1 INFO-VAX	Fri, 22 Oct 2004	Volume 2004 : Issue 587       Contents:% Re: Color Laser printers with OpenVMS ! Re: DEC-C: Null string constant ? ! Re: DEC-C: Null string constant ? ! Re: DEC-C: Null string constant ? ! Re: DEC-C: Null string constant ? ! Re: DEC-C: Null string constant ?  EMC - multi database locationsA Good string processing languages (was Question about Descriptors) 6 Re: Hello from the OpenVMS Technical Update in Austria* Re: HP announces new educational programme Re: J F on why USA is a regime Re: JAR files must be stream-lf  Re: JAR files must be stream-lf  Re: JAR files must be stream-lf 0 Re: Java JAR run across platforms, Zip exception, Re: memory performance Alpha, Itanic and SUN Re: OT: Short Article about Sun  Re: OT: Short Article about Sun  Re: OT: Short Article about Sun  Re: OT: Short Article about Sun  Re: pthreads and scheduling  Public DS10L ? Re: Question about Descriptors Re: Question about Descriptors Re: Question about Descriptors Re: Question about Descriptors Re: Question about Descriptors Re: Question about Descriptors Re: Question about Descriptors Re: Question about Descriptors Re: Question about Descriptors Re: Question about Descriptors Re: Question about Descriptors. Re: Retain Trust 'warranty' period almost over sales@careerpartner.co.uk " Re: SYSUAF => export passwords ???" Re: SYSUAF => export passwords ???. VAX 6000-610 For Sale/Trade in Seattle WA Area Re: What happened with SEVMS?  Re: What happened with SEVMS?  ZIP & work area? ZIP & work area? Re: ZIP & work area? Re: ZIP & work area? Re: ZIP & work area?  F ----------------------------------------------------------------------  % Date: Fri, 22 Oct 2004 02:31:17 -0400 ( From: David Froble <davef@tsoft-inc.com>. Subject: Re: Color Laser printers with OpenVMS, Message-ID: <4178A935.7050609@tsoft-inc.com>   John Smith wrote:    > Paul Anderson wrote: > E >>The HP web pages don't list OpenVMS as a supported operating system + >>for printers even though it obviously is.  >> >  >  > Paul,  >  > Nothing personal.... > N > Sometimes it's little details like this that drive us nuts. How difficult isM > it for somebody at HP to get the people who maintain a web page to say that M > VMS is a supported o/s for a product? It costs nothing (figuratively) to do N > so at the time the pages are created - after all what's the incremental cost8 > of typing 7 more characters (OpenVMS) onto a web page. > F > It goes to first impressions when prospective new customers look forK > solutions; it pops-up as another 'hit' when doing as search; it's another 0 > identifiable product to sell that's supported. >  >  >   N You're overlooking the hugh educational costs in informing the printer people Q that there is a product called VMS, that it's an HP product, and that it can use  	 printers.   Q One side effect will be the response of the printer people when someone tries to  L tell them that HP doesn't inovate and 'invent' anymore.  They'll go running G around saying "Oh, No, you're wrong.  We just got this 'brand new' OS."    Dave   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Fri, 22 Oct 2004 02:55:03 -0400 ( From: David Froble <davef@tsoft-inc.com>* Subject: Re: DEC-C: Null string constant ?, Message-ID: <4178AEC7.7090900@tsoft-inc.com>   briggs@encompasserve.org wrote:   W > In article <4177CBED.40109@tsoft-inc.com>, David Froble <davef@tsoft-inc.com> writes:  >  >>JF Mezei wrote:  >> >> >>>char *mybuffer ;  >>>mybuffer = "" >>>myroutine(mybuffer);  >>>  >>>----  >>>myroutine(char *mybuffer) >>>{1 >>>printf ("Address of string=>%ld\n", mybuffer);  >>>} >>> Q >>>Retults in a big fat value of 0, and of course, if you try to strlen(mybuffer)  >>>the program crashes.  >>> P >>>I was under the impression that "" yielded a valid string constant containingP >>>only the null character, and that you could use this to still provide a valid@ >>>pointer address to the various routines in your applications. >>>  >>7 >>First note that my prefered language is BASIC, not C.  >> >>A null is a valid character. >> > E > That statement is a tad strong.  A null _can be_ a valid character.     Q To be a bit more precise, a null character is a valid character.  ASCII value is   zero.     A > Whether it is valid depends on context.  It's a valid character  > in the ASCII code. > F > But, if your context is a null-terminated string, then a null is not3 > a valid character within the body of that string.  >  > 3 >>It takes one byte of storage for each character.  R >>It is a bit tough to have a string of nulls when using null terminated strings.  >> > G > Of course.  As it happens, I find null-terminated strings distasteful ! > as well.  For this very reason.  > L > I have developed a distaste for VMS "can't be longer than 2^16 characters"D > string descriptors as well.  It is a bit tough to encode a 100,000@ > character string when strings are limited to 65535 characters.    ? I confess to curosity.  What's the usage of such a long string?     @ > Ah well.  Such is life.  I don't much care for ASN.1 unlimitedG > length encodings either.  Guess there's no pleasing some people.  :-)  >  > * >>  Another shortcoming of that technique. >>< >>A null string is a string with no characters, zero length. >> > > > True.  And a null string encoded as a null-terminated string1 > is a buffer containing a single null character.  > J > A null string encoded using a VMS descriptor is a quadword with a lengthG > code of zero and pointer that doesn't have to point anywhere special.     O Well, you describe the descriptor, not the data.  I've not looked, but I think  O there might be a problem with a pointer to some location that will generate an  $ error, such as zero.  Just guessing.    B > A null string encoded as a counted string is a buffer containing > a single zero byte.  > ? > A null string encoded with ASN.1 is... I'd have to go look it @ > up.  *** sound of furious keyboard tapping *** ...  A two byte > buffer containing  > - >  0x4   (Universal, Primitive, Octet string) ! >  0x0   (Short format, length 0) " >        (String contents -- null) >  > 	John Briggs >        --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Fri, 22 Oct 2004 10:20:01 -0400 # From: sol gongola <sol@adldata.com> * Subject: Re: DEC-C: Null string constant ?+ Message-ID: <41791711.B1C750ED@adldata.com>    > >   > > A null is a valid character. >   . Valid Character? What is an invalid character?2 Valid/invalid character depend on the application.  @ In a name field, valid characters might be A-Z, maybe restricted/ to upper case/lower case, and some punctuation. J In a social security number (u.s.) valid characters are 0-9 and maybe "-".H Even these depend on whether you are using ASCII, EBCDIC, BCD, whatever.  E Any 8 bit configuration, including x"00", that fits in an 8 bit byte   is a valid character.   C So what if C uses a zero byte as a terminator. The reality is that  F this can only apply to alphanumeric character strings. For many types B of programs including data compression, encryption, communication B protocols, systems programming, the zero byte is part of the data , and the null termination concept is useless.  E > That statement is a tad strong.  A null _can be_ a valid character.  >  ... ... ...    sol    ------------------------------  % Date: Fri, 22 Oct 2004 10:51:22 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> * Subject: Re: DEC-C: Null string constant ?, Message-ID: <41791E67.1D9F9BAA@teksavvy.com>   David Froble wrote: R > To be a bit more precise, a null character is a valid character.  ASCII value is > zero.   N In C conventioanal strings, the null is not a valid character in the middle of4 the string since it specifies the end of the string.    N > > I have developed a distaste for VMS "can't be longer than 2^16 characters"F > > string descriptors as well.  It is a bit tough to encode a 100,000B > > character string when strings are limited to 65535 characters. > A > I confess to curosity.  What's the usage of such a long string?   K When I had read the paragraph, my brain automatically added the implied :-)  :-) :-)   M In fairness though, consider a very long document (50 pages at 2k per page is H about 100k characters). If you display it on decwidnows in a simple TextN widget, it will be manipulated as a single very long string. (however, you getN back a null terminated string, and then the descriptor's limitation only applyG if yo need to feed that oen very long string to some RTL/SYS$ routines.   M (Granted, when scaling to this level, one generally won't use the simple Text L widget and use a more complex one where you progressively supply text as theK user crolls so that the entire document isn't present at any point in time.    ------------------------------  % Date: Fri, 22 Oct 2004 16:17:37 +0100 - From: Roy Omond <Roy.Omond@BlueBubble.UK.Com> * Subject: Re: DEC-C: Null string constant ?+ Message-ID: <2tsmo8F242rnaU1@uni-berlin.de>    David Froble wrote:    > A > I confess to curosity.  What's the usage of such a long string?   D Sure would have been nice to have had such long strings available in@ the days when we were developing DNA-sequence handling software,' almost all of which started out on VMS.    ------------------------------    Date: 22 Oct 2004 11:13:16 -0500 From: briggs@encompasserve.org* Subject: Re: DEC-C: Null string constant ?3 Message-ID: <DExwPUbil2R7@eisner.encompasserve.org>   \ In article <41791E67.1D9F9BAA@teksavvy.com>, JF Mezei <jfmezei.spamnot@teksavvy.com> writes: > David Froble wrote: S >> To be a bit more precise, a null character is a valid character.  ASCII value is  >> zero. > P > In C conventioanal strings, the null is not a valid character in the middle of6 > the string since it specifies the end of the string. >  > O >> > I have developed a distaste for VMS "can't be longer than 2^16 characters" G >> > string descriptors as well.  It is a bit tough to encode a 100,000 C >> > character string when strings are limited to 65535 characters.  >>  B >> I confess to curosity.  What's the usage of such a long string? > M > When I had read the paragraph, my brain automatically added the implied :-) 	 > :-) :-)   2 You may have inferred a :-), but none was implied.  B I am perfectly serious that 65,535 is a limit that bothers me.  ItE hasn't managed to bite me in any applications yet.  But that's partly F because I know enough not to write anything that bends that particular limitation.   C The limit is large enough that you can _almost_ ignore it.  But not * large enough so that you can do so safely.  O > In fairness though, consider a very long document (50 pages at 2k per page is J > about 100k characters). If you display it on decwidnows in a simple TextP > widget, it will be manipulated as a single very long string. (however, you getP > back a null terminated string, and then the descriptor's limitation only applyI > if yo need to feed that oen very long string to some RTL/SYS$ routines.   C Yup.  There are plenty of applications where you are dealing with a A blob of data and an unbounded (or at least extravagently-bounded) B octet-string would be a nice data structure to use and an ordinary= string descriptor would be a nice way to try to reference it.    	John Briggs   ------------------------------    Date: 22 Oct 2004 09:44:55 -0700$ From: panjo03@hotmail.com (John Pan)' Subject: EMC - multi database locations = Message-ID: <dc85ec7c.0410220844.102590c9@posting.google.com>    Hi    = the site I am on is configured running OpenVMS 7.3-1 on GS160 @ clustered with other Alphas - mixture of 8400 and 4100 and ES40. The disk is serviced by EMC.  9 We are experiencing with our database an 'odd' situation.   B When a large (semi) table is spread over multiple locations on theF same physical disk, access seems to be quick. When the table is spreadB over two physical disks (as happened this morning) we get multipleA locking and performance problems. This got solved when we put the # table back in its correct location.   A This is not how I have experienced other systems before. Normally ? spreading the table over multiple locations over multiple disks B spreads io and generates better access - however not in this case.  ; Do I have a real problem, if so where in the EMC setup, lun  distribution? or what.  C Would appreciate any help/pointers in this area please as I have no  experience with EMC.   thanks   ------------------------------  % Date: Fri, 22 Oct 2004 16:13:43 +0100 - From: Roy Omond <Roy.Omond@BlueBubble.UK.Com> J Subject: Good string processing languages (was Question about Descriptors)+ Message-ID: <2tsmgtF23suc2U1@uni-berlin.de>    Steve Lionel wrote:   O > On Fri, 22 Oct 2004 00:24:50 -0400, David Froble <davef@tsoft-inc.com> wrote:  > ? >>You forgot BASIC, which may be the best at string processing.  > P > BASIC?  PL/I?  MUMPS?  Pikers.  For string processing, no language I have ever > seen beats SNOBOL4.   @ Well, my last brief exposure to Snobol was, IIRC, ca. 1971/1972,7 so maybe I'm not best qualified to make comparisons ...   0 But I'd say TPU must surely run it pretty close.   And how's about (VAX) SCAN ?  F And then there was IMP from Edinburgh University, in which was written% that Very Fine Operating System EMAS.    ------------------------------   Date: 22 Oct 2004 06:14:50 GMT2 From: "Dave Weatherall" <djw-nothere@nospam.nohow>? Subject: Re: Hello from the OpenVMS Technical Update in Austria ? Message-ID: <DTiotGxQ0bj6-pn2-21Wdd3uN633z@dave2_os2.home.ours>   C On Mon, 18 Oct 2004 12:44:47 UTC, susan_skonetski@hotmail.com (Sue   Skonetski) wrote:    > Dear Newsgroup,  > G > Hello, from Vienna Austria.  This is our fourth stop on the Technical G > update tour.  We have had an excellent time.  We started in Stockholm G > Sweden.  We were in the HP office in Kista.  The event sold out there B > and a few people ended up sitting on the floor, we then moved toH > Holland, Utrect (sp?)we had one spare seat on the first day.  Then offH > to London and now Austria.  Its funny no matter where I go I am alwaysF > touched at how kind the VMS people are.  The audience is always veryC > open to what we have to say about where are going and what we are < > doing now.  Hopefully people will post their own thoughts. > C > We are off to the last technical update day in Germany on Friday.  >  > Warm Regards,  > Sue   F Hope you have a good time. You've picked a good time weatherwise. Pity I'm too busy to drive up.    --   Cheers - Dave W.   ------------------------------  % Date: Fri, 22 Oct 2004 10:36:06 -0700 ' From: David Mathog <mathog@caltech.edu> 3 Subject: Re: HP announces new educational programme + Message-ID: <clbge6$a56$1@naig.caltech.edu>    John Smith wrote:     N > Higher education institutions using HP servers and workstations for teachingL > or research can now receive the HP-UX 11i operating system at no charge(1): > through the HP Campus Investment Program Academic Offer.  D After, of course, paying for the hardware.  What MORON would buy an : HP-UX for research these days?  Opteron or Pentium systems= have far more bang for the buck and linux works just fine for  research, thank you very much.     > H > The program also provides substantial discounts on HP-UX 11i operatingD > environments, software, licenses and support services for faculty,E > researchers and students to meet the general enterprise information / > technology needs of educational institutions.   F This may be the true purpose - an attempt to try to hold onto existing! enterprise level computing sites.   * > This program is designed to help supportA > the efforts of campuses to transform into 21st century learning  > environments."  E Ah yes, because PA-RISC has such a bright future for the remainder of 
 this century.    > @ > Colleges and universities using HP-UX 11i environments to meetM > communications and IT infrastructure needs - including e-mail, registration M > and financial processing - are eligible for up to 45 percent savings on all K > HP-UX 11i operating environments, selected layered products and HP-UX 11i K > clustering software, called HP Serviceguard. Educational institutions can K > use the HP Campus Investment Program discount to help build innovative IT F > systems to improve academic performance while also practicing fiscal > responsibility.   @ Again, buy some more hardware and we'll give you a break on the F software.  Versus, buy a bunch of Opterons (faster, for less) and run  Linux (for nothing.)     > L > "The HP Campus Investment Program Academic Offer gives our institution theD > ability to adapt and capitalize on future technologies quickly andN > cost-effectively," said Doug Eltoft, director of operations computer systemsH > and support, College of Engineering, University of Iowa. "This programM > enables our students, faculty and administrators to have access to the most H > advanced technology available to help support the University of Iowa's > academic mission."  ! And it will cure cancer!  Really!    > C > All Campus Investment Program software packages allow educational L > institutions one-time purchases of software and perpetual licenses that doA > not require yearly renewal, thereby reducing maintenance costs.   E Ok, that much is good.  The yearly renewals can be a problem in years  when the grants don't come in.  J > Additionally, education customers who have a system registered under theL > program may place their HP-UX system under an existing support contract orB > can take advantage of special discounts on HP Care Pack support.  D Translation, we know service costs are driving customers to cheaper G solutions.  So we'll offer a price cut in this guise and hope to retain  some of these customers.     > J > (1) HP reserves the right to remove software products from the HP CampusM > Investment Program Academic Offer at any time due to royalty obligations or L > other reasons, and provides no assurance that any product will continue to > remain in the program.   Run screaming for the exits.  I It's like buying a car from Ford and having them say, "Ford reserves the  8 right to remove your motor and/or brakes at any time..."  ; Those who do not remember the CSLG are doomed to repeat it.      Regards,   David Mathog mathog@caltech.edu   ------------------------------  # Date: Fri, 22 Oct 2004 10:31:26 GMT  From: reb <natron@ntlworld.com> ' Subject: Re: J F on why USA is a regime 0 Message-ID: <opsf9os5b8aqvj2s@news.ntlworld.com>  K On Thu, 21 Oct 2004 15:49:16 +0000 (UTC), <david20@alpha2.mdx.ac.uk> wrote:  > G > There was an interesting program on BBC 2 last night - The Power of   
 > Nightmares. t > http://www.bbc.co.uk/bbctwo/listings/programme.shtml?service_id=4224&FILENAME=20041027/20041027_2100_4224_43282_60 > I > The series explores the idea that the threat of a terror network is a    > myth.  > F > On the neo-conservative side it was interesting to see how all the  H > arguments currently being used were also used against the Soviet Union > in the 1980s. H > Hyping up of the threat by inventing or misinterpretting intelligence.D > Claims that All Terrorist groups were part of a global terrorist  1 > conspiracy being organised by the Soviet Union.     9 The bit I hadn't realised was that it was the same people 7 - Cheney, Rumfeld, Paul W - that lied to Reagan who are 7 now back in power with Bush. Some of their old speeches 1 sounded very familiar, just with different names.     8 plus ca change, as they probably don't say in Washington     Roger B.   ------------------------------  # Date: Fri, 22 Oct 2004 07:43:17 GMT + From: "Daniel Gustafsson" <daniel@mimer.se> ( Subject: Re: JAR files must be stream-lf5 Message-ID: <pS2ed.107005$dP1.402581@newsc.telia.net>   A "Carl Karcher" <karcher@thuria.waisman.wisc.edu> wrote in message 0 news:21OCT04.12354780@thuria.waisman.wisc.edu... > Bob Koehler wrote:F > ->    jar files on VMS must be stream-lf format.  Your FTP tool willB > ->    not do that.  So first transfer it in binary, and then use > ->( > ->       $set file/attribute=rfm=stmlf > K > You can also convert the existing files if you don't want to re-transfer:  > , >  $ CONVERT/FDL=SYS$INPUT file.jar file.jar	 >  RECORD  >  FORMAT STREAM_LF  >  <CTRL-Z>   D Isn't that a destructive method in this case? CONVERT may change theJ contents of the file, while SET FILE /ATTR does not touch the contents and& only change the meta-data of the file.   -- Daniel Gustafsson  http://developer.mimer.se    ------------------------------    Date: 22 Oct 2004 07:46:36 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) ( Subject: Re: JAR files must be stream-lf3 Message-ID: <hDF7rpzyziwS@eisner.encompasserve.org>   m In article <21OCT04.12354780@thuria.waisman.wisc.edu>, karcher@thuria.waisman.wisc.edu (Carl Karcher) writes:  > Bob Koehler wrote:F > ->    jar files on VMS must be stream-lf format.  Your FTP tool willB > ->    not do that.  So first transfer it in binary, and then use > ->  ( > ->       $set file/attribute=rfm=stmlf > K > You can also convert the existing files if you don't want to re-transfer:  > , >  $ CONVERT/FDL=SYS$INPUT file.jar file.jar	 >  RECORD  >  FORMAT STREAM_LF  >  <CTRL-Z>   F    Convert does not do the same thing as set file/attribute.  Convert H    actually can change the contents of the files.  Set file only changesK    which bytes RMS will consider metadata.  In general convert and set file C    are not alternative ways to do the same thiung, although in some     cases both may work.    ------------------------------   Date: 22 OCT 2004 13:26:54 GMT4 From: karcher@thuria.waisman.wisc.edu (Carl Karcher)( Subject: Re: JAR files must be stream-lf6 Message-ID: <22OCT04.13265497@thuria.waisman.wisc.edu>  1 In a previous article, "Daniel Gustafsson" wrote:   M ->> You can also convert the existing files if you don't want to re-transfer:  ->> . ->>  $ CONVERT/FDL=SYS$INPUT file.jar file.jar ->>  RECORD  ->>  FORMAT STREAM_LF 
 ->>  <CTRL-Z>  ->  F ->Isn't that a destructive method in this case? CONVERT may change theL ->contents of the file, while SET FILE /ATTR does not touch the contents and( ->only change the meta-data of the file.  > Well, yes and no. Yes, it changes the contents of the file but3 no, it doesn't change the contents of the records.    H The point being you need to convert the <CR><LF> at the end of each line@ (how text files are delimited on Windows/Dos) to just a <LF> (asH expected by unix and friends). Changing only file attributes in the fileD header requires you to re-transfer the files in binary mode. ConvertE "fixes" the broken files already transferred in ascii mode. Actually, H you can do a DIFF on file.jar after the convert and the records it reads will be identical.   --G -- Carl Karcher, Waisman Computing Services, Waisman Center, UW-Madison 8 --                 karcher.nomorespzm@waisman.wisc.edu     ------------------------------  # Date: Fri, 22 Oct 2004 07:51:44 GMT + From: "Daniel Gustafsson" <daniel@mimer.se> 9 Subject: Re: Java JAR run across platforms, Zip exception 5 Message-ID: <k_2ed.107006$dP1.402616@newsc.telia.net>   ? "David J Dachtera" <djesys.nospam@comcast.net> wrote in message % news:41769FAA.4D3D9B9F@comcast.net...  > Daniel Gustafsson wrote: > > 2 > > "Soterro" <soterroatyahoocom> wrote in message9 > > news:41766e8c$0$27334$4d4ef98e@read.news.ch.uu.net...  > > > Hello, > > > I > > > I'm trying to get a JAR file working on a VMS and a Windows machine  > > > (Unix shall come later) J > > > So I go and create the JAR file on the OpenVMS machine (JRE 1.4.1-2, VMS L > > > 7.3-2), everything works fine. I go and create the JAR file on WindowsI > > > (W2000, Eclipse 3.0.1 does that, JRE 1.4.2-03), then when I copy it  over' > > > to the VMS and I run the command:  > > >  > > > $ java -jar mything.jar  > > >  > > > I get: > > >  > > > JavaVM args:K > > >      version 0x00010002, ignoreUnrecognized is JNI_FALSE, nOptions is  4 - > > >      option[ 0] = '-Djava.class.path=.' 7 > > >      option[ 1] = '-Djava.class.path=mything.jar' 8 > > >      option[ 2] = '-Dsun.java.command=mything.jar'& > > > 1 micro seconds to InitializeJVM > > > Main-Class is '' > > > Apps' argc is 0 E > > > Exception in thread "main" java.util.zip.ZipException: error in  opening  > > > zip file; > > >          at java.util.zip.ZipFile.open(Native Method) < > > >          at java.util.zip.ZipFile.<init>(ZipFile.java)< > > >          at java.util.jar.JarFile.<init>(JarFile.java)< > > >          at java.util.jar.JarFile.<init>(JarFile.java) > > > L > > > The file is not corrupted, JAR is otherwise happy with it, it can listK > > > and extract the classes and I can create then a fully working jar out  ofJ > > > the unpacked classes. It's just I can't run on VMS the version I getG > > > from the Win machine. The Main-Class shows nothing, whatever that  means G > > > (although Manifest is there). The Win jar works fine on other Win J > > > machines. I don't know anymore where to look, although it looks like3 > > > being something very basic (just an unzip...)  > > >  > > > So I'm lost... Help? > > " > > What does DIR/FULL x.JAR say ? > >  > > The file must be Stream_LF.  > > 0 > > Run this after you transfer the file to VMS. > 1 > Remember to transfer the .JAR file as *BINARY*!   J If the file had been transferred with FTP ASCII there would have been muchK worse errors and the JAR file would have been totally unusuable. ;) In this  case there was a subtle error.  + > > SET FILE x.JAR /ATTR=(RFM:STMLF,RAT:CR)  > " > RAT is irrelevant, but harmless.   OK.    -- Daniel Gustafsson  http://developer.mimer.se    ------------------------------    Date: 22 Oct 2004 04:35:54 -0700' From: icerq4a@spray.se (David Svensson) 5 Subject: Re: memory performance Alpha, Itanic and SUN = Message-ID: <734da31c.0410220335.601e2678@posting.google.com>   q Andrew Harrison <andrew_remove__harrison@sun__.com> wrote in message news:<ckjfjv$hub$1@new-usenet.uk.sun.com>...  > David Svensson wrote: u > > Andrew Harrison <andrew_remove__harrison@sun__.com> wrote in message news:<ckebpv$qoa$1@new-usenet.uk.sun.com>...  > >  > >>Dirk Munk wrote: > >>) > >>>Look at this little table I found at  > >>> ; > >>>http://www.cs.virginia.edu/stream/top20/Bandwidth.html  > >>>  > >>= > >>The clue to why your contention may not be as safe as you 4 > >>think is hidden in the following compile options > >> > >>SGI Altix 3000 result : > >>-i8 -extend_source -O3 -mP2OPT_hlo_pref_hint=3 -openmp > >> > >>HP SuperDome result H > >>cc +DSitanium2 +DD64 +O3 +Odataprefetch -Wl,+pd,64M -c second_wall.cL > >>mpif90 -o stream_d.mpi +Ofaster +DSitanium2 -Wl,+pd,64M +DD64 +Onoopenmp4 > >>+extend_source +noppu stream_mpi.f second_wall.o > >> > >  > > G > > Are you deliberately hiding the truth by being vague and confusing?  > >  > B > What it doesn't help you understand is the bisectional bandwidth@ > of the system and so although the STREAMS bandwidth number mayB > be interesting to certain types of HPC tests they don't tell you* > much that helps you for other workloads.  E When benchmarks don't show the correct figures from a vendor view, it D is the benchmark which is wrong... It is well known that the currentA Sun large server don't have excellent bandwidth even without this 
 benchmark.   ------------------------------    Date: 22 Oct 2004 04:25:05 -0700' From: icerq4a@spray.se (David Svensson) ( Subject: Re: OT: Short Article about Sun= Message-ID: <734da31c.0410220325.447410f6@posting.google.com>   q Andrew Harrison <andrew_remove_.harrison@s_u_n.com> wrote in message news:<cl5stf$k4r$1@new-usenet.uk.sun.com>...  > David Svensson wrote: [ > > "John Smith" <a@nonymous.com> wrote in message news:<K82dnYh-wOJC1ujcRVn-rQ@igs.net>...  > > 4 > >>...run bigger databases with fancier graphics... > >  > > ;) > > J > > This is a bit sad, Sun was the last company who did something on theirH > > own, and did not bought the PC revolution. Now also they are turning > > into being a PC company. > > = > Lets assume that what you have posted is a deeply held view  > on your part.   C Yes, I am always very deep when it comes to newsgroups. And this is B not only a view on my part, I usually reflect what I hear and read from other people.  A > So despite your claims we would actually appear to be somewhere ? > between Sun's start point and Sun's adoption of SPARC. Hardly & > a dramatic change in the status quo.  B Sun used to downtalk Linux, were not selling x86 boxes, were not a? friend of Windows. Now, Sun sell x86 boxes, with Linux and even B support Windows on these boxes. PC is what atleast most people say. connected to the terms x86, Windows and Linux.  A > And the new AMD64 boxes will be all Sun's design not an off the @ > shelf system board mated to a peice of tin as is the case with > HP.   D While Sun's new boxes will be in-house, HP's current 4-way AMD64 boxC was the first (, and still is the only) in-house 4-way Opteron box. C Besides, there is not any significant differences between different F x86 vendors, it is still a common x86 box (PC) where people don't careB that much about who did the manufacturing if they get a good deal.  ? > In the meantime Sun continues to develop Niagara and Rock the = > two most radical CMT processors currently under development # > and continues to develop Solaris.   A Yes, Niagara and Rock are really interesting, but I think IBM has F shown that less cores can do very well. I don't expect Niagara to haveC any large impact on the market as whole and Rock is 2008, who knows  what others have in 2008.   = > Solaris 10 as you probably know contains a bunch of wholely > > new technologies currently unavailable in any other OS, ZFS,A > dtrace etc (and if you don't know where have you been) so there B > hardly seems to be a huge letup in the rate of development thereA > and then of course we have Java, JES and JDS all which with the = > exception of JDS are largely technologies that Sun invented  > and continues to develop.  >  > > > > In a few years, the whole world could run on Linux x86-64. > > Not fun. >  > You mean RedHat don't you,  E No, I don't mean Redhat, I rarely have anything to do with Redhat, it # is usually Suse or Debian nowadays.   3 > on a 2-4 way box then there will always be a need " > for something better than Linux.  B That's a laugh, even though Solaris may scale a little better now,@ there is good evidence that the Linux of today can scale good toF 32-way on commericial workloads. SGI is also doing a tremendous job on> Linux kernel scalability. Next year Linux will be even better.  B Solaris has an advantage in RAS over Linux today, that is where it@ excels Linux on something that people care about. IBM, and otherB companies are though on their way to make the RAS better on Linux.   ------------------------------    Date: 22 Oct 2004 04:27:38 -0700' From: icerq4a@spray.se (David Svensson) ( Subject: Re: OT: Short Article about Sun= Message-ID: <734da31c.0410220327.6c6e3932@posting.google.com>   U "Dr. Dweeb" <dr@dweeb.com> wrote in message news:<cl5u4l$10b4$1@news.cybercity.dk>...  > David Svensson wrote: 2 > > "John Smith" <a@nonymous.com> wrote in message, > > news:<K82dnYh-wOJC1ujcRVn-rQ@igs.net>...5 > >> ...run bigger databases with fancier graphics...  > > ;) > > J > > This is a bit sad, Sun was the last company who did something on theirH > > own, and did not bought the PC revolution. Now also they are turning > > into being a PC company. > > > > > In a few years, the whole world could run on Linux x86-64. > > Not fun. > M > Carley's mantra is that in a few years the whole world will run Windoze x86  > 3 > I suggest that this prospect is even less fun :-)   E Yes, ofcourse, but HP is already a PC company. It is a lost cause. ;)  Sun was atleast an alternative.    ------------------------------  % Date: Fri, 22 Oct 2004 16:32:08 +0100 9 From: Andrew Harrison <andrew_remove_.harrison@s_u_n.com> ( Subject: Re: OT: Short Article about Sun0 Message-ID: <clb95p$fjs$1@new-usenet.uk.sun.com>   David Svensson wrote:us > Andrew Harrison <andrew_remove_.harrison@s_u_n.com> wrote in message news:<cl5stf$k4r$1@new-usenet.uk.sun.com>...  >  >>David Svensson wrote:  >>Z >>>"John Smith" <a@nonymous.com> wrote in message news:<K82dnYh-wOJC1ujcRVn-rQ@igs.net>... >>>t >>> 4 >>>>...run bigger databases with fancier graphics... >>>n >>>;)( >>>uI >>>This is a bit sad, Sun was the last company who did something on theirAG >>>own, and did not bought the PC revolution. Now also they are turning  >>>into being a PC company.b >>>e >>= >>Lets assume that what you have posted is a deeply held viewe >>on your part.J >  > E > Yes, I am always very deep when it comes to newsgroups. And this isdD > not only a view on my part, I usually reflect what I hear and read > from other people. >  > A >>So despite your claims we would actually appear to be somewheren? >>between Sun's start point and Sun's adoption of SPARC. Hardlya& >>a dramatic change in the status quo. >  > D > Sun used to downtalk Linux, were not selling x86 boxes, were not aA > friend of Windows. Now, Sun sell x86 boxes, with Linux and evensD > support Windows on these boxes. PC is what atleast most people say0 > connected to the terms x86, Windows and Linux. >  > A >>And the new AMD64 boxes will be all Sun's design not an off the @ >>shelf system board mated to a peice of tin as is the case with >>HP.A >  > F > While Sun's new boxes will be in-house, HP's current 4-way AMD64 boxE > was the first (, and still is the only) in-house 4-way Opteron box.IE > Besides, there is not any significant differences between different-H > x86 vendors, it is still a common x86 box (PC) where people don't careD > that much about who did the manufacturing if they get a good deal. >   5 Hardly, HP's box is in fact an off the shelf sytstems 3 board mated to a standard HP chassis, this is ahead 8 of Dell but somewhat behind actually designing the whole system themselves.  7 And people do care if the vendor is able to package thei2 x86 box better or deliver equivalent functionality: at a lower cost. Thats what Dell or rather Dells suppliers have been rather good at.l   > ? >>In the meantime Sun continues to develop Niagara and Rock the = >>two most radical CMT processors currently under developmentn# >>and continues to develop Solaris.  >  > C > Yes, Niagara and Rock are really interesting, but I think IBM hastH > shown that less cores can do very well. I don't expect Niagara to haveE > any large impact on the market as whole and Rock is 2008, who knowso > what others have in 2008.o >   E Lets wait and see shall we. I am not predicting that Rock and Niagara.F will take the market by storm, if I did you could accuse me of being a= Rob Young clone, nor is it worth assuming that they will havet no impact either.,     > = >>Solaris 10 as you probably know contains a bunch of wholelyn> >>new technologies currently unavailable in any other OS, ZFS,A >>dtrace etc (and if you don't know where have you been) so thereuB >>hardly seems to be a huge letup in the rate of development thereA >>and then of course we have Java, JES and JDS all which with the4= >>exception of JDS are largely technologies that Sun inventedl >>and continues to develop.E >> >> >>= >>>In a few years, the whole world could run on Linux x86-64.- >>>Not fun.- >> >>You mean RedHat don't you, >  > G > No, I don't mean Redhat, I rarely have anything to do with Redhat, itC% > is usually Suse or Debian nowadays.  >   > Over 80% of the commercial Linux market is dependant on RedHat8 both from a platform perspective and from an ISV revenue standpoint.t  @ Therefore is quite fair to say that at the moment the commercial4 Linux market is RedHat at least in the server space. > 3 >>on a 2-4 way box then there will always be a need " >>for something better than Linux. >  > D > That's a laugh, even though Solaris may scale a little better now,B > there is good evidence that the Linux of today can scale good toH > 32-way on commericial workloads. SGI is also doing a tremendous job on@ > Linux kernel scalability. Next year Linux will be even better. >   = So provide some actual numbers that support your claim. Don'ta= use anything from SGI we are talking commercial workloads notI< embarassingly parallel HPC and don't include RAC for obvious reasons.  + Guess what you will not be able to respond.a  D > Solaris has an advantage in RAS over Linux today, that is where itB > excels Linux on something that people care about. IBM, and otherD > companies are though on their way to make the RAS better on Linux.  4 Really, so Linux has something that it equivalent to& greenline and dtrace, how interesting.   So how about security ?   5 Or performance for that matter, Solaris 10 absolutely>9 trashes Linux on network throughput because of fireengine  on the same hardware.>  8 If I was you I would get on the Software Express Program9 for Solaris 10 and try it on an x86 box, let me know what & you think. Should be very interesting.   Regardsg Andrew Harrisona   ------------------------------  % Date: Fri, 22 Oct 2004 13:03:39 -04000- From: JF Mezei <jfmezei.spamnot@teksavvy.com>o( Subject: Re: OT: Short Article about Sun, Message-ID: <41793D60.9E935130@teksavvy.com>   Andrew Harrison wrote:G > Lets wait and see shall we. I am not predicting that Rock and Niagara   > will take the market by storm,  L By the time they come out, won't they just be "competitive" chips as opposed to "revolutionary" ones ?r  N Since Power since to be the leader now with every sign that they will keep theK #1 spot for some time, and with the 8086 getting up there with the big boyssK (especially the AMD version), it seems that for every bump in features thateN the 8086 gets, it narrows the gap with the big chips, thus reducing the market niche for those chips.   The 8086 has large volume.F Power has medium volum with both enterprise systems and Apple systems.) Both Sparc and IA64 are low volume chips..  J Expectation of IA64 are now so low that I think people are happy if it canJ simply keep up within 2 years of its competitors. (When DID Power get dual& core ? When WILL IA64 get dual core ?)  M I will not be surprised if SUN eventually goes all 8086 when AMD has versions L that really scale to large enterprise systems. And I believe that until thisN happens, Sun will need Sparc to remain in the game, and Sun will want Sparc to remain in the game.   G SPARC is what differentiates Sun from Dell.  Dell can't get some of the2L serious enterprise sales because its offering just doesn't scale so high dueI to limits in the intel 8086. SUN can offer much larger systems because ofa Sparcs's added capabilities.  5 (Same applies to IBM with Power, and HP with PaRisc).:  B > Therefore is quite fair to say that at the moment the commercial6 > Linux market is RedHat at least in the server space.  J Agreed. However, where Linux has a big edge is that Linux expertise is nowL widely available because so many have turned thise 8086 game controller from Windows to Linux.   N In that sense, Linux has a far more succesful hobbyist programme than VMS. AndJ I can see why Mr Dachtera has been shouting for VMS to be available on the 8086 for so long.C   ------------------------------   Date: 22 Oct 2004 06:14:53 GMT2 From: "Dave Weatherall" <djw-nothere@nospam.nohow>$ Subject: Re: pthreads and scheduling? Message-ID: <DTiotGxQ0bj6-pn2-zL5fIxYncZy9@dave2_os2.home.ours>r  2 On Thu, 21 Oct 2004 10:31:28 UTC, "Johan Nilsson" ) <johan.nilsson@---.esrange.ssc.se> wrote:    > ? > "Dave Weatherall" <djw-nothere@nospam.nohow> wrote in messagen; > news:DTiotGxQ0bj6-pn2-MvFfTDfg3qpK@dave2_os2.home.ours...t5 > > On Tue, 19 Oct 2004 07:45:30 UTC, "Johan Nilsson" - > > <johan.nilsson@---.esrange.ssc.se> wrote:- > >- >  > [snip] >  > >cI > > I read the passage yesterday and it makes sense. UPCALLS apply to thejF > > kernel's ability resolve 'waits' bewteen threads. In the 'problem'F > > case it has to resolve an across-(VMS)process wait/block scenario.E > > Hoff might describe that as 'interesting'. I can imagine that theeG > > passage describes the 'worst-case' scenario and that the programmer F > > should design a solution with that in mind. I read it such that itJ > > describes the situation before UPCALLS were introduced. i.e. a processE > > WAIT condition was just that, i.e. the _process_ went into a wait F > > state, even if another thread could run, and nothing then happenedF > > until the process/thread scheduler was triggered again by the nextH > > time-slice allocation. This would explain why your other threads get > > some CPU time. > M > To be honest, I'm not equally sure that it makes sense ... at least not for J > me. If the I/O thread is not saturating the CPU (and it is not) then theJ > main thread should get some time pretty quick - and as soon as it entersL > $waitfr on the common event flag everything else "should" hang. Or is thatN > only if the entering thread has equal or higher priority than the others ... > or ...  F I'm not sure but I believe its related to how Decthreads uses AST's toF control the scheduling of the threads. I've always had the impression F that the delivery of the AST wakes the _process_ into the the context E of the thread scheduling, does its thread work and then goes back to m1 the process wait-state. How it does that I dunno.C  D Before upcalls were introduced, one needed to play some tricks with F AST's and QIO's to give decent throughput. Using QIOW, I saw behaviourA similar to yours, i.e. not full-stop but dramatic slowdown, so I  F issued a QIO with AST and then entered a  thread condition wait-state.? The i/o completion AST is then used to satisfy the thread wait.   A That was fine when we were still reading/writing to mag-tape via 0E QIO... I still do it in some cases now. My lowest common denominator aC is still VMS 6.2 and UPCALLS aren't available. Still looking to be c 7.3-1 by the end of the year.c  H > It sure would be nice to have a deeper explanation on the internals ofH > thread scheduling under VMS + interaction with process scheduling. I'mK > curious as to why the usage of a common event flag is an exception to theaI > otherwise good thread implementation (i.e. when upcalls are enabled). ItN > recall reading "VAX/VMS internals and data structures" (or similar) sometime > at version 5.<something>.r > : > Are there any recent good VMS internals books available?  C ISTR someone was revising it last year or am I confusing that with   Hoff's DCL book. t  I > > Anyway, it struck me that if you change your process interface to usehH > > $READEF, then you can have your main sleep for an appropriate periodJ > > of time , wake-up, read the relevant event flag cluster, check whetherF > > it's time to close down and act accordingly. The sleep can be doneA > > with a Pthread timed wait, preferable before upcalls, or  VMS  > > scheduled  wake-up.t > H > Did that (almost); now using a maximum wait for read I/O to complete +M > always check the event flag when the wait returns using $readef. I normally D > detest polling, but couldn't see any other way around the problem. > # > Thanks for the suggestion though.u  > Glad its working. Agree with you on polling. It usually seems  inelegant, particularly on VMS.s   -- ( Cheers - Dave W.   ------------------------------  % Date: Fri, 22 Oct 2004 14:14:05 +0100aE From: Tim ffrench-Lynch <tim-DOT-ffrench-HYPHEN-lynch@baesystems.com>o Subject: Public DS10L ?r. Message-ID: <4179079D.638E48D1@baesystems.com>  A I'm planning to try buy a used 617MHz DS10L to improve processingd- speed on the VAX/Alpha cluster I use at work.r  D Does anyone know of any public access DS10L systems that I could log6 into remotely to test performance with my application?   Tim-   ------------------------------  % Date: Fri, 22 Oct 2004 06:17:19 -0700)# From: "Tom Linden" <tom@kednos.com>i' Subject: Re: Question about Descriptors2( Message-ID: <opsf9wi5r2zgicya@hyrrokkin>  H On Fri, 22 Oct 2004 00:24:50 -0400, David Froble <davef@tsoft-inc.com>   wrote:   > Larry Kilgallen wrote: >a< >> In article <417548E8.E5B8CC1D@adldata.com>, sol gongola   >> <sol@adldata.com> writes: >>J >>> Descriptors are not "The Solution" to the problem. They are a way to  K >>> manage a program. Languages that supported strings worked fine before  tL >>> DEC had descriptors. Even null terminated strings are not the problem.  G >>> The problem is in the coding practice where simple checks are not  r8 >>> being performed to prevent undesirable data overlays+ >>> including buffer overrun type problems.l >>>rF >>  The problem is people with computers who are unwilling to automateD >> mundane tasks.  There have been languages with support for takingB >> care of string length details (Pascal, PL/I, Ada -- in order ofE >> participation of proponents in this discussion :-) for years.  ButwE >> some people insist on "bare-metal" programming in C-like languagescE >> for no good public reasons (but possibly good private reasons like ) >> a cost-overrun payment on a contract).w >> >t? > You forgot BASIC, which may be the best at string processing.e Better than PL/I?w >e@ > Careful with the conspiricy theories.  Ammunition for JF.  :-) >- > Dave >o       -- >C Using Opera's revolutionary e-mail client: http://www.opera.com/m2/    ------------------------------  # Date: Fri, 22 Oct 2004 13:47:11 GMT ! From: Nigel Barker <nigel@hp.com>l' Subject: Re: Question about DescriptorsI8 Message-ID: <1q3in0d2pmhund7pbsp9u72j0po3ges6qp@4ax.com>  H On Fri, 22 Oct 2004 06:17:19 -0700, "Tom Linden" <tom@kednos.com> wrote:  @ >> You forgot BASIC, which may be the best at string processing. >Better than PL/I?   Better than MUMPS?   -- Nigel Barker Live from the sunny Cote d'Azur-   ------------------------------  % Date: Fri, 22 Oct 2004 07:13:47 -0700n# From: "Tom Linden" <tom@kednos.com>l' Subject: Re: Question about Descriptorsu( Message-ID: <opsf9y49gczgicya@hyrrokkin>  D On Fri, 22 Oct 2004 13:47:11 GMT, Nigel Barker <nigel@hp.com> wrote:  J > On Fri, 22 Oct 2004 06:17:19 -0700, "Tom Linden" <tom@kednos.com> wrote: >eA >>> You forgot BASIC, which may be the best at string processing.  >> Better than PL/I? >  > Better than MUMPS? >eE Apparently no online copy of the ref manual for MUMPS?  I was curiousTE to compare.  What does MUMPS have that is not in PL/I? Or vice-versa?oJ If you go to the following table of contents you can see the functionality supported under String.L  N http://www.kednos.com/pli/docs/REFERENCE_MANUAL/6291pro_index_007.html#index_s   > -- > Nigel Barker! > Live from the sunny Cote d'Azur        -- tC Using Opera's revolutionary e-mail client: http://www.opera.com/m2/    ------------------------------  % Date: Fri, 22 Oct 2004 10:31:50 -0400t3 From: Steve Lionel <Steve.Lionel@REMOVEintelME.com>a' Subject: Re: Question about Descriptorso8 Message-ID: <ic6in0lbvgia5e0ndrhlas4p62lkmq09c9@4ax.com>  M On Fri, 22 Oct 2004 00:24:50 -0400, David Froble <davef@tsoft-inc.com> wrote:-  > >You forgot BASIC, which may be the best at string processing.  N BASIC?  PL/I?  MUMPS?  Pikers.  For string processing, no language I have ever seen beats SNOBOL4.g   Stevee   ------------------------------  % Date: Fri, 22 Oct 2004 07:37:02 -0700r# From: "Tom Linden" <tom@kednos.com> ' Subject: Re: Question about Descriptors ( Message-ID: <opsf9z70gczgicya@hyrrokkin>  2 On Fri, 22 Oct 2004 10:31:50 -0400, Steve Lionel  ' <Steve.Lionel@REMOVEintelME.com> wrote:.  J > On Fri, 22 Oct 2004 00:24:50 -0400, David Froble <davef@tsoft-inc.com>   > wrote: >e@ >> You forgot BASIC, which may be the best at string processing. > H > BASIC?  PL/I?  MUMPS?  Pikers.  For string processing, no language I   > have everX > seen beats SNOBOL4.o9 True, but did it do anything _besides_ string processing?, >  > Stever       -- nC Using Opera's revolutionary e-mail client: http://www.opera.com/m2/    ------------------------------  # Date: Fri, 22 Oct 2004 15:22:59 GMTo! From: Nigel Barker <nigel@hp.com>i' Subject: Re: Question about Descriptorsr8 Message-ID: <q49in013vfeccsfefngmkv5or526arp31r@4ax.com>  H On Fri, 22 Oct 2004 07:13:47 -0700, "Tom Linden" <tom@kednos.com> wrote:  E >On Fri, 22 Oct 2004 13:47:11 GMT, Nigel Barker <nigel@hp.com> wrote:  > K >> On Fri, 22 Oct 2004 06:17:19 -0700, "Tom Linden" <tom@kednos.com> wrote:c >>B >>>> You forgot BASIC, which may be the best at string processing. >>> Better than PL/I?g >> >> Better than MUMPS?5 >>F >Apparently no online copy of the ref manual for MUMPS?  I was curiousF >to compare.  What does MUMPS have that is not in PL/I? Or vice-versa?K >If you go to the following table of contents you can see the functionalityo >supported under String. >IO >http://www.kednos.com/pli/docs/REFERENCE_MANUAL/6291pro_index_007.html#index_sp  O I was joking a little. My memory of MUMPS is from when I was an expert about 20oN years ago. To many brain cells have died &/or the new knowledge has forced outP the old. I do recall it having powerful string manipulation functions as well asO a very fast text database of sparse arrays. Actually syntactically strings wereaP its only data type so it need powerful functions. It's now called 'M' & there is< an ANSII standard. I found some FAQs & other pointers here:-) http://www.mcenter.com/mtrc/mfaqhtm1.htmlt   -- Nigel Barker Live from the sunny Cote d'Azurd   ------------------------------  # Date: Fri, 22 Oct 2004 15:24:09 GMTc! From: Nigel Barker <nigel@hp.com>e' Subject: Re: Question about Descriptorsv8 Message-ID: <qe9in0tjh1vueikms0l8pps4btr18j5ru7@4ax.com>  0 On Fri, 22 Oct 2004 10:31:50 -0400, Steve Lionel' <Steve.Lionel@REMOVEintelME.com> wrote:y  N >On Fri, 22 Oct 2004 00:24:50 -0400, David Froble <davef@tsoft-inc.com> wrote: >r? >>You forgot BASIC, which may be the best at string processing.. > O >BASIC?  PL/I?  MUMPS?  Pikers.  For string processing, no language I have everg >seen beats SNOBOL4. >  >Steve  G Funnily enough in revisiting my prehistory by browsing the MUMPS FAQ atuB http://www.mcenter.com/mtrc/mfaqhtm1.html I found this statement:-  L good string handling capabilities: better than BASIC, not as good as SNOBOL4   -- Nigel Barker Live from the sunny Cote d'Azur    ------------------------------  % Date: Fri, 22 Oct 2004 11:58:19 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com>a' Subject: Re: Question about Descriptors , Message-ID: <41792E15.5CA67634@teksavvy.com>   Nigel Barker wrote:uI > Funnily enough in revisiting my prehistory by browsing the MUMPS FAQ at D > http://www.mcenter.com/mtrc/mfaqhtm1.html I found this statement:- > N > good string handling capabilities: better than BASIC, not as good as SNOBOL4  / How does one define "good string handling" ????a  M When I moved from Basic to C on DOS, I found trememdous improvement in string L handling due to the richer set of string handling library. This was in earlyT 1980s. (along with ability to name variables with more than 6-8 characters etc etc).  N Perhaps basic has evolved since then. Perhaps VAX/DEC Basic is a huge superset/ of "Basic" (was there ever a "basic" standard ?P  J And while all language mentioned so far as more or less ancient, how aboutO java or javascript ? How woudl their string handling compare with Snobol4 etc ?i   ------------------------------  % Date: Fri, 22 Oct 2004 11:18:54 -0400 ( From: David Froble <davef@tsoft-inc.com>' Subject: Re: Question about Descriptorse, Message-ID: <417924DE.2040507@tsoft-inc.com>   Tom Linden wrote:   4 > On Fri, 22 Oct 2004 10:31:50 -0400, Steve Lionel  ) > <Steve.Lionel@REMOVEintelME.com> wrote:t > 4 >> On Fri, 22 Oct 2004 00:24:50 -0400, David Froble   >> <davef@tsoft-inc.com>  wrote: >>A >>> You forgot BASIC, which may be the best at string processing.a >> >>I >> BASIC?  PL/I?  MUMPS?  Pikers.  For string processing, no language I  o >> have ever >> seen beats SNOBOL4. > ; > True, but did it do anything _besides_ string processing?s    O I took a look at the PL/I docs.  Just a brief look.  Forgive me for not trying -L to learn the language.  As for comparison, I did find one function that was O interesting, REVERSE, which reverses the characters in a string.  Not sure why 29 one would want to do that, but an interesting capability.a   Don't know MUMPS.w  N Don't know SNOBOL, but from the dim past I seem to remember comments about it % being implemeted for string handling.V  M I guess it comes down to what you're comfortable with using.  I'm limited to tO what BASIC offers, thus don't know what I may be missing.  Regardless, dynamic fQ and fixed strings using descriptors, cocantination, finding sub-strings, setting  O sub-strings, anything I've ever needed to do.  In addition, (no pun intended),  O arithmetic operations on strings of digits, though I don't use this capability.e  N   Not specific to strings, but one of the capabilities of BASIC is the use of O variables without the need to DECLARE the variables.  BASIC will automatically rN define the variables for you if you use default data typing.  For example, B% N indicates an integer, B$ indicates a string, and B indicates a floating point O number.  Numeric data types can have a default size, BYTE, WORD, LONG, SINGLE,  K DOUBLE, etc, within a progrom or routine.  Specific variables can still be . DECLARED as a different size.   N Not wanting to start a war with those who believe in declaring all variables, P but my opinion is, if the compiler can do this menial work, why should I do it. 5   Don't we use computers to automate the menial work?    Dave     -- e4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Roadh Vanderbilt, PA  15486t   ------------------------------  % Date: Fri, 22 Oct 2004 09:20:11 -0700n# From: "Tom Linden" <tom@kednos.com>o' Subject: Re: Question about Descriptorss( Message-ID: <opsf94zxaazgicya@hyrrokkin>  H On Fri, 22 Oct 2004 11:18:54 -0400, David Froble <davef@tsoft-inc.com>   wrote:   > Tom Linden wrote:2 >T6 >> On Fri, 22 Oct 2004 10:31:50 -0400, Steve Lionel   * >> <Steve.Lionel@REMOVEintelME.com> wrote: >>6 >>> On Fri, 22 Oct 2004 00:24:50 -0400, David Froble  ! >>> <davef@tsoft-inc.com>  wrote:, >>>iB >>>> You forgot BASIC, which may be the best at string processing. >>>y >>>nK >>> BASIC?  PL/I?  MUMPS?  Pikers.  For string processing, no language I   r
 >>> have ever  >>> seen beats SNOBOL4.>= >>  True, but did it do anything _besides_ string processing?a >  >VK > I took a look at the PL/I docs.  Just a brief look.  Forgive me for not  cD > trying to learn the language.  As for comparison, I did find one  L > function that was interesting, REVERSE, which reverses the characters in  I > a string.  Not sure why one would want to do that, but an interesting   
 > capability.   K All the string functions operate on either characters or bits.  A number of(J the BIFs (Builtin Functions) operate from left to right, so if you wanted,J for example to know the position of the last occurence oc some char, say   'A'h in a string S, you might write! length(S) - index(reverse(S),'A')aG then if you wanted to truncate the string starting with that char and  b ignore( the first b-1 characters you could write0 substr(s, b, length(S) - index(reverse(S),'A') )  J Note that since these are part of the language and not the runtime, data   types L are semantically validated and converted as appropriate in accordance with   ae very well defined set of rules  K http://www.kednos.com/pli/docs/REFERENCE_MANUAL/6291pro_017.html#express_24a  J S can be either fixed length or varying string with dynamic length prefix.  J Now if S were a bit string corresponding, e.g. to an address  (UNSPEC of aF pointer) then the above could be used to determine the alignement by   findingr  the position of the last '1' bit     >  > Don't know MUMPS.t >eH > Don't know SNOBOL, but from the dim past I seem to remember comments  0 > about it being implemeted for string handling. >tE > I guess it comes down to what you're comfortable with using.  I'm  fI > limited to what BASIC offers, thus don't know what I may be missing.   oK > Regardless, dynamic and fixed strings using descriptors, cocantination,  nJ > finding sub-strings, setting sub-strings, anything I've ever needed to  L > do.  In addition, (no pun intended), arithmetic operations on strings of  - > digits, though I don't use this capability.t >[J >   Not specific to strings, but one of the capabilities of BASIC is the  K > use of variables without the need to DECLARE the variables.  BASIC will   F > automatically define the variables for you if you use default data  I > typing.  For example, B% indicates an integer, B$ indicates a string,  iK > and B indicates a floating point number.  Numeric data types can have a  tL > default size, BYTE, WORD, LONG, SINGLE, DOUBLE, etc, within a progrom or  I > routine.  Specific variables can still be DECLARED as a different size.a >aF > Not wanting to start a war with those who believe in declaring all  J > variables, but my opinion is, if the compiler can do this menial work,  K > why should I do it.   Don't we use computers to automate the menial work?k >c > Dave >s >s       -- uC Using Opera's revolutionary e-mail client: http://www.opera.com/m2/s   ------------------------------  % Date: Fri, 22 Oct 2004 13:12:40 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com>h' Subject: Re: Question about Descriptors , Message-ID: <41793F7D.16D6E9A8@teksavvy.com>   David Froble wrote:iM > to learn the language.  As for comparison, I did find one function that wassP > interesting, REVERSE, which reverses the characters in a string.  Not sure why; > one would want to do that, but an interesting capability.a  K ALLIN1 also has that ability.  Why was it added ? To ease the processing ine0 languages where people write from right to left.  O > Not wanting to start a war with those who believe in declaring all variables,I  E Isn't that being discussed in another topic ? :-) :-) :-) :-) :-) :-).  Q > but my opinion is, if the compiler can do this menial work, why should I do it.-7 >   Don't we use computers to automate the menial work?r   Yes and no.   L Forcing you to formally declare your variables and structures does generallyM lead to better program organisation. And it also helps prevent mis-reuse of aeJ variable. (did you mean for that B$ variable to be newly allocated in thatJ subroutine, or did you mean to re-use the B$ in your mainline program thatL should have been globally defined instead of restricted to only the mainline procedure ?     K Yeah, having to constantly declare status, x, y, z , *ptr1, *ptr2, *ptr3 in L routines is a pain. But you know, it takes only a few seconds with cut/paste (or just type it in).   L Perhaps there is also a question of scaling. Do all languages scale to large projects the same ?>C Or are some language better suited for small independant programs ?i   ------------------------------  % Date: Fri, 22 Oct 2004 10:20:25 -0700+' From: David Mathog <mathog@caltech.edu>c7 Subject: Re: Retain Trust 'warranty' period almost over,+ Message-ID: <clbfgq$9ph$1@naig.caltech.edu>p   David Froble wrote:a > John Smith wrote:  > C >> Has the former Compaq, that entertaining enterprise now known as L >> "carly(tm)'s World", extended the trade-in and license transfer guaranteeF >> now that the 3-year 'warranty' period is near expiry (December 31,  >> 2004 last? >> I checked) and there still aren't any GA Itanic/VMS systems?p >>H > Maybe it says that the people implementing the program didn't imagine J > the itanic would still not be available after 3 years.  You're implying 4 > cunning, when ineptness easily explains the issue. >   0 Is ineptness a defense against a fraud charge???  8 It looks like HP would be open to legal action for falseD advertising and/or breach of contract if they simply let the program expire without providing:4   A. the promised systemso  F or, more likely (since its seems pretty clear they will not have such  systems to deliver):   B. something of "equal value"o   or, even more likely,w  @ C. refunds and or compensation (ie, $$$ to keep it out of court)     Regards,   David Mathog mathog@caltech.edu   ------------------------------  # Date: Fri, 22 Oct 2004 10:08:26 GMTi: From: sales@careerpartner.co.uk<sales@careerpartner.co.uk>" Subject: sales@careerpartner.co.uk8 Message-ID: <u_4ed.12020$Pk2.11715@newsfe5-win.ntli.net>   <HTML> <HEAD>A <META NAME="GENERATOR" Content="Microsoft DHTML Editing Control">  <TITLE></TITLE>  </HEAD>a <BODY> <P><A I href="mailto:sales@careerpartner.co.uk">sales@careerpartner.co.uk</A></P> M <P><A href="mailto:info@careerpartner.co.uk">info@careerpartner.co.uk</A></P>l <P><A M href="mailto:support@careerpartner.co.uk">support@careerpartner.co.uk</A></P>d
 <P>&nbsp;</P>. </BODY>  </HTML>t   ------------------------------    Date: 22 Oct 2004 08:54:46 -0500 From: briggs@encompasserve.org+ Subject: Re: SYSUAF => export passwords ???h3 Message-ID: <YLyDSCJ9mkFo@eisner.encompasserve.org>i  ^ In article <wB7ed.62$dq4.3482@tor-nn1.netcom.ca>, "Syltrem" <syltremzulu@videotron.ca> writes:< > You can update the record with just the password, with DCLK > Just need to know the offset and length of the password field in a sysuaf " > record. Easy enough to find out.  A You need more than the password.  You need the algorithm and salte as well.   	John Briggs   ------------------------------  % Date: Fri, 22 Oct 2004 11:22:58 -0400r* From: "Syltrem" <syltremzulu@videotron.ca>+ Subject: Re: SYSUAF => export passwords ??? 1 Message-ID: <2A9ed.70$dq4.3819@tor-nn1.netcom.ca>m  5 <briggs@encompasserve.org> a crit dans le message de - news:YLyDSCJ9mkFo@eisner.encompasserve.org... = > In article <wB7ed.62$dq4.3482@tor-nn1.netcom.ca>, "Syltrem"a" <syltremzulu@videotron.ca> writes:> > > You can update the record with just the password, with DCLF > > Just need to know the offset and length of the password field in a sysuaf$ > > record. Easy enough to find out. >0C > You need more than the password.  You need the algorithm and salt 
 > as well. >o
 > John Briggsu  . Even for the same username? (what to OP asked)< Extract pwd field from one uaf file and load into the other?  : The password field is 8 bytes long at offset 340, I found.  I It's not possible to do the update in DCL through DECnet because of this:fE $ open/read/write/share f vfm"system password"::sys$system:sysuaf.datv $ read f rec /k="SYLTREM"w; %RMS-E-NETBTS, network buffer too small for 644 byte recordu  J So you must save the password in a file, then log on to the remote note an read the file from there   Let's see how we can do it:a   Step1: get the pwd) $ open/read/share f sys$system:sysuaf.date $ read f rec/k="""SYLTREM""" $ pwd=f$extr(340,8,rec)h	 $ close fg   Step2: save it to a file $ open/write f pwd.txt
 $ write f pwda	 $ close f   F Step3: obtain the new password from the saved file, on the remote node $ set h XYZd Username: system	 Password:a5 SYS-VFM> open/read f atlas"syltrem password"::pwd.txt  SYS-VFM> read f pwdA SYS-VFM> close f  " Step 4: update it in user's record6 SYS-VFM> open/read/write/share f sys$system:sysuaf.dat! SYS-VFM> read f rec/key="SYLTREM"h SYS-VFM> rec[340,8]:="''pwd'"- SYS-VFM> write/sym/update f rec  SYS-VFM> close f   Step 5: test it: SYS-VFM> set h 0 Username: syltremc Password: (new password)#         Welcome to OpenVMS VAX V7.32     Works fine here..o   Regards, -- . Syltremd   OpenVMS 7.3-1 + Oracle 8.1.7.4H http://pages.infinit.net/syltrem (OpenVMS related web site, en franais)% ---zulu is not in my email address---a   ------------------------------  # Date: Fri, 22 Oct 2004 15:37:47 GMTb3 From: "James T. Sprinkle" <oicmrsnakes@hotmail.com>s7 Subject: VAX 6000-610 For Sale/Trade in Seattle WA Area 1 Message-ID: <fP9ed.15$OL5.2428@news-west.eli.net>t  E I am wondering if anyone would be interested in a VAX 6000-610 that InD recently purchased.  I am looking to possibly trade for a desktop orI deskside model.  This is for Pick-Up only, I will not ship it.  I do havedL the ability to load it onto a truck, however.  If you are interested, e-mail me.,   James T. SprinkleP   ------------------------------    Date: 22 Oct 2004 05:41:05 -0700( From: bob@instantwhip.com (Bob Ceculski)& Subject: Re: What happened with SEVMS?= Message-ID: <d7791aa1.0410220441.744430b1@posting.google.com>R  a JF Mezei <jfmezei.spamnot@teksavvy.com> wrote in message news:<41787F60.75713CA1@teksavvy.com>...o > Larry Kilgallen wrote:O > > What mandatory access control implementations are validated at B1 on IA32 ?u >  > That is not the issue. > N > The issue is that Sun now has bragging rights about very secure systems used > by the military. > L > VMS used to have such bragging rights, with mention fo VMS used on missionA > critical military msisions. That gave it an exclusive position.s > P > But now, if even Sun can claim this,  VMS no longer has any edge, and the fact@ > that VMS no longer has SEVMS means even fewer bragging rights. > B > This is all about strategy for bragging rights (aka: marketing). > N > There are projects which may not result in standa alone profits for that oneQ > project, but whcih are stretegic for the success of the produc line as a whole.     6 the only way to secure a sun box is to shut it off! :)4 The cert counts don't lie ... VMS still trounces all others!P   ------------------------------  % Date: Fri, 22 Oct 2004 11:45:43 -0400 ( From: David Froble <davef@tsoft-inc.com>& Subject: Re: What happened with SEVMS?, Message-ID: <41792B27.7080008@tsoft-inc.com>   JF Mezei wrote:e   > Larry Kilgallen wrote: > M >>What mandatory access control implementations are validated at B1 on IA32 ?h >> >  > That is not the issue. > N > The issue is that Sun now has bragging rights about very secure systems used > by the military. > L > VMS used to have such bragging rights, with mention fo VMS used on missionA > critical military msisions. That gave it an exclusive position.  > P > But now, if even Sun can claim this,  VMS no longer has any edge, and the fact@ > that VMS no longer has SEVMS means even fewer bragging rights. > B > This is all about strategy for bragging rights (aka: marketing). > N > There are projects which may not result in standa alone profits for that oneQ > project, but whcih are stretegic for the success of the produc line as a whole.t >   P VMS is being used in military systems, and at times the 'customer' would not be & pleased at publicity about that usage.  P Since this type of customer doesn't seem to like publicity, then with Sun being T so mouthy, maybe they are making noise, but don't really have the installed systems.   Dave   -- o4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Roade Vanderbilt, PA  15486a   ------------------------------  % Date: Fri, 22 Oct 2004 09:16:34 -0500 ( From: brandon@dalsemi.com (John Brandon) Subject: ZIP & work area?e1 Message-ID: <04102209163416@dscis6-0.dalsemi.com>t  + Frequently we find ourselves running low onu= disk space when using ZIP - typically when ZIPing huge files.t  + It seems ZIP uses your directory default as>( the location for work files (ZIxxxxxx.;)  4 Other than using SET DEFAULT is there a way to force2 ZIP into using a different disk for its work area?  : I am thinking of how SORT uses SORTWORKx files to do this.  2 I have tried changing SYS$SCRATCH and to no avail.     J*o*h*n B*r*a*n*d*o*nv VMS Systems Administratort* firstname.lastname.spam.me.not@dalsemi.com   ------------------------------  + Date: Fri, 22 Oct 2004 09:25:39 -0500 (CDT)  From: sms@antinode.org Subject: ZIP & work area? ) Message-ID: <04102209253944@antinode.org>o  ( From: brandon@dalsemi.com (John Brandon)   [...]a- > It seems ZIP uses your directory default ass* > the location for work files (ZIxxxxxx.;) > 6 > Other than using SET DEFAULT is there a way to force4 > ZIP into using a different disk for its work area? [...]   B    It's clear that you haven't been paying attention to the recent (12-OCT-2004) Zip discussions.  E    There's a (poorly documented) "-b <dir>" (or "/TEMP_PATH = <dir>")eE option which allows the user to specify a directory for the temporarytH file.  Alternatively, putting the temp file in the right place is one of< the fixes in the replacement files available for Zip 2.3 at:  4       http://www.antinode.org/ftp/info-zip/zip-2_3x//       ftp://ftp.antinode.org/info-zip/zip-2_3x/v  E    This fix should be included in Zip 3.0, which may become available  someday.  H ------------------------------------------------------------------------  4    Steven M. Schweda               (+1) 651-699-98183    382 South Warwick Street        sms@antinode-org     Saint Paul  MN  55105-2547    ------------------------------  % Date: Fri, 22 Oct 2004 10:47:08 -0400 # From: sol gongola <sol@adldata.com>e Subject: Re: ZIP & work area?k+ Message-ID: <41791D6C.6C18B47C@adldata.com>x   John Brandon wrote:e > - > Frequently we find ourselves running low on.? > disk space when using ZIP - typically when ZIPing huge files.o > - > It seems ZIP uses your directory default asu* > the location for work files (ZIxxxxxx.;) > 6 > Other than using SET DEFAULT is there a way to force4 > ZIP into using a different disk for its work area? > < > I am thinking of how SORT uses SORTWORKx files to do this. > 4 > I have tried changing SYS$SCRATCH and to no avail. >  > J*o*h*n B*r*a*n*d*o*n  > VMS Systems Administratore, > firstname.lastname.spam.me.not@dalsemi.com  $ zip 2.1 had this in its help output:   -b   use "path" for temp filer    6 zip 2.3 mentions it in the command string description:  > Zip 2.3 (November 29th 1999). Usage: zip=="$disk:[dir]zip.exe"N zip [-options] [-b path] [-t mmddyyyy] [-n suffixes] [zipfile list] [-xi list]J   The default action is to add or replace zipfile entries from list, which<   can include the special name - to compress standard input.B   If zipfile and list are omitted, zip compresses stdin to stdout.J   -f   freshen: only changed files  -u   update: only changed or new filesI   -d   delete entries in zipfile    -m   move into zipfile (delete files)    solh   ------------------------------  % Date: Fri, 22 Oct 2004 10:41:32 -0500r( From: brandon@dalsemi.com (John Brandon) Subject: Re: ZIP & work area?.1 Message-ID: <04102210413279@dscis6-0.dalsemi.com>    Steven M. Schweda wrote:D >    It's clear that you haven't been paying attention to the recent  > (12-OCT-2004) Zip discussions.  ) You are right on that one.  My apologies.e  G >    There's a (poorly documented) "-b <dir>" (or "/TEMP_PATH = <dir>")eG > option which allows the user to specify a directory for the temporaryS	 > file.  a  ; Thanks!  I missed that one - and pooly documented is right.        J*o*h*n B*r*a*n*d*o*nt VMS Systems Administrator'* firstname.lastname.spam.me.not@dalsemi.com   ------------------------------  % Date: Fri, 22 Oct 2004 13:28:39 -0400l# From: sol gongola <sol@adldata.com>e Subject: Re: ZIP & work area?c+ Message-ID: <41794347.8C9B7E76@adldata.com>e   John Brandon wrote:: >  > Steven M. Schweda wrote:F > >    It's clear that you haven't been paying attention to the recent" > > (12-OCT-2004) Zip discussions. > + > You are right on that one.  My apologies.x > I > >    There's a (poorly documented) "-b <dir>" (or "/TEMP_PATH = <dir>")nI > > option which allows the user to specify a directory for the temporaryc	 > > file.i > = > Thanks!  I missed that one - and pooly documented is right.t >  > J*o*h*n B*r*a*n*d*o*nn > VMS Systems Administrator-, > firstname.lastname.spam.me.not@dalsemi.com  - ftp://ftp.info-zip.org/pub/infozip/doc/MANUALt          -b pathA               Use  the  specified  path  for  the  temporary  zipn#               archive. For example:   (                      zip -b /tmp stuff *  A               will put the temporary zip archive in the directoryeA               /tmp, copying over stuff.zip to the current  direc-tA               tory  when  done.  This  option is only useful whentA               updating an existing archive, and the  file  systemeA               containing  this  old  archive does not have enoughaA               space to hold both old and new archives at the sames               time.t   ------------------------------   End of INFO-VAX 2004.587 ************************