1 INFO-VAX	Tue, 24 Dec 2002	Volume 2002 : Issue 710       Contents:1 Re: "VMS will be around long after we retire" ... 1 Re: "VMS will be around long after we retire" ...  Re: Digital server 5000  E-mailing pgm output Re: E-mailing pgm output Re: E-mailing pgm output' Re: F$GETDVI ("AVL") in an IF statement ! Re: First Hammer performance test ! Re: First Hammer performance test ! Re: First Hammer performance test ! Re: First Hammer performance test 9 Re: How to interpret extended File IDs - e.g. (86051,3,0) 9 Re: HP-CUO Membership (Re: DECUS^H^H^H^H^HCUO membership)  Re: Murphy's law at its best Re: Murphy's law at its best- Of Galaxy, Infiniband and Distributed Caching  Re: OpenVMS Branding Query- Re: OT: Whoa! Is Sun aiming at VMS's jugular? - Re: OT: Whoa! Is Sun aiming at VMS's jugular? - Re: OT: Whoa! Is Sun aiming at VMS's jugular? - Re: OT: Whoa! Is Sun aiming at VMS's jugular? - Re: OT: Whoa! Is Sun aiming at VMS's jugular? - Re: OT: Whoa! Is Sun aiming at VMS's jugular? 2 Re: SAN recommendations mixing VMS, Solaris, Linux Re: SCSI controller question Re: SCSI controller question' Re: Semaphores and Shared memory on VMS  Re: UTC time in DCL  Re: vax6k.openecs.org rebirth  Re: vax6k.openecs.org rebirth  Re: vax6k.openecs.org rebirth 2 Re: VMS has got apps ... 3300 and counting per HP!2 Re: VMS has got apps ... 3300 and counting per HP!2 Re: VMS has got apps ... 3300 and counting per HP!2 Re: VMS has got apps ... 3300 and counting per HP!2 Re: VMS has got apps ... 3300 and counting per HP!2 Re: VMS has got apps ... 3300 and counting per HP!F Re: Why does BACKUP/IMAGE need write access to INDEXF.SYS, BITMAP.SYS?@ Re: Your Multi-volume Tape Backups may be bad on all versions of [OT] Why Windows isn't ready  F ----------------------------------------------------------------------  % Date: Tue, 24 Dec 2002 16:05:07 +0100 $ From: "Dr. Dweeb" <Dweeb@nospam.com>: Subject: Re: "VMS will be around long after we retire" .../ Message-ID: <II_N9.41$M05.1127@news.get2net.dk>    Bill Todd wrote:4 > "Main, Kerry" <Kerry.Main@hp.com> wrote in message > L news:BE56C50EA024184DAF48F0B9A47F5CF402660C4B@kaoexc01.americas.cpqcorp.net. > .. > Bill,    .. clip   ? > In other words, she's pleased that HP has to beat off new VMS E > customers (and why wouldn't she be?), but, once again, says nothing 5 > about actually making any effort to *attract* them.  >  > - bill   On this point, sort of,   I Is it true that when Carley invited all the BIG customers from around the G world for a briefing in CA (San Diego ??) that she cancelled out (read: I could not be bothered turning up) and palmed them off on some lower level  suit ?   Dweeb.   ------------------------------  % Date: Tue, 24 Dec 2002 12:20:40 -0600 1 From: "David J. Dachtera" <djesys.nospam@fsi.net> : Subject: Re: "VMS will be around long after we retire" ...' Message-ID: <3E08A578.98A97DAE@fsi.net>    Bill Todd wrote: > [snip]I > In other words, she's pleased that HP has to beat off new VMS customers   ? Be aware that in some contexts, "beat off" is an American slang A euphemism for "masturbate", in which case I suppose that might be ' effective to some degree in some cases.    > (andL > why wouldn't she be?), but, once again, says nothing about actually making > any effort to *attract* them.   E Well, pornographic spam seems to be effective (otherwise why it be so D prevalent?). Perhaps that would be one way to get VMS "exposed" (pun intended) to a new audience.    9 Less than ideal, but since when does that matter anymore?    --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  # Date: Mon, 23 Dec 2002 19:50:33 GMT 9 From: Alan Adams <alan.adams@orchard-way.freeserve.co.uk>   Subject: Re: Digital server 5000? Message-ID: <d2717ea94b.Alan.Adams@orchard-way.freeserve.co.uk>   4 In message <atvpsl$38onb$1@ID-143435.news.dfncis.de>-           "Hans Vlems" <hvlems@iae.nl> wrote:   K > I have a temporary guest: a Digital Server 5000 system. It might become a J > permanent resident provided it will behave and run VMS. The system bootsK > from a VMS 7.3 cd, so that's encouraging. The problem is that I'm used to I > VAXes, VAXstations actually so this is all rather new, please have some ( > patience with the following questions. > J > Unfortunately the SCSI controller is missing, which leads to question 1., > What kind of SCSI controller is supported?I > I have two white flatcables with black connectors on them; 1"x1/8", are  > these SCSI cables?  K These sound a bit like the cables for old-style disks such as RA81 - sorry, K can't remember the proper name for the interface. They would have a smaller @ number of pins than scsi, e.g. 10 to 20, can't remember exactly.   > J > The firmware is at level 5.4 and VMS rather prefers 5.9. Is the firmware > available on-line?= > If so how do I install the firmware? (I have no books...:-(  >  > Hans Vlems >    --  
 Alan Adams& alan.adams@orchard-way.freeserve.co.uk http://www.nckc.org.uk/    ------------------------------    Date: 24 Dec 2002 08:02:59 -0800. From: Jack.Trachtman@vmmc.org (Jack Trachtman) Subject: E-mailing pgm output = Message-ID: <69d784c4.0212240802.4b3859cb@posting.google.com>   B We run all third party pkgs on our hosts, ie we can't modify them.  J Occassionaly I get user requests to be able to automatically e-mail output? form an app that can only either be put into a file or put into  a print que.  G What tricks are people using to redirect this kind of output to e-mail?    Thanks   ------------------------------    Date: 24 Dec 2002 10:34:20 -06003 From: bradhamilton@127.0.0.1 (Bradford J. Hamilton) ! Subject: Re: E-mailing pgm output 3 Message-ID: <YQUVHovC181L@eisner.encompasserve.org>   n In article <69d784c4.0212240802.4b3859cb@posting.google.com>, Jack.Trachtman@vmmc.org (Jack Trachtman) writes:D > We run all third party pkgs on our hosts, ie we can't modify them. > L > Occassionaly I get user requests to be able to automatically e-mail outputA > form an app that can only either be put into a file or put into  > a print que. > I > What tricks are people using to redirect this kind of output to e-mail?  >  > Thanks  J How about writing a batch job that traverses an output directory, e-mails K output to the appropriate individual(s), deletes (or archives) the output,  6 and re-submits itself to run every (say) five minutes?  F I'm sure there will be other, more elegant solutions that will present themselves shortly.   O I realize that this is not strictly "re-direction of output", but you *did* say L that the package could not be modified - if you were truly able to re-direct: output, would that not be a modification of a package?	:-)   ------------------------------  # Date: Tue, 24 Dec 2002 17:40:02 GMT 1 From: Michael Austin <maustin@firstdbasource.com> ! Subject: Re: E-mailing pgm output 2 Message-ID: <3E089BC7.9FA9CFE6@firstdbasource.com>   Jack Trachtman wrote:  > D > We run all third party pkgs on our hosts, ie we can't modify them. > L > Occassionaly I get user requests to be able to automatically e-mail outputA > form an app that can only either be put into a file or put into  > a print que. > I > What tricks are people using to redirect this kind of output to e-mail?  >  > Thanks  
 $set noverify H $runapp :== $dev:[dir]appname  !note the leading "$" before the dev name $define sys$output x.x< $runapp p1 p2 p3 p4 p5 p6  !if there are any input variables) $... do cleanup of file here if necessary " $mail x.x /subj="subject goe here" user@domain.com 	 $del x.x;  $exit   , $submit runapp /param=("''p1'","''p2'",....)  A and submit this to a batch job.  -- and if you are really good at G security and web servers, this can be submitted from a web page.  Works  great!   --   Regards,  6 Michael Austin            OpenVMS User since June 19847 First DBA Source, Inc.    Registered Linux User #261163    ------------------------------    Date: 24 Dec 2002 07:54:47 -0800. From: spamsink2001@yahoo.com (Alan E. Feldman)0 Subject: Re: F$GETDVI ("AVL") in an IF statement= Message-ID: <b096a4ee.0212240754.3ef64d4c@posting.google.com>   ` "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message news:<3E07E791.9B6F1009@fsi.net>... > "Alan E. Feldman" wrote: > > d > > "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message news:<3E068939.25BCE6E7@fsi.net>... > > > JF Mezei wrote:  > > > >  > > > > Z wrote:D > > > > > It fails big time because F$GETDVI causes an error instead > > > > > of returning "FALSE."  > > > > I > > > > But why does F$GETDVI cause an error instead of returning false ?  > > > L > > > Because you're confusing "exists" with "available". A device can existJ > > > but not be available - see HELP SET DEVICE /AVAILABLE. A device thatM > > > does not exist cannot be examined to see whether or not it's available.  > > I > > Ah, but if you can't find it, it's certainly not available. It's also H > > not mounted. Now to determine whether it's a file-oriented device or) > > not, for that you have to examine it.  > E > So, we're making brash assumptions about non-existant DCBs, are we?   E No. OK. This all depends on what you consider to be possibilities for D the "value" of "availability". You can consider the possibilities to? be limited to true and false, *OR*, you can consider them to be = limited to true, false, and "meaningless" (w/ error message).   A DEC chose the latter. That's fine. But the former is a legitimate C alternate point of view. JF apparently belongs to the first school,  you to the second.  C NOW, if you choose the first school, your only choices are true and C false. So, if a device doesn't exist, certainly you can't call that D true. Therefore, it must be false. Therefore, this first school saysF that something that doesn't exist certainly isn't available because ifC it were available, it must exist! There's nothing brash about that.    [...]  > Note that: > , > $ PIPE SHOW DEVICE | SEAR SYS$PIPE MOUNTED > H > ...will not display shadow-set members, which are MOUNTed, but displayJ > as shadow-set members - the word "Mounted" does not appear in the statusF > column or any place else on the line or any relevant lines displayed > sunsequently.   B Ah! But I didn't say *how* to list them. You could write a programC using DCL lexical functions and temporary files to list all mounted ) devices. You're making brash assumptions!    [...] H > > Now, please. We're *really* getting too deep into semantics here. ItJ > > seems to me that DEC simply chose a simple way to do this: If a deviceJ > > doesn't exist, other F$GETDVI functions for that device will return anI > > error. It's a choice the VMS engineers made. And this choice provides H > > more information from the F$GETDVI call. You can get true, false, or? > > an error. If you get an error, you know something is wrong.  [...] > > > > Remember: a device must exist before you can examine its > > > characteristics. > > = > > Yeah, but availability is a status, not a characteristic.  > $ > Stati constitute characteristics.  [...]   E Perhaps. But if the item isn't in the box, you don't have to find the D item to determine that it isn't in the box. You only need to look inD the box. (This is the motivation for the "first school", of course.)B Why do you consider that a brash assumption? Perhaps I should have5 said "a status is not an *intrinsic* characteristic".     [...] ? > Capt. Kirk:	Norman, everything Harry Mudd tells you is a lie.  > 3 > Harry Mudd:	Now listen to me, Norman: I am lying.    That *was* a fun episode!   
 > > [snip]F > > Correct, except for cases like F$GETDVI("NODEX::DKA200:","EXISTS") > > which returns a warning. > E > You can't us ethis to examine devices on another node. Documented /  > expected behavior.  E I never said it wasn't. Someone in this thread said it always returns @ true or false. I was simply providing a counter-example and said* nothing about its being documented or not.   [...]   @ > Logic is present, just not the logic you would seem to prefer.  E I actually like it the way it is. I'm just saying that DEC could have F chosen either way. The first school of thought requires an additional, deductive step of logic.  D > > > How can you know that until you know whether or not you have a > > > kramblefratz?  > > I > > Well, F$GETDVI could have been designed to check for existence if the I > > device is not found, report it as unavailable. But it wasn't designed 
 > > that way.  > , > True. It does not make brash assumptions.    What brash assumptions?     > Rather than lying to you aboutH > a not existant device, you are informed that the device indicated does
 > not exist.    B According to the second school, that is true. But according to theB first school, if it doesn't exist, it isn't available and it isn'tF mounted. But I still like it the way it is on VMS (second school). DEC could have chosen either way.   A You're arguing in favor of the second school here and I like your F argument. This is an excellent argument in favor of the second school.  6 > Availability is irrelevant for non-existant devices.  F Well, I wouldn't put it that way. I would say that it's more important) to be told that the device doesn't exist.   F However, a point in favor of the first school: If it doesn't exist, itB can't be available. Please explain how a device that doesn't exist could be available or mounted? [...]    Disclaimer: JMHO  &-)  Alan E. Feldman    ------------------------------  % Date: Tue, 24 Dec 2002 01:34:27 -0500 ( From: David Froble <davef@tsoft-inc.com>* Subject: Re: First Hammer performance test, Message-ID: <3E07FFF3.5090903@tsoft-inc.com>   Shane Smith wrote:   > Now I E > just wonder what the new 'industry standard' CPU is going to be for F > what is now known as the Wintel market. Perhaps in two years we will+ > talk about the WinAmd or WinAth market ?      Q Intel will never let this happen.  But, when they decide to compete with Hammer,   guess what dies?  L It's never been IA-64 against anything.  The whole issue will be protecting  their wintel market.   Dave   ------------------------------  % Date: Tue, 24 Dec 2002 08:24:33 +0100  From: Dirk Munk <munk@home.nl>* Subject: Re: First Hammer performance test8 Message-ID: <3p2g0vkhnfa465r2vvojpn0tc9m10thhrt@4ax.com>  A On 23 Dec 2002 18:58:16 -0800, bob@instantwhip.com (Bob Ceculski)  wrote:  _ >Dirk Munk <munk@home.nl> wrote in message news:<fc680vs3ulrssfthkirarejoj73lcci54e@4ax.com>... I >> The German magazine CT has issued a first comparisation test between a D >> 1.2 GHz Hammer, a 1.2 GHz Athlon XP and a 2.2 GHz Pentium 4 (real >> clock speeds).  >>  F >> All tests were done in 32 bit mode, Later tests will use the 64 bit >> mode too of course.   >>  H >> The Hammer was a lot faster than the Athlon XP, and a bit slower thanD >> the Pentium. Using AMD's normal way of branding CPU's, the HammerE >> could be branded as a 2000+ processor (=comparible to a 2 GHz P4).  >>  G >> The memory access in particular was extremely fast, also much faster  >> as with the P4. >>  E >> Keeping in mind that this Hammer was just a prototype, the editors I >> expect that the entry model 2 GHz Hammer can be branded as a 3300+ CPU I >> when it hits the market in Q1/Q2 2003. And since the Hammer also has a F >> Alpha EV7 like possibilty for interprocessor communication, I can'tH >> wait to see the first tests for multi-cpu servers and workstations.   >>  H >> Somewhere in the next year I will buy myself a new PC. At that time IG >> will have the choice between pure 32 bit CPU's, a CPU that is a very H >> fast 32 bit CPU and (most likely) a (very?) fast 64 bit CPU (so I can@ >> run 32 bit and 64 bit windooz as well as 64 bit Linux), and aI >> (expensive) CPU that is a very slow 32 bit CPU, and a fast 64 bit CPU.  >>  I >> Three guesses what I'm going to buy if the price is somewhat ok. Now I F >> just wonder what the new 'industry standard' CPU is going to be forG >> what is now known as the Wintel market. Perhaps in two years we will + >> talk about the WinAmd or WinAth market ?  > @ >and I am waiting to see what lawsuits hp files againset amd ...> >some alpha people are there and most likely stole alpha stuffA >just like intel did, only Palmer was paid off to let them do it, ? >but Carly will be different ... hp should collect some hefty $  >for patent violations ...  D Don't forget that AMD had a quite substantial agreement with Digital7 about using the Alpha bus design for their processors.     ------------------------------  % Date: Tue, 24 Dec 2002 11:59:21 -0500 $ From: "rob kas" <rob@netcarrier.net>* Subject: Re: First Hammer performance test3 Message-ID: <3e089269$0$1408$8e9e3842@news.atx.net>   
        Bob  &    Do you every think before you post?                         Rob  "  > A > and I am waiting to see what lawsuits hp files againset amd ... ? > some alpha people are there and most likely stole alpha stuff B > just like intel did, only Palmer was paid off to let them do it,@ > but Carly will be different ... hp should collect some hefty $ > for patent violations ...    ------------------------------   Date: 24 Dec 2002 18:38:28 GMT( From: bill@cs.uofs.edu (Bill Gunshannon)* Subject: Re: First Hammer performance test5 Message-ID: <aua9j4$5o682$2@ID-135708.news.dfncis.de>   3 In article <3e089269$0$1408$8e9e3842@news.atx.net>, ' 	"rob kas" <rob@netcarrier.net> writes:  >        Bob > ( >    Do you every think before you post? >  >                       Rob   % Your question was rhetorical, right??    bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  % Date: Tue, 24 Dec 2002 12:53:05 -0600 1 From: "David J. Dachtera" <djesys.nospam@fsi.net> B Subject: Re: How to interpret extended File IDs - e.g. (86051,3,0)' Message-ID: <3E08AD11.10138FEE@fsi.net>    Larry Kilgallen wrote: > [snip]C >         ( UNSIGNED_LONGWORD ( FIB.FID.FID_RVN.FID_NMX ) * 65536 ) 5 >             + UNSIGNED_LONGWORD ( FIB.FID.FID_NUM )   = Hhmmm... looks amazingly like the formula for a longword UIC:   ' 	(group_number * 65536) + member_number   ' Does anyone remember the RAD50 formula?    --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------    Date: 24 Dec 2002 06:41:03 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) B Subject: Re: HP-CUO Membership (Re: DECUS^H^H^H^H^HCUO membership)3 Message-ID: <0tt6NSq60Cxh@eisner.encompasserve.org>   j In article <3E07BFC5.3239660D@ipl.demon.co.nospam.uk>, Steve Reece <SYSTEM@ipl.demon.co.nospam.uk> writes:  F > This week, you should receive a copy of Ping, the technical magazineF > published by the HP/Works user group.  This is being sent out in theI > hope that HP-CUO members will find it useful and educational.  I have a = > vested interest here in that I have an article in there....  > F > A copy of HP User, the magazine from the HP-CUA user group should be: > finding its way to 'our' members in the near future too.  A Are these real magazines (as distinguished from non-text email) ?    ------------------------------  # Date: Mon, 23 Dec 2002 21:21:30 GMT = From: "Colin Butcher" <colinDOT.butcherAT@xdeltaDOT.coDOT.uk> % Subject: Re: Murphy's law at its best ; Message-ID: <u7LN9.329$ri1.2254900@news-text.cableinet.net>    So where's CAMPAG then?    -- Hope this helps. Cheers, Colin. ' (colinDOT.butcherAT@xdeltaDOT.coDOT.uk)     = "JF Mezei" <jfmezei.spamnot@vl.videotron.ca> wrote in message ) news:3E05836A.BD613A2C@vl.videotron.ca...  > I > This week, a disk drive on my mac broke. Bank froze my accounts because  theyK > thought I had widthdrawn too much money and wouldn't give me time of day.  They( > lost a customer. When things go bad... > G > Anyhow, I get new "old" disk drives, including one of those big 5.25"  bricksL > that is 10 gig, perfect for my all mighty microvax, to replace the 250 megC > hard drive that was the system disk (from an old Apple computer).  > L > Well, not quite. I have another brick, Futjitsu 670-meg. So I plan to makeI > that one the new system disk, and move the data from the fujitsu to the  new 10D > gig seagate. But I only have 2 slots for drives, so I have to plan carefully... >  > Cast of characters on VELO > DATA1 -> new 10 gig drive  > SHIMANO -> 670- meg futjitsu3 > MARINONI -> 250 meg (apple labeled quantum drive)  > - > PEDALS -> 1 gig Seagate on node BIKE (3100)  >  > The plan is to have:4 > MARINONI on the 670 meg fujitsu (formerly Shimano), > SHIMANO on the new 10 gig seagate (DATA1). > L > But my all mighty Microvax 2 has only 2 scsi drive bays. So I have to planL > those things carefully and ensure that the Mvax-II can in fact handle a 10 gig 6 > drive. (in its days, 650 meg was as big as they got) > L > First step was to removed MARINONI to make room for DATA1 so I powered off the L > machines, but I wanted to keep the router and cable modems etc going, so II > fiddled with the switch in the back, hoping to find a config that would  keepL > the power to the "utility" plugs while keeping the actual VAX powered off.% > Unfortunatly, that was to no avail.  > H > So, i power the stuff back on, and as I had configured, the new 10 gig drive , > takes 50 seconds to start spinning. Yeah ! > G > Get into the DILOG config menu (I never do remember exactly the magic J > incantations (deposit stuff into memory then branch to "200"). Start the low ? > level format. Boy, that took a LONG time (measured in hours).  > D > Then, I booted standalone backup from SHIMANO. And Issued a simpleL > backup/image from SHIMANO to DATA1. That seemed to work. (Interesting that onJ > a MVAX-II, the backup really does reach a chunk, and then write a chunk,& > instead of doing both at same time). > J > While this was going on, I was thinking about the next step. With VELO'sI > system disk out, I couldn't really test DATA1 properly. So I set out to  allow K > VELO to boot as a satellite of BIKE. Hadn't used Cluster_config in a long  > while. > G > (Suggestion to engineers: in cluster_config, when you add a satellite  node, it/ > should remind one to think about licences !).  > G > So, i finally manage to boot VELO from BIKE, after having wasted time  because K > I had entered the wrong ethernet address (I had written down the real one  asG > the AA- one, but the real ones are the 08- ones, with the AA- are the  DECNET > modified ones).  > K > Meanwhile, my MAC doesn't seem to be able to access the internet anymore.  BindL > seems to work on BIKE, but it too cannot resolve stuff from outside. I try toK > telnet to the router, but to no avail. So I start checking all cables. To 	 no avail.  > L > There is a serial cable into the router for backup maintenance (usually go inJ > via telnet), but VELO is down for its maintenance... Once VELO does boot and K > reconfigures itself (that took a long time with autogen etc), I manage to  goK > in, with only OPA0 operational due to lack of licences. And with the VELO D > system disk physically disconnected, I don't have access to VELO's licences...  > L > So, by the time I do get into the router, I find out it is totally screwed up. J > Random IP adreess, and when i access certain menus, it spews out garbage and  > reboots etc etc. Oh shit ! > H > The other thing is that the serial ports on VELO didn't get configured since K > the VMS system parameters (TTY_xxxxxx) are the bland vanilla ones instead  ofH > the ones I had setup on the MARINONI system disk temperary stored on a	 shelf :-(  > K > The Router allows on to use XMODEM to send the config or firmwars through  the I > serial port. (usually one would this with FTP via TPCIP, but the router  was G > totally unresponsive, even when I reconfigured BIKE to be in the same  (random)8 > subnet as what the router had decided it wanted to be. > H > Unfortnuatly, the comms apps for VELO are stored on MARINONI, which is sitting K > on a shelf... I have no access to internet to download new ones since the L > router is down.  But i eventually did find an older one, as well as a copy ofI > Y modem (VVSB).  However, the XMODEM one doesn't do CRC and besides, it  never - > was able to even nogotiate with the router.  > I > What i ended up doing is fiddle with anotherserial port's config to get H > transfers to work to my PSION PDA. Transfered the router config to it, thenK > plugged the PSION onto the router to download using XMODEM-CRC (which the J > PSION has). And voila, the router rebooted with a good enough config. OfE > course, since that one happened, I did make changes to filters etc.  Argh...  > J > I think i'll leave the 10 gig drive running while I catch some sleep and  > continue its testing tororrow. > I > Any hints on how I can easily test its ability to handle 10 gigs ? Just J > CREATE/FDL of very large files ? Woudl ANA/DISK then detect if the disk, orK > SCSI controller (DILOG 739) "wrap around" and start to write at beginning  of	 > drive ?  >  > F > Oh, and I also tried to give my 3100 a 4 gig hard drive, but it just wouldn'tJ > want to format it. Perhaps it is incomaptible, perhaps the drive is bad. Never ! > did trust micropolis that much.  > L > In the end, it is my PSION that saved the day, not only because it had theJ > ability to upload the config back to router, but also because it has all myJ > ethernet adresses for my LAN etc and that was a godsend when I was doingK > cluster_config on BIKE while VELO booted with standalone backup without a 6 > system disk to allow me to look into its parameters. > K > I am not sure I trust the stairs on my way to bedroom. With my luck, they  will break.    ------------------------------  % Date: Mon, 23 Dec 2002 17:19:49 -0400 ( From: JF Mezei <jfmezei@vl.videotron.ca>% Subject: Re: Murphy's law at its best / Message-ID: <3E077DED.864C3C66@vl.videotron.ca>    Colin Butcher wrote: >  > So where's CAMPAG then?   H Probably next disk drive I add. I could then use full name CAMPAGNOLO...   ------------------------------    Date: 24 Dec 2002 12:54:51 -0600+ From: young_r@encompasserve.org (Rob Young) 6 Subject: Of Galaxy, Infiniband and Distributed Caching3 Message-ID: <5n776H2yH2Pp@eisner.encompasserve.org>   @         Recently we heard rumors of Galaxy not making its way to?         Itanium.  This is either true or not, doesn't matter to          what might be...  A         Looking at a few wrinkles that may make Galaxy less of an          advantage.  @         Infiniband , depending on source is 2 to 10 micorsecondsA         in latency.  2+ us to send a disk sector, 3 GByte/sec per 1         direction when 12x appears in production:   P http://groups.google.com/groups?selm=39372B56.B6E8CEE1%40ibm.net&oe=UTF-8&output =gplain   J > Interesting.  2.5*10^9 bits/s.  If the signal propagates at 0.833c, thenO > each meter of wire holds 10 bits of data -- enough for one byte encoded 8-10. K > (3*10^8*0.833 = 2.5*10^8)  That's assuming that the 2.5Gb/s refers to the F > 10 and not the 8 (otherwise one byte fits in 0.8m: 10cm/bit!).  WithH > that assumption, each wire transmits one byte in 250MB/s and a 512byte; > disk sector in about 2us (plus latency in the switching).   H Yes, the 2.5 GBits per second is on the wire after encoding.  So the 250F MBytes/second per wire is correct.  And of course the speed freaks canF have 12 physical lanes for 3 GBytes/second/direction over distances ofG 10 meters or so using copper cables or 100 meters using fiber optics.      ---   C         Using Infiniband as an interconnect would deliver bandwidth F         far higher than Memory Channel with the benefits of offloadingE         the CPU.  As Jan Vorbrueggen says, something CI has had since 
         1983:   P http://groups.google.com/groups?selm=3D9A9978.B21BBD15%40mediasec.de&oe=UTF-8&ou tput=gplain   >         Regarding a good distributed cache, Bill has discussed!         on a number of occasions:   P http://groups.google.com/groups?selm=7nec6g%24p2n%241%40pyrite.mv.net&oe=UTF-8&o utput=gplain  L         Central cache versus distributed cache.  The concept of XFC residingJ         in Galactic shared memory may be less than ideal.  Unless there is8         some overlap in a COMA fashion as Bill mentions:  2 "In terms of Greg's discussion, the central sharedI cache is equivalent to *always* "having the data in the wrong place", and H the only justification for even considering it is that this place is farL less 'wrong' (by virtue of having access latency only about 3x that of local memory)"    E         But surely each node would manage portions of XFC in Galactic D         shared memory.  But then why even stick it in shared memory?E         XFC futures talk of XFC becoming much like AIX cache in using G         all available free memory (today XFC is limited to 1/2 physical C         memory) so in a sense you would handicap XFC by sticking it I         in Galactic shared memory as GSM must be configured at bootup and H         certainly would be a subset of a distributed caching scheme that+         takes advantage of all free memory.   L         As another example, RAMDISK in Galactic Shared memory.  As Bill has O         noted before , a good distributed cache will beat that everytime as you <         don't travese the driver/code paths to get the LBNs.  ;         Galaxy would be communicating through shared memory F         for clustering.  SMCI shows latency of 94 us for lock traffic.B         Certainly, Infiniband could come just as close.  Can it do.         better than MC II at 120 us?  Why not?  G         What's left?  CPUs that will migrate from instance to instance. H         Perhaps hooks to drop CPUs into firmware where others could look+         for them and grab them.  Maybe not.   ,                         New tech advantages:  K                                 1)  Infiniband - offloads CPU, low latency/ E                                      high, approaching or better than K                                      shared memory.  Also eliminates memory >                                      traffic for messaging/IO.  J                                 2)   Distributed cache.  Take advantage ofI                                         all memory.  Smarts would migrate @                                         pages in a COMA fashion.  )                         Galaxy advantage:   1                                 1)  Migrate CPUs.     9         Offloading the CPU may be the greatest advantage.   F         Maybe the way things are going migrating CPUs isn't such a bigE         deal.  Sun marketing shows us what we know about Galaxy, i.e. J         "your CRM application is becoming saturated.. with Dynamic DomainsF         you can grab CPU resources."   We know you have to have enoughH         horsepower for peaks.  Even if you have to rent it and return itM         as many Web sites do on Superbowl Sunday.  Otherwise, you are turning G         away customers.  Migrating CPUs is an advantage.  The niche may H         be narrow as more applications will take advantage of many nodes-         in an Oracle Parallel Server fashion.   G         Galaxy may be coming to Itanium, it is on roadmaps.  That said, C         a number of its great advantages could be further narrowed.   B         It wouldn't be a stretch to see large boxes of the future F         hardware partitioned deliberately for the majority of the OSesC         including VMS, another big advantage there is availability.   #                                 Rob     B Men with walkie-talkie                  I'm home again to you babeC Men with flashlights waving             You know it makes me wonder G Up upon the tower                       Sittin' in the quiet slipstream > The clock reads daylight savin'         Rollin' in the thunder  -                                 -- Neil Young    ------------------------------  % Date: Mon, 23 Dec 2002 15:22:19 -0400 ( From: JF Mezei <jfmezei@vl.videotron.ca># Subject: Re: OpenVMS Branding Query / Message-ID: <3E07626A.6E3B6406@vl.videotron.ca>    Hans Vlems wrote:  > 9 > The DECWindows logo changed to Compaq with VAX/VMS 7.3.   G Isn't that ironic ... Didn't 7.3 get released at a time where Curly had ; already announced his intentions to sell himself to Carly ?n   ------------------------------  % Date: Tue, 24 Dec 2002 01:43:38 -0400e( From: JF Mezei <jfmezei@vl.videotron.ca>6 Subject: Re: OT: Whoa! Is Sun aiming at VMS's jugular?/ Message-ID: <3E07F3FE.F1E9E332@vl.videotron.ca>I  * Alan Winston - SSRL Admin Cmptg Mgr wrote:N  I'm not arguing against a Hammer port; I was specifically arguing against theK > claim that QNX running on IA32 meant that VMS could (easily) run on IA32.   N I have yet to see a VMS engineer claim that VMS could NOT be ported to today'sL 8086. Today,s 8086 supports virtual memory, supports 4 protection modes etc.L You have to give it to Intel: they have been able to transform the 8086 game, controller into a somewhat respectable chip.  G Considering that they still maintain VAX-VMS, woudl generating a 32 bitj? version from the 64 bit code-base be such a huge undertaking ?    M I think that VMS should be ported to Hammer or the 64 bit version of the 8086lN whenever Intel announces it.  Once Hammer is out (or Intel releases its 64 bitM 8086), then it will be pointless to continue any work on IA64, and of course, 4 pointless to think about porting to the 32 bit 8086.   ------------------------------  % Date: Mon, 23 Dec 2002 13:29:21 +0100 5 From: "Chris Clifford" <chris.clifford@openvms.co.uk> 6 Subject: Re: OT: Whoa! Is Sun aiming at VMS's jugular?* Message-ID: <3e0701bc@news.swissonline.ch>  K "Andrew Harrison SUNUK Consultancy" <Andrew_No.Harrison_No@nospamn.sun.com>i9 wrote in message news:3E06E3C5.9020507@nospamn.sun.com...H >  >  > Chris Clifford wrote:d' > > "Andrew Harrison SUNUK Consultancy" ' <Andrew_No.Harrison_No@nospamn.sun.com> = > > wrote in message news:3E008710.9060803@nospamn.sun.com...- > >- > >-E > >>SunFire Link is currently available for the F6800 and the F15000.  > >>L > >>The SunFire Link card replaces an I/O boat on either system. It connectsI > >>directly into the Sun Fire backplane, it isn't a PCI based system for2 > >>example. > >>F > >>Because of the direct connection and because of the performance ofB > >>the interconnect one card will give you up to 2.8 GB/s, with 2I > >>links, 2 cards which is the normal config in a cluster for resiliance 4 > >>with link level striping will give you 4.8 GB/s. > >>D > >>It supports direct node to node connections for up to 3 nodes or3 > >>via a switch which currently gives you 8 nodes., > >> > >  > > <blah blah blah> > >e > > Three points:I > >tJ > > 1. This Sun Fire Link is only of use if your cluster does not span any > > reasonable distance. >o< > Of course but then its orders of magnitude faster than any5 > OpenVMS cluster interconnect so cut it some leaway.-  L Haven't seen it yet but not doubt it'll turn up soon enough - we've loads of$ 6800s so it's only a matter of time.   >r9 > It isn't designed for long distances, it is designed toD8 > allow a relatively small number of very large nodes to > scale in a cluster.H >a9 > You blah, blah response to what is currently one of theo6 > fastest intersystem interconnects available tends to- > suggest that you should get out a bit more.t  K Fastest doesn't mean most reliable. And you're comparing a garden shed with?I a 4-storey house. With Sun Cluster being the former of course. You'll gettI around very quickly in a garden shed but you'll be able to do a hell of ay lot more in a 4-storey house.s   > K > > 2. Sun Cluster only runs with any reasonable performance if you give it K > > dedicated private links and if we're talking ethernet, you've got to goD forrF > > Gigabit. You don't necessarily need as fast interconnects with VMS (we'reL > > using 100Mb ethernet as the interconnect within our two mission-critical& > > clusters, each spanning 11 miles). >mC > You also don't necessarely need fast interconnects for SunClusterDC > either, it depends entirely on what you are intending to acheive.e  K I wonder then why we've had to install a private gigabit network to get any  half-decent performance.   >w< > If you want to run 9i RAC and get it to scale then aim for: > the fastest interconnect you can afford or is available,= > it will pay cost dividends if it improves your scalability.s >h% > This applies to Solaris or OpenVMS.'  I Not yet on RAC on either platform - will provide the truth once we get ita" running, not just some sales talk.   >u: > If you want a failover cluster then slower interconnects > work.e  L In VMS a cluster is a cluster. You can use it for failover or a multitude ofJ other purposes without any special configuration. Our 100Mb ethernet-based9 clusters are mission critical and they perform perfectly.i   >u >a >AC > > 3. Fibrechannel's only used as a storage interconnect with VMS.w > >hI > > Speaking from experience here. I've never seen such a group of amazedc and F > > overjoyed people as Sun when they got 'clustering' working over 11 miles.I > > And that's clustering in a relatively loose sense of the word, albeito better* > > than previous versions of Sun Cluster. > >n >aG > Really I have just come back from our manufacturing plant in Scotland-E > the test lab there has a 100 KM cluster running which the engineersm> > get to use. So I rather doubt that anyone was very suprised.  G Oh believe me they were pretty astonished. If 100km's been managed thentG great - I can happily approach our Sun sales guy for details in the newg year.    ------------------------------    Date: 24 Dec 2002 06:43:28 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) 6 Subject: Re: OT: Whoa! Is Sun aiming at VMS's jugular?3 Message-ID: <fK9NTsUreUr$@eisner.encompasserve.org>-   In article <00A18E38.86A165BC@SSRL.SLAC.STANFORD.EDU>, winston@SSRL.SLAC.STANFORD.EDU ("Alan Winston - SSRL Admin Cmptg Mgr") writes:pZ > In article <SACN9.2547$I64.1400@news.get2net.dk>, "Dr. Dweeb" <Dweeb@nospam.com> writes:N >>There was a BLISS-IA32 compiler for years.  COMPAQ killed it along with lotsK >>of other stuff during the preliminary phases of the kill Alpha procedure.  > E > Which managed to whack the NT version of Rdb as collateral damage. uM > (Although I really thought the Rdb group could have afforded to pick up onerI > guy to maintain the IA32 Bliss compiler, they didn't seem to think so.)t  F But that compiler is still used for the former DEC (now Intel) Fortran compiler on Windows, right ?   ------------------------------  % Date: Tue, 24 Dec 2002 13:59:47 +0000,' From: Andrew Harrison SUNUK Consultancya6 Subject: Re: OT: Whoa! Is Sun aiming at VMS's jugular?. Message-ID: <3E086853.2090506@nospamn.sun.com>   Greg Cagle wrote: * > Andrew Harrison SUNUK Consultancy wrote: >  >> >> >> Bob Koehler wrote:e >>G >>> In article <3E03387C.40808@nospamn.sun.com>, Andrew Harrison SUNUK   >>> Consultancy writes:i >>>n5 >>>> Or alternatively sometime you will realise that:  >>>>8 >>>> SPECfp has almost no relevance to the vast majority? >>>> of commercial apps (the buld of what runs on Alpha's/Suns)s >>>> because they don't use FP.b >>>c >>>> >>>a >>>a? >>>    What I realise is that you live in a tiny world of a fewe >>>    commercial apps.  >>>c >> >>: >> I have worked in Telco, Retail Government and FinancialB >> services accounts and have worked with many of the applications >> vendors in those markets. >>9 >> If you want I can list the apps I have worked with but-9 >> it seems a pretty pointless excercise since your pointi >> is hardly valid is it.g >  > 9 > So you have no experience with scientific, engineering,p0 > simulation, or design applications, then? Hmm. >   2 Yes, I did quite a lot of work with Redifusion and2 Singer. Plus Ford and Rover. I have worked for Sun/ for 15 years, nearly 16 and we were origionally # an engineering workstation company.   2 And you are right they do have systems that are FP- dependant, but even there if you take out thek- engineering desktops (not really a market for + OpenVMS anyway) and just concentrate on the , servers in something like Auto manufacturing+ they still end up with more systems runninge7 integer commercial apps than FP technical applications.m  8 The HPTC market is much much smaller than the commercial5 server market, numbers I have seen put it at 5-10% oft the total server market.  3 Ironically even in HPTC the GS series don't do welln/ because they are too big, too expensive and tool/ slow. If people are using Alphas in this markety, then they are mostly using ES40/45's running0 Tru64 clustered. There arn't to my knowledge any. GS based systems or clusters of systems in the0 top500 list for example. On the other hand there, are clusters of large Sun's, HP's and IBM's.   Regardsh Andrew Harrisona   ------------------------------    Date: 24 Dec 2002 14:28:57 -00004 From: Doc.Cypher <Use-Author-Address-Header@[127.1]>6 Subject: Re: OT: Whoa! Is Sun aiming at VMS's jugular?5 Message-ID: <20021224142857.5405.qmail@nym.alias.net>e  6 On Tue, 24 Dec 2002, Andrew Harrison SUNUK Consultancy. <Andrew_No.Harrison_No@nospamn.sun.com> wrote: >Greg Cagle wrote:  : >> So you have no experience with scientific, engineering,1 >> simulation, or design applications, then? Hmm.m  3 >Yes, I did quite a lot of work with Redifusion andt3 >Singer. Plus Ford and Rover. I have worked for Sunn0 >for 15 years, nearly 16 and we were origionally$ >an engineering workstation company. >n3 >And you are right they do have systems that are FPl. >dependant, but even there if you take out the. >engineering desktops (not really a market for, >OpenVMS anyway) and just concentrate on the- >servers in something like Auto manufacturing , >they still end up with more systems running8 >integer commercial apps than FP technical applications.  E Ah, that would be the installation of that well-known resource-hungrye device,t  	 /dev/SAP.'  @ Apparently it behaves similarly to /dev/null in terms of output.     Doc. --  : Time and money, the psychotropics of the business world...K ~ VAXman                                             https://vmsbox.cjb.nett   ------------------------------  % Date: Tue, 24 Dec 2002 15:57:10 +0100D$ From: "Dr. Dweeb" <Dweeb@nospam.com>6 Subject: Re: OT: Whoa! Is Sun aiming at VMS's jugular?/ Message-ID: <eB_N9.39$M05.1433@news.get2net.dk>   , "Alan Winston - SSRL Admin Cmptg Mgr" wrote:? > In article <SACN9.2547$I64.1400@news.get2net.dk>, "Dr. Dweeb"  > <Dweeb@nospam.com> writes:E >> There was a BLISS-IA32 compiler for years.  COMPAQ killed it alongiE >> with lots of other stuff during the preliminary phases of the kill  >> Alpha procedure.s >WD > Which managed to whack the NT version of Rdb as collateral damage.F > (Although I really thought the Rdb group could have afforded to pickE > up one guy to maintain the IA32 Bliss compiler, they didn't seem tor > think so.) >e	 > -- Alang  G They did not even need to do that.  They already had a couple of ex-DECDL BLISS compiler guys to do the work.  CompaQ wanted an exhorbitant amount for, the code & rights - US$ 7 figures I believe.  K Orcale though that that was extortion and decided not to play.  I think the L CompaQ - Oracle relationship was pretty sour for quite a while.  It probably> was right until the end of CompaQ.  No idea how it is with HP.   Dweeb    ------------------------------  % Date: Tue, 24 Dec 2002 08:20:46 +0100e From: Dirk Munk <munk@home.nl>; Subject: Re: SAN recommendations mixing VMS, Solaris, Linuxo8 Message-ID: <ha2g0vsgil59poojok7qrbbuonddh9ej90@4ax.com>  E On Tue, 24 Dec 2002 00:31:22 -0500, "Main, Kerry" <Kerry.Main@hp.com>  wrote:   >Carl, >i2 >You might be interested in this SAN design guide:I >ftp://ftp.compaq.com/pub/products/storageworks/techdoc/san/AA-RMPNE-TE.pa >dfe >o' >Reference pg. 6-13 (note SGI included)oD >"ACS works in a heterogeneous environment that includes Tru64 UNIX,D >OpenVMS, Microsoft Windows NT and Windows 2000, Novell NetWare, SunD >Solaris, HP-UX, SGI IRIX, IBM AIX, Linux x86, and Linux Alpha. This- >application is at the storage system level."o  F You're right Kerry, but the ACS software is for the HSG80, not for theD EVA. And the EVA is a much more sofisticated  product with much moreE performance. In this document the EVA software is not listed as beings< compatible with SGI. However new firmware versions with more0 possibilities appear quite reguarly for the EVA.   >t	 >Regards,  >i >Kerry Maino >Senior Consultant >Hewlett-Packard (Canada) Co.o" >Consulting & Integration Services >Voice: 613-592-4660 >Fax   : 613-591-4477  >Email: kerryDOTmain@hpDOTcom . >    (remove the DOT's and replace with "."'s) >B >i >a >-----Original Message----- = >From: Carl Karcher [mailto:karcher@thuria.waisman.wisc.edu] h  >Sent: December 23, 2002 2:35 PM >To: Info-VAX@Mvb.Saic.Com8 >Subject: SAN recommendations mixing VMS, Solaris, Linux >s > H >We would like to implement a SAN (4 TB initially with growth to 100 TB) >with the following goals: >o< >1. Support the following platforms as hosts within the SAN:F >   Sun/Solaris, Sgi/Irix [1], Linux, Windows 2000 [2] and HP Openvms. >iC >2. Support a SAN shared file system between multiple heterogeneous I >   clients, Solaris/SPARC and Linux/IA32 currently, as well as serve thesD >   same files out using NFS and CIFS. The filesystem should supportG >   large volumes and files, and be dynamically expandable. The abilitybE >   to add failover clustering and Hierarchical Storage Management ist >   also highly desirable. >e/ >   Possible shared file system candidates are:a; >   SGI CXFS (would require supporting SGI/Irix on the SAN). >   ADIC StorNext FS >   Sun QFS + Tivoli SANergy >- >Notes:- >-E >   [1] Sgi/Irix is only necessary if CXFS is chosen as a SAN shared b >       file system. > / >   [2] Windows is not essential but desirable.g >DH >I've been hoping HP's EVA would fit the bill however it doesn't supportH >SGI. So we are looking for a shared unix file system that the EVA wouldA >support between Solaris and Linux. As HP is closed this week I'm : >wondering if anyone else has suggestions/recommendations.   ------------------------------  % Date: Mon, 23 Dec 2002 15:33:41 -0500s' From: Stuart Fuller <stufuller@usa.net>i% Subject: Re: SCSI controller questionp/ Message-ID: <5vr7ua.in4.ln@dadsys.fuller.local>-  
 mhr wrote:K > Intraserver controllers need to be the VMS-capable type, ie., a couple ofoH > roms added to the board-these of course are the most expensive roms inJ > history. Note the controller will work with the various linux as well as > Tru64c > mhr / > "Hans Vlems" <hvlems@iae.nl> wrote in messager1 > news:au7lvl$5375e$1@ID-143435.news.dfncis.de...iH >> My new Alphaserver 5000 (not too dissimilar from an Alpha 1200) got aG >> SCSI controller today. It is a device with on-board ethernet and twoPJ >> Ultra SCSI ports, available as external outlets or as internal ports. I >> use boths > portssD >> internally. The console SHOW DEV command lists all 5 disks on the > controller, >> and passes a diagnostic test (TEST PCI1).) >> But VMS boots with the error messages:U >>8 >> SCSi chip is SYM53C875 Operating Mode is SE UltraSCSI >> ROM Checksum read error >>K >> for both the PKB0 and PKC0 devices and they stay off line. The disks area' >> (obviously) not available under VMS.t* >> VMS SHOW DEV PkB0 shows: KZPCM off lineK >> I have tried the controller on pci0 and pci1, both cases resulted in the  >> same error message. >>1 >> The onboard ethernet interface works properly.  >>% >> Any idea what causes this problem?c >> >> Hansc >>  C I believe that the module also needs these ROMs to work with TRU64.r           Stu>   ------------------------------  # Date: Mon, 23 Dec 2002 21:19:43 GMTe= From: "Colin Butcher" <colinDOT.butcherAT@xdeltaDOT.coDOT.uk>a% Subject: Re: SCSI controller questiono; Message-ID: <P5LN9.325$051.1998074@news-text.cableinet.net>   F You need the version of the board with the minimum acceptable firmware version for VMS.  K Why more than one version? The VMS one is properly tested. Other OS's don'ty+ necessarily care so much about your data...    -- Hope this helps. Cheers, Colin. ' (colinDOT.butcherAT@xdeltaDOT.coDOT.uk)o    - "Hans Vlems" <hvlems@iae.nl> wrote in messagel/ news:au7lvl$5375e$1@ID-143435.news.dfncis.de... L > My new Alphaserver 5000 (not too dissimilar from an Alpha 1200) got a SCSIL > controller today. It is a device with on-board ethernet and two Ultra SCSIG > ports, available as external outlets or as internal ports. I use botht ports C > internally. The console SHOW DEV command lists all 5 disks on theg
 controller+ > and passes a diagnostic test (TEST PCI1).l( > But VMS boots with the error messages: >p7 > SCSi chip is SYM53C875 Operating Mode is SE UltraSCSIu > ROM Checksum read error0 > J > for both the PKB0 and PKC0 devices and they stay off line. The disks are& > (obviously) not available under VMS.) > VMS SHOW DEV PkB0 shows: KZPCM off lineoJ > I have tried the controller on pci0 and pci1, both cases resulted in the > same error message.t >w0 > The onboard ethernet interface works properly. > $ > Any idea what causes this problem? >f > Hans >l   ------------------------------  # Date: Tue, 24 Dec 2002 18:57:08 GMTi- From: "John E. Malmberg" <wb8tyw@qsl.network>y0 Subject: Re: Semaphores and Shared memory on VMS= Message-ID: <862O9.36686$VA5.5745003@news1.news.adelphia.net>c   Lyndon Bartels wrote:o  / > I'm working on the port of PostgreSQL to VMS.s > F > It needs Semaphore operations, as well as Shared memory operations. 7 > (sem_wait, sem_init, sem_close, shmctl, shmget, etc.)r > D > Has anybody down a "port", "copy", "clone" or whatever you want to > call it to VMS?  > : > I've been looking around and haven't found anything yet.  4 The only one that I am aware of, does not work well.  7 A reference on what these functions do can be found at n http://www.opengroup.org/f  I You need to register (free) to access the site, and then you can look up n, the official definitions of these functions.    G sem_init() appears to map to the OpenVMS function of SYS$ENQW(), and a a few other functions.  G sem_wait() also appears to map to SYS$ENQW() Convert lock function, it ,E must also be able to be interrupted by a signal.  PostgreSQL may not  2 require the ability to be interrupted by a signal.  B sem_close appears to clean up everything allocated by sem_init(), C including deallocating the lock.  This cleanup only takes place if o0 sem_unlink() has been called after a sem_open().    C The Shared memory functions either seem to be able to be mapped to  C global sections, or lock value blocks, or a combination of the two.p  G If I were doing this, I would try to see if lock value blocks could be -" used for the shared memory values.   -John  wb8tyw@qsl.network Personal Opinion Onlyn   ------------------------------  # Date: Mon, 23 Dec 2002 20:22:08 GMTe9 From: Alan Adams <alan.adams@orchard-way.freeserve.co.uk>t Subject: Re: UTC time in DCL? Message-ID: <df5581a94b.Alan.Adams@orchard-way.freeserve.co.uk>i  7 In message <4oe80vk15682ief7hrcn3pff87800vkdoo@4ax.com> )           Dirk Munk <munk@home.nl> wrote:o  G > One of the problems I encounter in DCL is that it is only possible towF > get the local time from F$TIME, and only in the shape of dd-mon-yyyy > hh:mm:ss.tt .a > B > Of course I know it is possible to use F$CVTIME, but that is notB > enough. I would like to be able to extract UTC time for example,@ > without having to write my own calculations. It is possible toE > calculate the UTC time, but the resulting syntax will be yyyy-mm-dd F > etc.. Since F$CVTIME is a one-way routine it is rather cumbersome to( > get it back in the dd-mon-yyyy syntax.  f I may have misunderstood the question, but I think you need to look into the second parameter to F$CVT  G The parameter can be "ABSOLUTE" or "COMPARISON", maybe some others. ThekA default is "COMPARISON" which is indeed yyyy-mm-dd. Try absolute.g   > G > Other time syntaxes would be nice too, for instance the syntaxes thatn  > are in use with SMTP and NCL.  > E > To start with, maybe it is possible to extend the F$TIME routine tog > F$TIME("UTC") ?  > 6 > Maybe Hoff can give us his thoughts on this matter ?   -- e
 Alan Adams& alan.adams@orchard-way.freeserve.co.uk http://www.nckc.org.uk/    ------------------------------  + Date: Tue, 24 Dec 2002 11:56:14 +0000 (UTC)t" From: greenaum@BOLLOCKSyahoo.co.uk& Subject: Re: vax6k.openecs.org rebirth3 Message-ID: <3e0f265c.7463269@news.btopenworld.com>t  7 On Sun, 22 Dec 2002 20:16:37 -0600, "David J. Dachtera"r! <djesys.nospam@fsi.net> sprachen:i  H >Beyond that, unless you've sworn a blood-oath to keep VAX-only alive inG >this way, I'd shop around for used Alpha 2100s, 4100s, used CIPCAs andM0 >other CI gear. Alphas are a LOT cheaper to run.  F I dunno what a VAX 6K is, but I can only assume it's some old, classic? computer, and it's being run for sortof nostalgia and enjoymenti5 reasons, not as a commercial server for some purpose.g  C Even if, technically, he could run it on a PC with an emulator, I'mb? sure there's lots of pleasure in knowing your bytes are runningtB through some old transistors or something. Actually there probably8 aren't any transistorised computers than can run Telnet, unfortunately.  = Tho there was that guy who rigged an Atari 2600 up to a videotF digitiser and ran the joysticks from the parallel port. That wuz k3wl!    H >> There are about 50 registered users on the vax6k.openecs.org system.  >tD >Perhaps a monthly subscription to cover costs would be appropriate.  D Yup, that's what I'd do. 2 Euros a month is nothing to anybody, it's< less than you generally have in your pocket at any one time.  E The problem with subs, is getting people to actually walk to the posteA box, and in arranging some sort of international cheque. Maybe heoF could take yearly subs, and get people to risk posting cash. Or open a, US bank account, or use Paypal or something.  C There STILL isn't a small-change Internet currency! I can't believe E that! I'd be glad to, say, buy a card from a shop, like those callinghC cards, with say 10 quid on it, and then use a number printed on therC card to buy things. There must be squillions to be made in startings your own currency. t  : It'd just need existing strong encryption. And credit-cardC verification manages with an incredibly weak verification algorithmO (the Luhn algo).  ? Perhaps some big banks could do it, in association with Visa or E something. Mobile phone companies manage to sell millions of pre-paid C vouchers, probably 80% of the British mobile phone market runs thato way.    D Visa make tons of money, but the system I want wouldn't even be thatF complicated. Everything already exists to support it. I just dunno why nobody's done it.   D Maybe you could sell your vouchers, and then use one of the existingE merchant cash-moving systems to actually pay sellers for their goods.aE The pr0n market's already got virtual card-schemes doing well. All it D needs, is for some way of paying cash to buy small amounts of creditD in local shops. Oh, and the vast cash reserve that'd give you, would) make you very rich on the interest alone!o  C Hmmm, maybe I should apply for a small business grant. I reckon 100o% million would cover my startup costs.t  H ------------------------------------------------------------------------  H if love is a drug, then, ideally, it's a healing, healthful drug... it's@ kind of like prozac is supposed to work (without the sexual side5 effects and long-term damage to the brain and psyche)    ------------------------------  # Date: Mon, 23 Dec 2002 21:22:02 GMTm) From: Charles Richmond <richmond@ev1.net>s& Subject: Re: vax6k.openecs.org rebirth' Message-ID: <3E079A67.CC4B01EA@ev1.net>    "David J. Dachtera" wrote: > - >         [snip...]    [snip...]     [snip..]p > E > Perhaps a monthly subscription to cover costs would be appropriate.I > < How about signing up corporate sponsors, or selling web site advertising space???   -- w? +-------------------------------------------------------------+ ? |     Charles and Francis Richmond     <richmond@plano.net>   |e? +-------------------------------------------------------------+B   ------------------------------  % Date: Tue, 24 Dec 2002 12:49:37 -0600t1 From: "David J. Dachtera" <djesys.nospam@fsi.net>f& Subject: Re: vax6k.openecs.org rebirth' Message-ID: <3E08AC41.3B6716EE@fsi.net>e  # greenaum@BOLLOCKSyahoo.co.uk wrote:c > [snip] > Actually there probablyt: > aren't any transistorised computers than can run Telnet, > unfortunately.  F Assuming you mean those little three-legged metal-can things about theD size of a pencil eraser soldered to the circuit boards found in 60'sH vintage consumer electronics and such, yes, I agree. Otherwise, I'm sureB we've all read about how many million transistors are found in CPU chips, ASICs, etc.   --   David J. Dachterah dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/,   ------------------------------  # Date: Tue, 24 Dec 2002 14:30:34 GMTt- From: bdc@world.std.com (Brian 'Jarai' Chase)p; Subject: Re: VMS has got apps ... 3300 and counting per HP!w& Message-ID: <H7MoAy.8vn@world.std.com>  = In article <d7791aa1.0212231850.7cef5f62@posting.google.com>,S) Bob Ceculski <bob@instantwhip.com> wrote:r  G > [...] The current OpenVMS apps base is ~3,300 apps, 100 of which werevG > added in the past year. An additional ~15 apps are being added to theb > portfolio each week.  J Post proof or retract!  (Sorry, it's a common quip from another newgroup).G So obviously HP have a list of these 3300 apps if they're able to countt, them; do you think they could let us see it?   -brian.  -- lF --- Brian Chase | bdc@world.std.com | http://world.std.com/~bdc/ ----->       IBM 1130 Club Motto:  "Computer People Are Happy People"   ------------------------------    Date: 24 Dec 2002 09:34:00 -0500- From: Ken Robinson <kenrbnsn1@rcn.com.nospam>n; Subject: Re: VMS has got apps ... 3300 and counting per HP! ; Message-ID: <Xns92EE639C7D78Aksrobinerenjcom@66.150.105.98>-  0 bdc@world.std.com (Brian 'Jarai' Chase) wrote in news:H7MoAy.8vn@world.std.com: f  ? > In article <d7791aa1.0212231850.7cef5f62@posting.google.com>,s+ > Bob Ceculski <bob@instantwhip.com> wrote:  > H >> [...] The current OpenVMS apps base is ~3,300 apps, 100 of which wereH >> added in the past year. An additional ~15 apps are being added to the >> portfolio each week.D > A > Post proof or retract!  (Sorry, it's a common quip from anotheruF > newgroup). So obviously HP have a list of these 3300 apps if they're< > able to count them; do you think they could let us see it?    / Easy enough to find on the OpenVMS Web site at nD <http://www.openvms.compaq.com/solutions/matrix/solution_cur_a.html>  5 Although you might have to count them yourself... :-)k   Ken Robinson   ------------------------------  % Date: Tue, 24 Dec 2002 11:49:25 -0500o$ From: "rob kas" <rob@netcarrier.net>; Subject: Re: VMS has got apps ... 3300 and counting per HP!t3 Message-ID: <3e089015$0$1401$8e9e3842@news.atx.net>.  5 "Bob Ceculski" <bob@instantwhip.com> wrote in message-7 news:d7791aa1.0212231850.7cef5f62@posting.google.com...f> > this is from skhpc on the openvms.org site ... for those who > say vms has no apps ...h            /  Microsoft plunks $500 million down on partnersa  G http://www.infoworld.com/articles/hn/xml/02/07/12/020712hnmspartner.xml   J update  MICROSOFT IS KICKING off its Fusion 2002 conference in Los AngelesI with a bang Friday, heralding plans to infuse its partner programs with an cool $500 million.  J The investment, which will be rolled out in phases over two years startingJ this month, represents a 50 percent increase over last year's annual spendE on partner development and support, according to Microsoft officials.0  K The money will be allocated across three areas, including hiring more field G reps to work directly with Microsoft partners; developing new Web-based J training tools and the eLearning Center; and boosting the amount and types of partner support.r  E Microsoft plans to allot 20 percent of the money toward the headcountpE increases, including a $4 million shot in the arm for partner accountpI manager training. Close to 30 percent of the investment will be sunk intodL Web-based training, including the eLearning Center, which features 11 online courses to date.  G Other areas being funded are private newsgroups for partners and betterr phone support, officials said.  J Trying to drive more partners to its Web site makes sense for Microsoft asJ it attempts to manage relationships with more than 810,000 partners, whichI run the gamut from the industry's top companies to much smaller entities,sL according to Dwight Davis, vice president of Summit Strategies, in Kirkland, Wash.r   ------------------------------  # Date: Tue, 24 Dec 2002 16:58:27 GMTs" From:   VAXman-  @SendSpamHere.ORG; Subject: Re: VMS has got apps ... 3300 and counting per HP!t0 Message-ID: <00A18ECC.3CC275CB@SendSpamHere.ORG>  k In article <Xns92EE639C7D78Aksrobinerenjcom@66.150.105.98>, Ken Robinson <kenrbnsn1@rcn.com.nospam> writes:s1 >bdc@world.std.com (Brian 'Jarai' Chase) wrote ini  >news:H7MoAy.8vn@world.std.com:  >e@ >> In article <d7791aa1.0212231850.7cef5f62@posting.google.com>,, >> Bob Ceculski <bob@instantwhip.com> wrote: >> eI >>> [...] The current OpenVMS apps base is ~3,300 apps, 100 of which wereiI >>> added in the past year. An additional ~15 apps are being added to thea >>> portfolio each week. >> cB >> Post proof or retract!  (Sorry, it's a common quip from anotherG >> newgroup). So obviously HP have a list of these 3300 apps if they're = >> able to count them; do you think they could let us see it?o >n >t0 >Easy enough to find on the OpenVMS Web site at E ><http://www.openvms.compaq.com/solutions/matrix/solution_cur_a.html>f >b6 >Although you might have to count them yourself... :-)  H There are at least a dozen products that I have my hands in that are notH listed.  My guess is that for every dozen not listed there are others onH that list that don't belong there because they are nolonger supported on VMS.    --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMi            o5   "Well my son, life is like a beanstalk, isn't it?"     ------------------------------  # Date: Tue, 24 Dec 2002 18:13:45 GMT # From: "John Smith" <a@nonymous.com>m; Subject: Re: VMS has got apps ... 3300 and counting per HP!iI Message-ID: <tt1O9.62143$F2h1.54746@news01.bloor.is.net.cable.rogers.com>a  / "rob kas" <rob@netcarrier.net> wrote in messagel- news:3e089015$0$1401$8e9e3842@news.atx.net... 2 >>  Microsoft plunks $500 million down on partners >  > F http://www.infoworld.com/articles/hn/xml/02/07/12/020712hnmspartner.xm l  >rD > update  MICROSOFT IS KICKING off its Fusion 2002 conference in Los Angeles D > with a bang Friday, heralding plans to infuse its partner programs with a > cool $500 million. >sC > The investment, which will be rolled out in phases over two yearss startingF > this month, represents a 50 percent increase over last year's annual spendw< > on partner development and support, according to Microsoft
 officials. >rB > The money will be allocated across three areas, including hiring
 more field? > reps to work directly with Microsoft partners; developing newa	 Web-based.F > training tools and the eLearning Center; and boosting the amount and types  > of partner support.a >r= > Microsoft plans to allot 20 percent of the money toward thet	 headcounti? > increases, including a $4 million shot in the arm for partnera accountuF > manager training. Close to 30 percent of the investment will be sunk intoD > Web-based training, including the eLearning Center, which features	 11 onlineh > courses to date. >nB > Other areas being funded are private newsgroups for partners and better  > phone support, officials said. >-? > Trying to drive more partners to its Web site makes sense forh Microsoft asF > it attempts to manage relationships with more than 810,000 partners, which A > run the gamut from the industry's top companies to much smaller 	 entities,lD > according to Dwight Davis, vice president of Summit Strategies, in	 Kirkland,- > Wash.-    D Versus the virtually $0.00 CHumPaq spends with 3rd party application developers for VMS.    ------------------------------    Date: 24 Dec 2002 10:47:53 -0800- From: jodonnell@hrblock.com (Jason O'Donnell)h; Subject: Re: VMS has got apps ... 3300 and counting per HP!'< Message-ID: <9059bf6b.0212241047.11b8f9f@posting.google.com>  ` "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message news:<3E07E8F2.2BC11A72@fsi.net>...% > Bob Ceculski wrote (quoting skhpc):0
 > > [snip] > > The<H > > current OpenVMS apps base is ~3,300 apps, 100 of which were added inJ > > the past year. An additional ~15 apps are being added to the portfolio > > each week. > : > ...but nobody knows about them due to stealth marketing.  C Yessireee-bob, we have HP commercials that do not name any products B and we have HP commercials that name their Compaq brand of Windoze PCs.  0 Where the $$#^%$^&% are the OpenVMS commercials?   ------------------------------  # Date: Mon, 23 Dec 2002 20:27:05 GMTa9 From: Alan Adams <alan.adams@orchard-way.freeserve.co.uk>eO Subject: Re: Why does BACKUP/IMAGE need write access to INDEXF.SYS, BITMAP.SYS? ? Message-ID: <e7c981a94b.Alan.Adams@orchard-way.freeserve.co.uk>t  < In message <b096a4ee.0212201422.504ac8c7@posting.google.com>9           spamsink2001@yahoo.com (Alan E. Feldman) wrote:t  g > Jeff Cameron <JCam90502@jcameron.com> wrote in message news:<BA21D1B4.246E%JCam90502@jcameron.com>...l > [...]tL > > I believe what John has said is correct. The INDEXF.SYS file is open forN > > write access as to prevent any changes to the file system while the backupN > > image makes a complete mapping of the file system in it's internal tables. > [...]R > G > So what happens if some files are deleted or added to the disk duringsG > the writing-to-the-save-set phase of the save operation? What happensnG > to those files as regards the backup save set and what error messages. > would one get, if any?    Files added are silently missed.  = Files deleted result in "file not found" messages in the log.r   > < > And why does BACKUP/IMAGE need write access to BITMAP.SYS? > C > And why does BACKUP/FAST need write access to INDEXF.SYS, but not 
 > BITMAP.SYS?u >  > Thanks again.c >  > Alan E. Feldmanr > spamsink2001 yahoo com > ? > [read the following quote slowly in a moderately deep voice:]2 > E > Benjamin Franklin: <pause> The only president of the United States,hC > <pause> who never was president of the United States.  --Firesign. > Theater (IIRC)   -- ,
 Alan Adams& alan.adams@orchard-way.freeserve.co.uk http://www.nckc.org.uk/n   ------------------------------  # Date: Mon, 23 Dec 2002 20:29:27 GMTd9 From: Alan Adams <alan.adams@orchard-way.freeserve.co.uk>hI Subject: Re: Your Multi-volume Tape Backups may be bad on all versions ofe? Message-ID: <5d0182a94b.Alan.Adams@orchard-way.freeserve.co.uk>u  ( In message <maTbzdUKtXRA@elias.decus.ch>4           p_sture@elias.decus.ch (Paul Sture) wrote:  b > In article <BA2409AD.253C%JCam90502@jcameron.com>, Jeff Cameron <JCam90502@jcameron.com> writes: >  > <snip> >  > > O > > I have asked the VMS Engineer who is responsible for correcting the code in K > > BACKUP.EXE, to provide me with corroborative references to this problemaL > > which will be fixed in the up coming patch/update. A previously reportedM > > problem was reported by another customer, which is attributed to the samee. > > piece of code in BACKUP, can be found at : > > P > > http://h18000.www1.hp.com/support/asktima/operating_systems/CTI_SRC991104002 > > 366.html > >  > J > Thanks for the reference. It _appears_ that we should be safe from this, > since: > H > a) we do not (to my knowlege, I'd best ask the production team though)F > have any volumes large enough that a backup/image would span 3 tapes > (DLT IV, with compression) > C > b) all our backups specify the tape label names, using the syntaxg' > /LABEL=(LABEL1,LABEL2,LABEL3,LABEL4).i > E > Something deep down is telling me that there was a reason that many H > years ago I chose to use /LABEL to override BACKUP's default labelling+ > mechanism, but I cannot remember details.e  - Two possibilities immediately sprint to mind.h  9 LABEL1 would by default be followed by LABE02 LABE03 etc.e  J putting label1 back in could result in it being overwritten, since it is a9 valid continuation label. (Use expiration dates as well.)    > H > On the subject of MME, I see it mentioned in passing in the BACKUP APID > documentation, and also in the I/O manual but nowhere do I find an1 > adequate description of it. What exactly is it?S > B > Thanks to the reorganization of the documentation site, I cannot? > provide more details right now. Time to reach for the CDs :-(    -- v
 Alan Adams& alan.adams@orchard-way.freeserve.co.uk http://www.nckc.org.uk/e   ------------------------------  % Date: Tue, 24 Dec 2002 09:36:06 -0800y% From: Dean Woodward <deanw@rdrop.com>n% Subject: [OT] Why Windows isn't ready ( Message-ID: <3E089B06.8030302@rdrop.com>  G It's not that it's not stable enough- the right combination of drivers hD and hardware can make an Wintel box stable for the vast majority of F business needs.  I've got a Wintel box that's nothing special and has C been up for- let's see, I was last there in October '01, so better   than a year.  G No, it's the way a [nominally] user-level process can wedge the system  E up so tight that a power cycle is the only way to unwedge it- unless -D it unwedges it self by crashing the machine.  Never mind protecting G the system from hostile external forces- until Microsoft can protect a b* system from supposedly benign user apps...  D Well, it's done rebooting, so I'll end the rant and try to get some  work done now.   ------------------------------   End of INFO-VAX 2002.710 ************************