      Island Computers US Corporationd 2700 Gregory Street  Savannah GA 31404e Tel: 912 447 6622e Fax:912 201 0096 sales@islandco.com   ------------------------------  % Date: Sat, 03 Mar 2001 10:46:28 +0000l) From: Christof Brass <brass@infopuls.com> " Subject: Re: Low cost workstations, Message-ID: <3AA0CB84.C09EC6FB@infopuls.com>   Malcolm Dunnett wrote: > . > In article <3AA0560E.481DC5AF@infopuls.com>,1 >     Christof Brass <brass@infopuls.com> writes:e   [SNIP]  7 > > Have you heard of the concept of creating a market?s > M >    Sure all you have to do is sell the box cheap enough ( probably not thateJ > hard to accomplish ) and make thousands of applications available for itN > so people will have something to use it for ( next to impossible probably ). > M >    Personally I'd hate to see Compaq take off on a tangent like this beforeiM > they show they're capable of addressing all the concerns we've stated about D > the cost of the existing systems. Once they figure out how to sellK > a DS10 size system but with dual processors, 1GB of memory, 36GB disk(s), P > unlimited use VMS licenses and clustering for under $6000 then, and only then,B > should they think about low-cost VMS workstations. I didn't justF > dream up that number - Dell sells a dual processor P1000 server withK > a similar hardware config and Linux for just over $4000, so even at $6000-, > it's a stiff premium over the competition. > I >    Once you understand that Compaq has no interest in expanding the VMS:F > market except into narrowly defined high-end niches you'll see why a8 > $1000 VMS workstation will never see the light of day.  S To be able to do something and wanting to do something are two different things ...9 I understood your previous post to which I was responding as an arguing against the possibility of having a VMS workstation market.i   ------------------------------  $ Date: Sat, 3 Mar 2001 08:10:29 -0500- From: "Island Computers" <sales@islandco.com>u" Subject: Re: Low cost workstations/ Message-ID: <ta1r9teq25cvf9@corp.supernews.com>e   Geez  0 OK - How about this  - I disagree with Junk Mail   I get too much of it It kills trees) It annoys the crap out of everyone I knowtB It gets thrown in the trash before it gets to the right individual  F Basically - I have a little more respect for our environment and would/ employ a "spam" company before I did papermail.tJ (Don't worry - we don't Spam  - just post to the newsgroups - thoug some I am sure disagree)l     -- David Turner Island Computers US Corporatione 2700 Gregory Streets Savannah GA 31404r Tel: 912 447 6622  Fax:912 201 0096 sales@islandco.comB "David J. Dachtera" <djesys.nospam@earthlink.net> wrote in message' news:3AA05A73.8CC72F4B@earthlink.net..., | Island Computers wrote:R | > 	 | > Davidi | > L | > We have dealt once with a company named computer intelligence - who wereA | > also claining to be the largest  direct mail house in the USAe |c% | No, Dave, you didn't pay attention:t |oJ | There is but one Direct Marketing Association. Metromail is its database% | home. There is no question of this.v |tE | There is but one National Consumer Database. Metromail is its home.u | There is no question of this.> |t# | What else can I clear up for you?g |d | -- | David J. Dachterag | dba DJE Systems  | http://www.djesys.com/ |C< | Unofficial Affordable OpenVMS Home Page and Message Board:! | http://www.djesys.com/vms/soho/  |tH | This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings | is to be expected. |sB | Feel free to exercise your rights of free speech and expression. |aH | However, attacks against individual posters, or groups of posters, are | strongly discouraged.a   ------------------------------    Date: 03 Mar 2001 21:18:32 +0800, From: Paul Repacholi <prep@prep.synonet.com>" Subject: Re: Low cost workstations- Message-ID: <87k8676lef.fsf@prep.synonet.com>   3 nothome@spammers.are.scum (Malcolm Dunnett) writes:c  7 > > Have you heard of the concept of creating a market?s  A It is no longer about *creating* a market. It is about surviving.   M >    Sure all you have to do is sell the box cheap enough ( probably not that J > hard to accomplish ) and make thousands of applications available for itN > so people will have something to use it for ( next to impossible probably ).  C Well, consider this. The are getting all these DS10L and DS10 MB as.D part of the upgrade. So drop em in a box, add mem, reasonable 24 bit video, disk.  @ Drop em out for $925 a pop. Add $75 for CDs, VMS-unlimited, FullB Cluster, TCPIP, DECNET FF, decset, complilers. Just like the decus kit.   Real cost? $300-400 done right./  < If they don't want to 'own' them, drop em on an Island ;) or something...  C And before you screem 'used, unclean...' go look at you IBM kit andd6 find the hard to see, tucked away sticker and read it.   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.e@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------   Date: 3 Mar 2001 18:11:27 GMTs2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)" Subject: Re: Low cost workstations, Message-ID: <97rc4f$qdj@gap.cco.caltech.edu>  X In article <3AA05747.105F2CD2@infopuls.com>, Christof Brass <brass@infopuls.com> writes: >John Vottero wrote:  L Remember, this is all for SMALL systems, which I stated in the initial post.   >> >D >> >                        VMS             Solaris          WindowsD >> > Lots of software       1  no           7  yes           10  yes: >    Not necessary see my other post to Davids HW analysis  L Count up the number of available and useful Word processors, spread sheets, J databases, CAD programs, drawing programs, etc.  I'll stand by my numbers.   >oE >> > speed of software      3  low          8  high           9  highc' >                              Not true   H Depends on what they do.  If they write to files with any frequency thisI is true.  If they are CPU bound then it's a dead heat.  Weight it all and I the numbers come out about right.  I included in this the speed of moving J files over the net, which is very common since user files tend to live on  remote servers.e     >oD >> > cost of software       1  ultrahigh    5  medium         9  low: >                              Not true if you look at TCO  E Yes true, because TCO on a small VMS system is dominated by software cF development costs.  The TCO numbers Compaq waves around are for large K systems using packaged software (Oracle and the like.)  It does not fit my  B situation at all - where I have to port and handhold every single  application.   >sD >> > cost to obtain system  1  expensive    4  low-medium     8  lowD >> > cost to obtain OS      1  expensive   10  free           3  low' >                           exaggerated  NOT!N                             >$1000          $0                $100 with system   > G >> > cost to maintain*      5  medium       5  medium         6  medium7 >    sJ >> > reliability            9  high         9  high-delta     1  don't askJ >> > marketing             -3  oxymoron     7  good          10  excellentJ >> > mindshare              1  don't ask    8  high          10  max'd outB >    What do you mean by that? Meeting other people to hold hands?  J No I mean developer, manager, and customer mindshare.  I mean the ability G to get somebody to write software for your platfrom, and the ability to F "sell" a new system based on VMS to either your boss or your clients.    >eJ >> > vendor's management    1  horrific     7  good           9  excellent' >                           exaggeratedo   I wish that it were.   Regards,   David Mathog mathog@seqaxp.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech nJ **************************************************************************J *                                RIP VMS                                 *J **************************************************************************   ------------------------------    Date: 03 Mar 2001 22:45:33 +0800, From: Paul Repacholi <prep@prep.synonet.com>' Subject: Re: more "disk" IO comparisons - Message-ID: <877l267vxu.fsf@prep.synonet.com>i  4 mathog@seqaxp.bio.caltech.edu (David Mathog) writes:  E > I've run the "mybenchmark" disk/file IO speed tests on a variety oftE > systems.  While this has mostly served to illustrate just how badlynC > VMS sucks at this particular type of fundamental operation it haso0 > also turned up a few surprises.  For instance:  E >   On Tru64 these tests run at exactly the same speed on a DS10 as a E >   DS20 A 466 Mhz DS10 is as fast, or faster, than an Itanium Blazer  >   running at 666 Mhz.I  C > The DS20 supposedly has twice the memory bandwidth of a DS10, ande? > when running on Unix mostly what this test measures is memoryaD > bandwidth and file cache efficiency.  I think what this is tellingC > us is that each CPU on a DS20 has the same bandwidth as that of a E > DS10, but that both can run at full speed.  Which doesn't help here C > because the code is single threaded.  Or maybe there's some otheryB > explanation.  But I was really expecting the DS20 to be twice as > fast as the DS10.w  C The DS/ES systems use a cross point switch between CPUs, memory andoC IO. Your bench mark is point bandwidth limited by the CPU, as totalaE memory BW is greater than CPU bandwidth. If you run 2 copys on a DS20o. you will get to or closer to the memory limit.  D For the DS20 to have twice the BW of the 10, the memory must be ableF to sustain greater than twice the CPU bandwidth. Equal won't quite cut# it due to cache probes victims etc.n  * No suprise it's faster than the bubbler...   -- PB Paul Repacholi 1 Crescent Rd., +61 (08) 9257-1001 Kalamunda.  West= Australia 6076 Raw, Cooked or Well-done, it's all half baked.l   ------------------------------    Date: 03 Mar 2001 22:30:24 +0800, From: Paul Repacholi <prep@prep.synonet.com> Subject: Re: New Sun Blade- Message-ID: <87bsri7wn3.fsf@prep.synonet.com>b  3 nothome@spammers.are.scum (Malcolm Dunnett) writes:e  D >    Any way they could fit an Alpha in one of those IPaq boxes? :-)  C Yes, but... How do you hold it? and do you have a fork lift licence  for the battery? ;)B  F Notice too the sly move to embrace the gatesters crud. It is a plug inF option. So PHBs can do their wordwanking, but going to a PC, they loseC most of the Sun stuff. So there is incentive for them to go that ifw( the 'one box' bug bites their tiny mind.  @ This one could kill Compaq stone mother dead in the unix market.   -- _< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.i@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  % Date: Sat, 03 Mar 2001 15:07:00 +0100   From: Paul Sture <paul@sture.ch> Subject: Re: OpenVMS goodies+ Message-ID: <VA.000002df.9f156fc5@sture.ch>l  D In article <fjiv9to6okd1p3qfl6k7evojc64ecrbqcv@4ax.com>, Alan Greig  wrote:' > From: Alan Greig <a.greig@virgin.net>o > Newsgroups: comp.os.vmse > Subject: Re: OpenVMS goodies' > Date: Fri, 02 Mar 2001 16:30:55 +0000o > B > On Fri, 02 Mar 2001 15:58:55 +0000, Nic P Clews <nclews@csc.com> > wrote: >  > >So far I have:  > >e > >An umbrella to keep me dry. > >A shirt to wear.e > >A hat for the bald patch. > >A pen to write with.  > >A book to write in.) > >A flashing bouncing ball to play with.a  > >Posters to decorate my walls. > >e > >But no pants. > >a$ > >Then again, OpenVMS is not pants. > >h$ > >(UK joke, may not work elsewhere) > $ > Sounds like one for Dave Letterman > : So who is Dave Letterman? Or is that another UK reference? ___t
 Paul Sture SwitzerlandR   ------------------------------   Date: 3 Mar 2001 11:53 CST' From: carl@gerg.tamu.edu (Carl Perkins)m Subject: Re: OpenVMS goodies, Message-ID: <3MAR200111533820@gerg.tamu.edu>   paul@sture.ch writes...sE }In article <fjiv9to6okd1p3qfl6k7evojc64ecrbqcv@4ax.com>, Alan Greig   }wrote:6C }> On Fri, 02 Mar 2001 15:58:55 +0000, Nic P Clews <nclews@csc.com>v	 }> wrote:i }> >So far I have: }> > }> >An umbrella to keep me dry.r }> >A shirt to wear. }> >A hat for the bald patch.b }> >A pen to write with. }> >A book to write in.<* }> >A flashing bouncing ball to play with.! }> >Posters to decorate my walls.  }> > }> >But no pants.  }> >% }> >Then again, OpenVMS is not pants.0 }> >% }> >(UK joke, may not work elsewhere)0 }> 2% }> Sounds like one for Dave LettermanB }> e; }So who is Dave Letterman? Or is that another UK reference?9 }___ }Paul Sture  }Switzerland  D US late night talk show host for the last N years (where N is fairly? large). His production company is Worldwide Pants Incorporated.a  & As far as I know, he does not use VMS.   --- Carl   ------------------------------  % Date: Sat, 03 Mar 2001 15:06:57 +0100   From: Paul Sture <paul@sture.ch>, Subject: Re: SAMBA 2.0.6 seems to hate me...+ Message-ID: <VA.000002db.9f156680@sture.ch>k  N In article <5.0.2.1.0.20010302110747.0253f858@24.8.96.48>, Dan Sugalski wrote:$ > From: Dan Sugalski <dan@sidhe.org> > Newsgroups: comp.os.vmsJ. > Subject: Re: SAMBA 2.0.6 seems to hate me...' > Date: Fri, 02 Mar 2001 11:19:17 -0500p > 1 > At 03:48 PM 3/2/2001 +0000, David Mathog wrote:mK > >In article <5.0.2.1.0.20010301105553.019f6eb8@24.8.96.48>, Dan Sugalski g > ><dan@sidhe.org> writes:I > > >I've been trying to get Samba 2.0.6 up and running, with close to no.O > > >success (smbclient works, which is something), and it's beginning to annoy N > > >me. I know folks have this working so I know it can work, but it beats me > > >what's wrong. > > >iL > > >*) I changed the workgroup to my local workgroup (Though it still fails > > >with XILE)w& > > >*) I set encrypt passwords to No. > >eN > >If you do that then you have to turn off encryption on all the workstations > >which will connect. > L > We have Samba on a lot of Unix boxes around here. Encrypted passwords are  > generally turned off.. > " > > >*) I set load printers to No.D > > >*) I commented out the printers, c-itoh, and frontport sections > >s, > >These two aren't going to be the problem. > E > Didn't think so, but you never know. (Old tech support habit. It's  A > sometimes the dopey, trivial things that cause the failures...)  > Q > > >When I try to connect with SMBCLIENT on the same host with the command line:r > > >r3 > > >  smbclient //monsoon/dsugalski "-U" dsugalskiu > >oG > >Do you know that that form will connect to a Unix or WNT SMB server?  > D > Yep. I can connect successfully to both Unix and WNT SMB machines. >  > > >it throws the error:: > > >o# > > >   tree connect failed: code 0h > > >g( > > >and the smb.log file has the entry: > > >)J > > >[2001/03/01 10:07:54, 1] CMS_ROOT:[SAMBA.SOURCE.SMBD]SERVER.C;2:(643)# > > >   smbd version 2.0.6 started.S+ > > >   Copyright Andrew Tridgell 1992-1998v > >DL > >The general form of the smbclient, at least last time I used it, was like > >this: > > C > >smbclient "//machine/C$" "-U" "username%password" "-W" "MACHINE"n > M > I haven't had to do that yet with SMBCLIENT. The plain -U form prompts for yJ > a password, or I can specify it on the command line. Both ways work whn M > connecting to other (all unix or NT) systems, neither way works to connect , > to this one. > M > > >and nothing else. Attempting to connect from a Win2KSP1 machine, the win I > > >machine yells with an error "the specified network name is no longer P > > >available" and the connection fails. The above log file message is all that > > >makes it into smb.log.  > >yG > >I don't know if it should be able to connect to W2K, try WNT or W95.-I > >Microsoft seems to make a point of diddling the SMB protocols slightlyt/ > >everytime the Samba guys catch up with them.. > M > SMBCLIENT on a Unix box failed as well, as did my last stab at this with a 5M > WinNT client. (Don't have one handy this time, as it's been side-graded to c > Win2K) > P I've had various struggles in this area too. I had to load the registry key for M plain text passwords on my NT box (and reboot it too) to get in on the Alpha.S  L At one point I discovered that the Disuser flag was set on the SAMBA__GUEST O account and changed that. Unfortunately I did something else at the same time,  L which got me connected, and backtracking those changes didn't reproduce the  problem.  Q I also had a problem where the workgroup master would move to the Alpha and "get 0P stuck". Killing everything in sight and rebooting the NT box seemed the way out  of that. ___ 
 Paul Sture Switzerlandc   ------------------------------  $ Date: Sat, 3 Mar 2001 14:39:00 +0100" From: "Hans Vlems" <hvlems@iae.nl>( Subject: Re: Satellite MV2000 won't boot( Message-ID: <97qs1k$6fb$1@news.IAEhv.nl>   What boot command did you use?
 >>> b esa0 ?  No qualifiers? Hans  & Randy Park heeft geschreven in bericht) <97p5ha$74s$1@slb6.atl.mindspring.net>... . >I can boot the local disk and SET HOST to the, >node I want to boot from, so the networking% >hardware appears to be working fine.o >h, >Hans Vlems <hvlems@iae.nl> wrote in message# >news:97oseo$3b6$1@news.IAEhv.nl...(I >> A wild guess: the ethernet switch on the back is in the wrong position # >> (so it uses AUI instead of BNC)?  >> >> Hansw >>) >> Randy Park heeft geschreven in berichtb, >> <97oqfd$lpm$1@slb7.atl.mindspring.net>...3 >> >I have an old MicroVax 2000 that I fire up fromo2 >> >time to time to do some 'old version' testing.2 >> >I usually boot it's own internal disk and that. >> >works fine.  Today I tried to boot it as a6 >> >satellite in a cluster, which has always succeeded6 >> >until today. I typically do a satellite boot about4 >> >once a year.  Today it failed with the following >> >error message: >> >! >> >    %VMB-F-ERR, PC = 00000765e! >> >    %VMS-I-STS, R5 = 00000056I >> > >> >    ?06 HLT INST >> >        PC = 00000E64f >> >7 >> >It's been several years since I've done any cluster-5 >> >management and the docs are not readily availablel2 >> >(other than on the web).  The TEST commands on2 >> >the MV2000 show everything is working ok.  The2 >> >boot node is running VMS 6.1. (I know it's old8 >> >but as a developer, I have my reasons.)  I've looked5 >> >at the SYSGEN parameters, /SCS and /CLU, and theyt >> >look ok. >> >0 >> >Can someone point my in the right direction? >> > >> >Thanks in advance. >> > >> > >> >> >n >t   ------------------------------  % Date: Sat, 03 Mar 2001 10:56:58 -0800-- From: Jack Patteeuw <jjpatteeuw@peoplepc.com> L Subject: Re: Sun Blade 100 with a VMS monitor ? (was: Low cost workstations), Message-ID: <3AA13E7A.5974DC60@peoplepc.com>   Larry Kilgallen wrote:  ? > Since I am already swimming in VMS monitors (they work, but I.? > can only sit in front of a limited number at a time), I wouldr; > prefer to use the Sun Blade 100 with an existing monitor.  > K > Does anyone have experience using a Sun workstation with a VMS-compatiblee > monitor ?a  F "VMS Compatible" monitor ?  I didn't know VMS cared specifically about7 the monitor specifications (sorry for the dig, Larry !)a    E Digital used a variety of monitors over the past 10+ years on VMS andnH "Unix" workstations (check http://www.monitorworld.com/Monitors/dec/ forG specifications).  For a long time all DEC monitors were "sync on green"aA with a fixed Vertical frequency of 60, 66 or 72 Hz (If I remember A correctly the VRT1x-HA would switch between 66 and 72 Hz).  LaterPC monitors (such as the beautiful VRC21), I believe, would do sync ons0 green or SVGA (RGB + Horizontal and Video sync).  A As long as Solaris can tell the video board to operate at a fixedoB frequency and you are willing to do a little soldering, your in !!     Searching the web, you'll find  3 http://www.bithose.com/serfaq/REPAIR/F_SyncGrn.html-D http://www.geocities.com/SiliconValley/Foothills/4467/fixedsync.html ^oB +-- this one even has notes on how to convert a VRT19 to RGB+HV !!     This one requires no power !!t  7 http://www.hut.fi/Misc/Electronics/circuits/sync_r.htmls     Cheap video cables   http://www.cableclub.com/      Who says "home brew" is dead !   Jack   ------------------------------  % Date: Sat, 03 Mar 2001 15:06:59 +0100t  From: Paul Sture <paul@sture.ch>  Subject: Re: Umbrella in Germany+ Message-ID: <VA.000002de.9f156ec1@sture.ch>   B In article <3a9ff899.12814586@News.CIS.DFN.DE>, Stefan Bill wrote:& > From: bill@freesurf.ch (Stefan Bill) > Newsgroups: comp.os.vmsL" > Subject: Re: Umbrella in Germany% > Date: Fri, 02 Mar 2001 19:46:35 GMTa > D > On Fri, 02 Mar 2001 15:59:11 +0100, Michael Unger <unger@decus.de> > wrote: > E > >I got my OpenVMS umbrella yesterday (Thursday, 1-Mar-2001) -- sents  > >from Zurich (Sitzerland) too. > = > From Zrich? I live in Zrich. Where can I get my umbrella?; >rC I work in Zuerich. Gimme the address and I'll save them postage by   picking one up myself :-)h  ? > >Many thanks to Sue Skonetski and the OpenVMS promotion team!i > > > Promotion Team?  Never heard. Please send an umbrella to me. >  > Stefan >    ___K
 Paul Sture Switzerland    ------------------------------  % Date: Sat, 03 Mar 2001 12:41:39 +0000e) From: Christof Brass <brass@infopuls.com>-@ Subject: Re: VMS V7.2+ bug - directory modification date changes, Message-ID: <3AA0E683.83ACEEE7@infopuls.com>   John Santos wrote: > % > On Sat, 3 Mar 2001, Ed Wilts wrote:d >  > > John Santos wrote: > >DE > > > I have discovered that in VMS (7.2-1 on Alpha, 7.2 on VAX), thetJ > > > modification date on a directory file changes whenever the directoryG > > > is extended or truncated by creating or deleting files in it.  NoiH > > > explicit change to the directory, just create or delete some files > > > in it. > > L > > It's worse than this.  I've got frequent directory modifications even ifJ > > the directory file size is not changing, and this is on AXP V7.1-2.  IM > > reported this here, and to sum up the response I got from Hoff, "life's aoM > > bitch and then you die, and not only that, but the problem will get worse N > > with the new government work that's being done".  That's obviously a grossM > > simplification of what he said, but it sums it up pretty well (and you're N > > welcome to correct me Hoff - I know you're watching :-)).  It's not a bug, > > and it won't be fixed. > >sO > > The solution is to use /noincremental. For those of you using Tapesys, it'ssM > > time to get the patch released this week, since /noincremental before the J > > patch was ignored (I knocked multiple hours off my backup times when I6 > > applied the Tapesys patch for the bug I reported). > >h > >         .../Ed > L > Okay, so then the modify date on .dir files is now absolutely meaningless?K > How about throwing it away then, and save the overhead of maintaining it?d > 	 > ------- ' > 2 years later (post VMS V7.4 release)y > H > Ooh!  Ooh!  I have an idea!  Since there is now a vacant 64-byte field= > in the file header of a .dir file, how about using it for a H > "user-modification date"?  This date would get updated whenever a userI > modifies a .dir file (rename, change protection code or ACL or owner or E > max versions, etc.) but not when the executive (F11BXQP) creates or I > deletes files in the directory or expands or compresses it.  This would  > be a *useful* date.  >  > --
 > John Santos> > Evans Griffiths & Hart, Inc. > 781-861-0670 ext 539  [ I can't believe this stupidity! Is there at least a global switch to get the old behaviour?    ------------------------------  % Date: Sat, 03 Mar 2001 15:06:58 +0100m  From: Paul Sture <paul@sture.ch>3 Subject: Re: [INFO] wrong answer in Mgmt agents FAQe+ Message-ID: <VA.000002dc.9f1569f1@sture.ch>   F In article <3AA064A7.85FB2F8B@earthlink.net>, David J. Dachtera wrote:9 > From: "David J. Dachtera" <djesys.nospam@earthlink.net>h > Newsgroups: comp.os.vms 5 > Subject: Re: [INFO] wrong answer in Mgmt agents FAQ,' > Date: Fri, 02 Mar 2001 21:27:35 -0600s >  > Didier Morandi wrote:* > >  > > The text in page:*M > > www.openvms.compaq.com/openvms/products/mgmt_agents/user_man.html#trouble  > > 8 > > gives an incorrect answer to the following question: > >  > > [start of quote] > > Q:? > > I am getting a PCSI error part-way through my installation:3 > > ; > > The following product will be installed to destination:mG > > COMPAQ AXPVMS V72_MGMTAGENTS V2.0-16 DISK$ALPHASYS-72:[VMS$COMMON.]i, > > Portion done: 0%...10%...20%...30%...40%" > > %PCSI-E-READERR, error readingK > > $6$DKA300:[SYS0.][SYSUPD]COMPAQ-AXPVMS-V72_MGMTAGENTS-V0200-16-1.PCSI;1m' > > -DDIS-E-TNF, invalid element syntaxn& > > %PCSI-E-OPFAILED, operation failedH > > Terminating is strongly recommended. Do you want to terminate? [YES]= > > %PCSI-E-CANCEL_WIP, termination resulted in an incompleted > > modification to the system3 > > %PCSI-E-S_OPCAN, operation cancelled by requesteG > > %PCSIUI-E-ABORT, operation terminated due to an unrecoverable errory
 > > condition  > >  > > A:J > > You probably have a corrupt kit. The PCSI kit must be in fixed-length,P > > 8192-byte CR format. Please verify the file attributes and copy a new kit to > > your target machine. > > [end of quote] > > ! > > To me, the correct answer is:e > > A > > NOTE:  An OpenVMS saveset or PCSI installation file is storedm8 > > on the Internet in a self-expanding compressed file. > > < > > For OpenVMS savesets, the name of the compressed saveset9 > > file will be kit_name.a-dcx_vaxexe for OpenVMS VAX oro= > > kit_name.a-dcx_axpexe for OpenVMS Alpha. Once the OpenVMS ; > > saveset is copied to your system, expand the compressed C > > saveset by typing RUN kitname.dcx_vaxexe or kitname.dcx_alpexe.o > > @ > > For PCSI files, once the PCSI file is copied to your system,@ > > rename the PCSI file to kitname-dcx_axpexe.pcsi, then it canE > > be expanded by typing RUN kitname-dcx_axpexe.pcsi.  The resultanthH > > file will be the PCSI installation file which can be used to install > > the ECO. > + > Here again, the correct syntax would be: l > @ > > rename the PCSI file to kitname.pcsi-dcx_axpexe, then it can7 > > be expanded by typing RUN kitname.pcsi-dcx_axpexe. o > A > Note that both Netscape and IE will truncate the extension fromtK > ..PCSI-DCX_%%%EXE to just .PCSI. Don't know if it's a browser bug or YAWBy > (Yet Another Windows Bug). >fI I've had similar when downloading VMSINSTAL kits using Windows Netscape.  4 eco_name.A-DCX_%%%EXE comes down as file eco_name.A.  Q This has more than once led me to try using VMSINSTAL direct on that file, which tM usually results in error messages such as BACKUP-E-READERRS, excessive error   rate reading <filename>. Bah!t  / IIRC, no problem with Netscape on VMS or Linux.h ___e
 Paul Sture Switzerlandm   ------------------------------   End of INFO-VAX 2001.124 ************************