0 INFO-VAX	Tue, 31 Jan 2006	Volume 2006 : Issue 62      Contents: Re: Cluster timeouts question  Re: Cluster timeouts question  Re: Cluster timeouts question  Re: Cluster timeouts question  Re: Cluster timeouts question  Re: Connectivity problem?  Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing: Re: HP, Intel becoming laughing stock of computer industry Re: IA64: Montecito info Re: IA64: Montecito info< Re: Intel boycott along with HP if they try to kill OpenVMS!< Re: Intel boycott along with HP if they try to kill OpenVMS!< Re: Intel boycott along with HP if they try to kill OpenVMS!< Re: Intel boycott along with HP if they try to kill OpenVMS!< Re: Intel boycott along with HP if they try to kill OpenVMS! Re: Open VMS programing in C: Re: p2c (dave gillespie's pas to c converter) & VMS pascal: Re: p2c (dave gillespie's pas to c converter) & VMS pascal: Re: p2c (dave gillespie's pas to c converter) & VMS pascal: Re: p2c (dave gillespie's pas to c converter) & VMS pascal: Re: p2c (dave gillespie's pas to c converter) & VMS pascal: Re: p2c (dave gillespie's pas to c converter) & VMS pascal$ Re: Problems with DECnetPlus-over-IP Re: Splitting serial cables ?  Re: Splitting serial cables ?  Re: Splitting serial cables ?  Re: Splitting serial cables ? # The Pentium Chronicles (2006.01.31)   F ----------------------------------------------------------------------  % Date: Tue, 31 Jan 2006 11:48:31 +0000 ! From: Baldrick <none@[127.0.0.1]> & Subject: Re: Cluster timeouts question' Message-ID: <drniqg$bs2$1@lore.csc.com>    Tom Linden wrote:   D > On Mon, 30 Jan 2006 18:37:40 +0000 (UTC), Phillip Helbig---remove  > I > Then explain this.  This morning I was getting excessive packet loss on H > one of my alpha nodes,  so I tried reseating the ethernet cable simplyJ > by sliding  it in an out.  The node crashed.  Now the router is a cisco,K > the orange light came on for some length of time which I interpreted to    > meanI > that it hadn't acquired sync, just a WAG.  It would be nice to have a   
 > function+ > like, "I am going to move you so hang on"   H You can increase (dynamically) RECNXINTERVAL on all nodes using SYSMAN, G i.e. PARAM USE ACTIVE, PARAM SET RECNXINTERVAL 500, PARAM WRITE ACTIVE  * when you are in a SET ENV/CLUSTER setting.  G However, the connection manager is sensitive to packet corruption, and  H particularly if you are in a CISCO environment which "packet snoops" to F dump non IP multicast packets "when busy" you are on dodgy ground and , the systems are not guaranteed to remain up.  C The whole point about a crash is that the system would rather shut  6 itself down that be in any doubt about data integrity.  I In your case Tom I expect there is another cause of the symptom, you had  C packet loss and I would guess significant errors against PEA0, and  G moving the cable was the straw breaking the camel's back, bringing the  	 box down.    --  E Regards, Nic Clews a.k.a. Mr. Car Park Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------    Date: 31 Jan 2006 08:20:33 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) & Subject: Re: Cluster timeouts question3 Message-ID: <Gj0J6BrbaRgk@eisner.encompasserve.org>   K In article <drl35u$3rs$1@lore.csc.com>, Baldrick <none@[127.0.0.1]> writes:  >>  I >> Unusually Nic, that's wrong.  check page 4-18 of VAXCluster Principles I >> (""The KFQSA provides an interface between a Q-bus and a DSSI bus, and I >> supports only VMS-to-device communication.") or page 4-11 (as at v7.1) H >> of "Guidelines for OpenVMS Cluster Configurations" ("The KFQSA cannot@ >> be used for node-to-node cluster comunication.  An additionalE >> interconnect must be configured between systems that use KFQSA for  >> access to shared storage.") > ...  >> Steve > J > I haven't come across the QBUS DSSI controllers, I've used the embedded ? > VAX and the Alpha (EISA and PCI), but I wasn't aware of this  A > restriction, I'll always maintain I have things to learn. This  J > controller is another special case, I would imagine it is of the era of H > the 3400 to 3900 VAX series, and designed for those systems, to allow K > storage interoperability between their later cousins, while allowing the  . > (storage) interface to gain some popularity. >   H    I had one in a MicroVAX II just so I could have a cluster common disk3    while I worked on the VAX 3400 on the other end.    ------------------------------  + Date: Tue, 31 Jan 2006 16:55:03 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)& Subject: Re: Cluster timeouts question$ Message-ID: <dro4p7$q0s$2@online.de>  F In article <ops37xazx5zgicya@hyrrokkin>, "Tom Linden" <tom@kednos.com> writes:   I > Then explain this.  This morning I was getting excessive packet loss on H > one of my alpha nodes,  so I tried reseating the ethernet cable simplyJ > by sliding  it in an out.  The node crashed.  Now the router is a cisco,K > the orange light came on for some length of time which I interpreted to    > meanI > that it hadn't acquired sync, just a WAG.  It would be nice to have a   
 > function+ > like, "I am going to move you so hang on"    How many nodes in your cluster?    ------------------------------   Date: 31 Jan 2006 17:50:47 GMT( From: bill@cs.uofs.edu (Bill Gunshannon)& Subject: Re: Cluster timeouts question+ Message-ID: <449mbnF13l7qU1@individual.net>   $ In article <dro4nl$q0s$1@online.de>,S 	helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) writes: 7 > In article <43DE91BD.5500D1F0@teksavvy.com>, JF Mezei ) > <jfmezei.spamnot@teksavvy.com> writes:   > 2 >> Phillip Helbig---remove CLOTHES to reply wrote:I >> > VMS is robust.  I remember when I accidentally pulled the SCSI cable K >> > from the box housing the system disk of a VAX.  Apart from a couple of 7 >> > mount-verification messages, nothing bad happened.  >>  D >> Not that robust :-) Yesterday, while stringing ethernet cables, II >> accidentally pulled the plug of a power bar feeding a vaxstation and a I >> mac. Both shut down and needed to be rebooted. :-) So VMS is no better " >> than MacOS in that regards. :-) > H > SOME VAX models could RESTART instead of REBOOT after a power outage, < > the contents of memory having been preserved in batteries.  E Not new with the VAX, the PDP-11 did that.  While I have no batteries 4 ont hem I have a number of the BBU's for my 11/44's.   bill      --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------    Date: 31 Jan 2006 12:32:16 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) & Subject: Re: Cluster timeouts question3 Message-ID: <iCuAfB1Mmlnf@eisner.encompasserve.org>   w In article <dro4nl$q0s$1@online.de>, helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) writes: 7 > In article <43DE91BD.5500D1F0@teksavvy.com>, JF Mezei ) > <jfmezei.spamnot@teksavvy.com> writes:   > 2 >> Phillip Helbig---remove CLOTHES to reply wrote:I >> > VMS is robust.  I remember when I accidentally pulled the SCSI cable K >> > from the box housing the system disk of a VAX.  Apart from a couple of 7 >> > mount-verification messages, nothing bad happened.  >>  D >> Not that robust :-) Yesterday, while stringing ethernet cables, II >> accidentally pulled the plug of a power bar feeding a vaxstation and a I >> mac. Both shut down and needed to be rebooted. :-) So VMS is no better " >> than MacOS in that regards. :-) > H > SOME VAX models could RESTART instead of REBOOT after a power outage, < > the contents of memory having been preserved in batteries.  H    That's what I needed:  a little VAX 2000 with a great big UPS next to
    it.    8-)   D    But I spent a good bit of time reading crash dumps from my 11/780I    only to figure out that a planned power source switch was too slow for H    unprotected memory, but too fast for the CPU to trigger a cold start.  F    Still, at the time there were IBM 360/75 in the room that needed 45/    minutes to do an IPL after the power switch.    ------------------------------  % Date: Tue, 31 Jan 2006 17:02:05 +0100 ( From: Paul Sture <paul.sture@bluewin.ch>" Subject: Re: Connectivity problem?+ Message-ID: <449fvuF12m8mU1@individual.net>    Bill Gunshannon wrote:* > In article <ops36b8nrnzgicya@hyrrokkin>,( > 	"Tom Linden" <tom@kednos.com> writes: > G >>I rebooted a 7.3-2 node and when I tried to connect to it using PuTTY : >>SSH I got "server unexpectedly closed server connection" >>F >>But from the attached kybd I can login directly; however when I tred >>$ SET HOST node ? >>%SYSTEM-F-UNREACHABLE, remote node is not currently reachable  >>K >> From the attached kybd if I do a SET HOST whatever  It gives the message  >>%SYSTEM-F-DEVNOTMOUNT  >> >>So what might have happened? >  > I > Sounds like your networking didn't come back  up on reboot.  Looks like  > time to read the logs. >   G Does this relate to the node which crashed when you were reseating the   ethernet cable?   D I used to see the network fail to come up fully on one cluster (2 x H 4100s acting as boot servers for ~20 PWS systems), due to a full shadow C copy on the system disk being provoked by the satellites rebooting.   I When the lot was coming up at once, some timeout somewhere would kick in  ? (I forget exactly what) and the network wouldn't come up fully.   G The fastest way to recover was to drop one shadow member of the system  D disk and reboot the affected nodes, bringing the shadow member back  online once everything was up.  H The cure was to change the satellites to DOSD (Dump Off System Disk) to  their local disks.   ------------------------------    Date: 31 Jan 2006 08:53:09 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) ' Subject: Re: FOR070.DAT files appearing 3 Message-ID: <4xI7G1jLHVE9@eisner.encompasserve.org>   U In article <447v3gFs35lU1@individual.net>, bill@cs.uofs.edu (Bill Gunshannon) writes:  > E > What?  Where in the header of the file does it say what program and 0 > which routine within that program created it?   J    I was explictly talking about those OPEN options that can be picked up G    from an existing file, not it's content or purpose.  If they weren't 1    in the header then they couldn't be picked up.  > F > Must be a VMS extension cause there is no "IMPLICIT NONE" in my copyH > of the standard. :-)  but, in any case that just supports exactly whatG > I said in the paragraphs above.  Nothing should be left to implicit.  + > Everything should be explicitly declared.   C    IMPLICIT NONE has been in every Fortran compiler I've used since G    1976.  If it's not in the standard it certainly should be.  And I've D    already told you why your adding unecessary maintenance issues to=    code.  It's folks that won't listen that cause the cost of #    maintaining a system to stay up.   8 > Buffer overflows are not the only example of bad code.  D    Yes.  You insistence on over specifying is another example of bad    code.   ------------------------------   Date: 31 Jan 2006 15:47:29 GMT( From: bill@cs.uofs.edu (Bill Gunshannon)' Subject: Re: FOR070.DAT files appearing + Message-ID: <449f4hF125haU1@individual.net>   3 In article <4xI7G1jLHVE9@eisner.encompasserve.org>, > 	koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:W > In article <447v3gFs35lU1@individual.net>, bill@cs.uofs.edu (Bill Gunshannon) writes:  >>  F >> What?  Where in the header of the file does it say what program and1 >> which routine within that program created it?   > L >    I was explictly talking about those OPEN options that can be picked up I >    from an existing file, not it's content or purpose.  If they weren't 3 >    in the header then they couldn't be picked up.    OK.    >>  G >> Must be a VMS extension cause there is no "IMPLICIT NONE" in my copy I >> of the standard. :-)  but, in any case that just supports exactly what H >> I said in the paragraphs above.  Nothing should be left to implicit. , >> Everything should be explicitly declared. > E >    IMPLICIT NONE has been in every Fortran compiler I've used since  >    1976.    F I checked what I had close at hand, Prime FTN, Prime F77 and U SalfordI F77.  None of them have it.  They are all from the early 80's  From X3J3: + "    The form of an IMPLICIT statement is:  7          IMPLICIT typ (a [,a]...) [,typ (a [,a]...)]... M               where: typ is one of INTEGER, REAL, DOUBLE PRECISION, COMPLEX,  ,               LOGICAL, or CHARACTER [*len] "   Nope, no "NONE".  @ >          If it's not in the standard it certainly should be.    , Actually, I think it should be the standard.  G >                                                              And I've F >    already told you why your adding unecessary maintenance issues to? >    code.  It's folks that won't listen that cause the cost of % >    maintaining a system to stay up.   F Maybe you mis-understood what I was advocating.  I believe in explicitI definitions.  You know like Pascal, C, Ada and pretty much every language K since Fortran has required.  Declare before use.  No implicit declarations. H If that is going to "cause the cost of maintaining a system to stay up",9 then why did all these languages decide it was necessary?    > 9 >> Buffer overflows are not the only example of bad code.  > F >    Yes.  You insistence on over specifying is another example of bad
 >    code.  C Now I am sure you mis-understood what I said.  Where did I advocate A "over specifying"?  I merely said that the Fortran concept of not D requiring things like variables (and yes, external files) to be pre-F declared before use was a bad idea dn led to much harder to understandH programs.  Especially after the passage of time and possibly the passage of the original writer.    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: 31 Jan 2006 08:32:28 -0800 From: davidc@montagar.com ' Subject: Re: FOR070.DAT files appearing B Message-ID: <1138725148.592726.63140@o13g2000cwo.googlegroups.com>  G This is very bad, since a user could create the FOR070 logical to point = about anywhere and corrupt the system.  You might want to try D "define/exec/system for070 make_it_throw_up:foo.bar" and try to findC it.  Since NL: hides the problem, a real directory doesn't help the G problem, give it an invalid location to open and see what crashes then.    ------------------------------    Date: 31 Jan 2006 12:34:41 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) ' Subject: Re: FOR070.DAT files appearing 3 Message-ID: <3zZl5oxr4Bxz@eisner.encompasserve.org>   V In article <449f4hF125haU1@individual.net>, bill@cs.uofs.edu (Bill Gunshannon) writes: > H > Maybe you mis-understood what I was advocating.  I believe in explicitK > definitions.  You know like Pascal, C, Ada and pretty much every language M > since Fortran has required.  Declare before use.  No implicit declarations. J > If that is going to "cause the cost of maintaining a system to stay up",; > then why did all these languages decide it was necessary?  >  >>  : >>> Buffer overflows are not the only example of bad code. >>  G >>    Yes.  You insistence on over specifying is another example of bad  >>    code.  > E > Now I am sure you mis-understood what I said.  Where did I advocate C > "over specifying"?  I merely said that the Fortran concept of not F > requiring things like variables (and yes, external files) to be pre-H > declared before use was a bad idea dn led to much harder to understandJ > programs.  Especially after the passage of time and possibly the passage > of the original writer.   -    OK, so we're chasing two different horses.    ------------------------------    Date: 31 Jan 2006 08:24:17 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) C Subject: Re: HP, Intel becoming laughing stock of computer industry 3 Message-ID: <eaIBRCtS2NHo@eisner.encompasserve.org>   r In article <1138548360.018354.309660@g47g2000cwa.googlegroups.com>, "rcyoung" <rcyoung@aliconsultants.com> writes:H > Well all I know if we have  an entry level I64 running OpenVMS 8.2 andE > used for development purposes, and it runs circles around the other E > development systems ( a Vax 4000 and a dual processor Alpha 2100A). I > Based on that, they needed to retire the Alphas and come up with a more E > modern design...whether they did it correctly by "rolling their own G > unique processor design" is the BIG question...the answer to which is / > starting to look all too obvious I am afraid.  > G > A pity since the Itanium does seem to be a real speedy little box....   G    Lots of Alphas will also run circles around that 2100A.  It's pretty     old.    ------------------------------  % Date: Tue, 31 Jan 2006 09:33:02 +0000 ! From: Baldrick <none@[127.0.0.1]> ! Subject: Re: IA64: Montecito info ' Message-ID: <drnasj$b3q$1@lore.csc.com>    Robert Deininger wrote: 7 > In article <43DD92BC.6A0DA374@teksavvy.com>, JF Mezei ' > <jfmezei.spamnot@teksavvy.com> wrote:  > J >>Question: Since that IA64 is finally going dual core, will VMS licencing >>get any changes ?  >  > 7 > Each core will count as a CPU for licensing purposes.   E No it is each socket. When it(anium) is 4 core, it will still be one  I socket and one license. As you populate a server with 2 or 4 chips, then  J you'll need to license those regardless of the cores present on each chip.   --  E Regards, Nic Clews a.k.a. Mr. Car Park Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------    Date: 31 Jan 2006 08:43:22 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) ! Subject: Re: IA64: Montecito info 3 Message-ID: <5uFhnXXyNhAY@eisner.encompasserve.org>   _ In article <1138628500.243226.325020@g44g2000cwa.googlegroups.com>, bob@instantwhip.com writes: < > what about companies that need 1P?  We do not want to have= > to pay for two cpu licenses when we only need one for small > > workgroups ... were is the 1P line of cheap 2K boxes we were > promised?   B    I only heard that promise in the discussions of what developers    need, and DSPP delivered.   ------------------------------    Date: 31 Jan 2006 00:21:06 -0800$ From: "AEF" <spamsink2001@yahoo.com>E Subject: Re: Intel boycott along with HP if they try to kill OpenVMS! C Message-ID: <1138695666.065220.213840@f14g2000cwb.googlegroups.com>    Dave Froble wrote: > AEF wrote: > > Dave Froble wrote: > >  > >>bob@instantwhip.com wrote: > >>F > >>>AMD will always be the processor of choice for us if Intel thinks/ > >>>their phony itanium strategy will work ...  > >>> < > >>>http://www.digitimes.com/bits_chips/a20060127PR201.html > >>>  > >>J > >>boob, if you're going to just give URLs, at least give those all of usL > >>can read, not paid sites.  If you want to post something, please include > >  > >  > > Hmmm, it was free for me.  > >  > > 1 > >>the details in your post, or just don't post.  > >  > > H > > Agreed. It would be nice to have *some* introductory material in the	 > > post.  > >  > >  > >>--8 > >>David Froble                       Tel: 724-529-04508 > >>Dave Froble Enterprises, Inc.      Fax: 724-529-0596B > >>DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com > >>170 Grimplin Road  > >>Vanderbilt, PA  15486  > >  > >  > > AEF  > >  > G > Ok, I tried again.  The page that displays has what appears to be the H > lead-in to stories, but not the full story.  Regardless, at the top of > the page:  > G > Sorry, the page you are trying to open is available only for our paid  > subscribers.  ; Here's what I get (ads on the sides and this in the middle)    [begin quoted material]   C The processor wars Part V, Intel's Itanium: To be or not to be? Q&A  with Nebojsa Novakovic  Printer friendly       Related stories    	  Comments       Email to a friend      Latest news    : Chris Hall, DigiTimes.com, Taipei [Friday 27 January 2006]    < To the outside observer, improvements in PC architecture areD evolutionary but logical. Processors advance inevitably in speed and> performance, in happy accordance with Moore's Law. For NebojsaD Novakovic, a consultant in high-end computing systems, that's hardlyE the case. The demise of the DEC Alpha processor is a case in point. A 4 performance leader was killed off by corporate whim.  C Novakovic argues that despite the turn by AMD to a 64-bit platform, E vindicated by the success of the Athlon 64, AMD had better watch out. E Intel will be fighting back in 2006 with Merom, Conroe and Woodcrest, G as processor design adopts a multicore approach. Even so, life won't be E a bed of roses for Intel either. Rumors abound in the industry of the E "retirement" of the Intel Itanium, an expensive attempt to break away D from the x86 architecture at the high end, and Intel sorely lacks an: interconnect that could compete with AMD's HyperTransport.  C Novakovic comments on the major technology issues, in this in-depth A interview, as Intel and AMD square off for the next rounds of the  processor wars.   G This is Part V of a five-part interview. Part I appeared on 23 January, ? Part II on 24 January, Part III on 25 January and Part IV on 26  January.    F [end quoted material]  [I'm assuming [that dreaded word!] this is fair use.]   D and then the interview is presented. You don't get any of the above?   ------------------------------    Date: 31 Jan 2006 01:38:54 -0800+ From: "Dr. Dweeb" <comp.os.vms@hotmail.com> E Subject: Re: Intel boycott along with HP if they try to kill OpenVMS! C Message-ID: <1138700334.213006.165280@g44g2000cwa.googlegroups.com>   G I just checked and it worked for me.  Maybe it knows I am not in the US  and lets me look.    ------------------------------  % Date: Tue, 31 Jan 2006 12:57:23 +0100 + From: Karsten Nyblad <nospam@nospam.nospam> E Subject: Re: Intel boycott along with HP if they try to kill OpenVMS! = Message-ID: <43df509e$0$78279$157c6196@dreader1.cybercity.dk>    Dr. Dweeb wrote:I > I just checked and it worked for me.  Maybe it knows I am not in the US  > and lets me look.  > F I can't see it, and I think we are both using Cybercity in Denmark as I ISP.  I could see the article when Niel Rieck posted links for it in the  H thread `The processor wars Part 1, "The death of Alpha."  Perhaps they C will only let you see new articles or a certain number of articles  H without paying?  Anyway paying $370 for reading that article is a waste 	 of money.    ------------------------------  % Date: Tue, 31 Jan 2006 10:14:36 -0500 ' From: Dave Froble <davef@tsoft-inc.com> E Subject: Re: Intel boycott along with HP if they try to kill OpenVMS! / Message-ID: <EOSdnZ6uV9Iy40LeRVn-vg@libcom.com>   
 AEF wrote: > Dave Froble wrote: >  >>AEF wrote: >> >>>Dave Froble wrote:  >>>  >>>  >>>>bob@instantwhip.com wrote: >>>> >>>>F >>>>>AMD will always be the processor of choice for us if Intel thinks/ >>>>>their phony itanium strategy will work ...  >>>>> < >>>>>http://www.digitimes.com/bits_chips/a20060127PR201.html >>>>>  >>>>J >>>>boob, if you're going to just give URLs, at least give those all of usL >>>>can read, not paid sites.  If you want to post something, please include >>>  >>>  >>>Hmmm, it was free for me. >>>  >>>  >>> 1 >>>>the details in your post, or just don't post.  >>>  >>> G >>>Agreed. It would be nice to have *some* introductory material in the  >>>post. >>>  >>>  >>>  >>>>--8 >>>>David Froble                       Tel: 724-529-04508 >>>>Dave Froble Enterprises, Inc.      Fax: 724-529-0596B >>>>DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com >>>>170 Grimplin Road  >>>>Vanderbilt, PA  15486  >>>  >>>  >>>AEF >>>  >>G >>Ok, I tried again.  The page that displays has what appears to be the H >>lead-in to stories, but not the full story.  Regardless, at the top of >>the page:  >>G >>Sorry, the page you are trying to open is available only for our paid  >>subscribers. >  > = > Here's what I get (ads on the sides and this in the middle)  >  > [begin quoted material]  > E > The processor wars Part V, Intel's Itanium: To be or not to be? Q&A  > with Nebojsa Novakovic >  Printer friendly  >  >  >  Related stories >  >  >  Comments  >  >  >  Email to a friend >  >  >  Latest news >  > < > Chris Hall, DigiTimes.com, Taipei [Friday 27 January 2006] >  > > > To the outside observer, improvements in PC architecture areF > evolutionary but logical. Processors advance inevitably in speed and@ > performance, in happy accordance with Moore's Law. For NebojsaF > Novakovic, a consultant in high-end computing systems, that's hardlyG > the case. The demise of the DEC Alpha processor is a case in point. A 6 > performance leader was killed off by corporate whim. > E > Novakovic argues that despite the turn by AMD to a 64-bit platform, G > vindicated by the success of the Athlon 64, AMD had better watch out. G > Intel will be fighting back in 2006 with Merom, Conroe and Woodcrest, I > as processor design adopts a multicore approach. Even so, life won't be G > a bed of roses for Intel either. Rumors abound in the industry of the G > "retirement" of the Intel Itanium, an expensive attempt to break away F > from the x86 architecture at the high end, and Intel sorely lacks an< > interconnect that could compete with AMD's HyperTransport. > E > Novakovic comments on the major technology issues, in this in-depth C > interview, as Intel and AMD square off for the next rounds of the  > processor wars.  > I > This is Part V of a five-part interview. Part I appeared on 23 January, A > Part II on 24 January, Part III on 25 January and Part IV on 26 
 > January. >  > H > [end quoted material]  [I'm assuming [that dreaded word!] this is fair > use.]  > F > and then the interview is presented. You don't get any of the above? >   D When the 5 sections came out I was able to read them using the URLs H given at the time, but using the above URL I got the message about paid I subscribers.  I didn't try going further.  I assumed the statement about   paid subscribers was factual.    --  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: Tue, 31 Jan 2006 10:27:40 -0700 " From: GreyCloud <mist@cumulus.com>E Subject: Re: Intel boycott along with HP if they try to kill OpenVMS! 0 Message-ID: <J_2dnTG2SIQ0A0LeRVn-rw@bresnan.com>   Dr. Dweeb wrote:I > I just checked and it worked for me.  Maybe it knows I am not in the US  > and lets me look.  >   B Still, he could of done us all a favor and at least post the most  pertinent part.    --   Where are we going?   And why am I in this handbasket?   ------------------------------  % Date: Tue, 31 Jan 2006 07:24:45 -0500 ) From: "Neil Rieck" <n.rieck@sympatico.ca> % Subject: Re: Open VMS programing in C 9 Message-ID: <%FIDf.3483$Iw6.316602@news20.bellglobal.com>   B "Mike Rechtman" <michael.rechtman.nospam@hp.com> wrote in message   news:43DF7067.4FC2B09D@hp.com... > Neil Rieck wrote:  >> [...snip...] > F > I apologize to EDT. I started on RSTS myself, and the first editor IJ > used was a line editor on an LA120...  When we got VT100s and KED (later% > went to EDT) it was a real jump up. G > My fingers still only know EDT, which is why I always 'set ke edt' in " > tpu, and use my own TPU$SECTION. >  > Mike > M I'm in the same boat. Like after learning to play a piano, my fingers "know"  M their way around the EDT keypad. My first DEC editor was TECO and we used it  2 to modify DEC-EX diagnostic scripts on the PDP-11.  ; (but you'll never catch me typing $EDIT/TECO on OpenVMS :-)   
 Neil Rieck Kitchener/Waterloo/Cambridge,  Ontario, Canada.9 http://www3.sympatico.ca/n.rieck/links/cool_openvms.html     ------------------------------  % Date: Tue, 31 Jan 2006 08:27:17 +0000 0 From: Chris Sharman <chris.sharman@sorry.nospam>C Subject: Re: p2c (dave gillespie's pas to c converter) & VMS pascal 4 Message-ID: <drn715$st9$1$8300dec7@news.demon.co.uk>   John Reagan wrote: > Chris Sharman wrote: > C >> Has anyone had experience of this, and got (or failed to get) a   >> working configuration ? >>I >> I've got it compiled & working under fedora, and would like to port a  L >> number of our vms pascal programs to there (while retaining them on vms). >>D >> My initial problems are with ! delimited line comments, and with @ >> [WEAK_GLOBAL] and other attributes (it doesn't like the '['). >>H >> It does have a language VAX option, but it's very old, and there's a ) >> note "not tested with large programs".  >>G >> I suspect other recent enhancements (%if etc) are going to cause me   >> trouble too.  >>I >> If anyone's already spent time looking at this, or has any tips, that  1 >> would be great. Meantime I'll continue trying.   K > I looked at this once many years ago just for fun.  Yes, it doesn't know  K >  much about any OpenVMS Pascal extensions.  When it says VAX, I think it  F > is dealing with the VAX Pascal V1 extension set, not the much newer 9 > language definition that came along with VAX Pascal V2.  > @ > In addition, the "!" end-of-line comments and %IF conditional I > compilation feature came along MUCH later than V2.  So even if he knew  F > about VAX Pascal V2 when p2c was written, I'm not surprised that it ( > can't handle your code that uses them. > B > You might be better off using one of the VAX emulator solutions.  I Not made much headway - !, %if, and [attributes] look like too much work  I to fix. It didn't seem to like default argument values either, if I read  H the errors right. I don't think I really want to run a VAX emulation on D a time-critical real-time application on our linux webserver - it's G going to be a hand re-write in C, I think. Probably the right approach  H anyway, seeing as the application has to do parsing, and endian-correct # bit extraction, among other things.   < Here's as far as I got with it, if anyone else wants to play Synonym UNSIGNED16 word  Synonym UNSIGNED8 byte Synonym INTEGER16 sword  Synonym INTEGER8 sbyte ReplaceBefore "VALUE" ":="5 Language VAX ## don't know whether this does any good    Thanks Chris    ------------------------------   Date: 31 Jan 2006 12:03:54 GMT( From: bill@cs.uofs.edu (Bill Gunshannon)C Subject: Re: p2c (dave gillespie's pas to c converter) & VMS pascal * Message-ID: <44921aFvvmgU1@individual.net>  4 In article <drn715$st9$1$8300dec7@news.demon.co.uk>,3 	Chris Sharman <chris.sharman@sorry.nospam> writes:  > J >                   I don't think I really want to run a VAX emulation on ? > a time-critical real-time application on our linux webserver  >     ^^^^^^^^^^^^^ ^^^^^^^^^                    ^^^^^ ^^^^^^^^^ Your joking, right?    bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  % Date: Tue, 31 Jan 2006 10:28:56 -0500 ' From: Dave Froble <davef@tsoft-inc.com> C Subject: Re: p2c (dave gillespie's pas to c converter) & VMS pascal / Message-ID: <W4KdnciUOcqRH0LeRVn-jQ@libcom.com>    Bill Gunshannon wrote:6 > In article <drn715$st9$1$8300dec7@news.demon.co.uk>,5 > 	Chris Sharman <chris.sharman@sorry.nospam> writes:  > J >>                  I don't think I really want to run a VAX emulation on ? >>a time-critical real-time application on our linux webserver   > @ >     ^^^^^^^^^^^^^ ^^^^^^^^^                    ^^^^^ ^^^^^^^^^ > Your joking, right?  >  > bill >   7 Well, I'm glad I wasn't the one to make that statement.   G There may be reasons for not using a current VAX system, but there are  E some VAX systems, the MicroVAX 3100 and such, that are small, do not  " need special power, and just work.  F There seems to be considerable success with Charon VAX, exceeding the  speed of any VAX system.  I Fixing something that ain't broke, such as a re-write in (hawk, spit) C,  B seems to me to be a totally worthless redundant effort.  But if a @ "time-critical real-time application" on a "linux webserver" is % involved, I'm guessing anything goes.    --  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: Tue, 31 Jan 2006 12:04:18 -0500 2 From: "Stanley F. Quayle" <squayle@insight.rr.com>C Subject: Re: p2c (dave gillespie's pas to c converter) & VMS pascal . Message-ID: <43DF5242.2445.5D0BD4A4@localhost>  + On 31 Jan 2006 at 10:28, Dave Froble wrote: G > But if a "time-critical real-time application" on a "linux webserver" + > is involved, I'm guessing anything goes.    B Emulating a VAX is a full-time job.  Running CHARON-VAX on a host @ system (other than Alpha) doing anything else is an unsupported @ configuration.  Usually works fine, sometimes not.  Sounds like $ suicide for a real-time application.  ' I have customers doing it, of course...   
 --Stan Quayle  Quayle Consulting Inc.  
 ----------8 Stanley F. Quayle, P.E. N8SQ  Toll free: 1-888-I-LUV-VAX3 8572 North Spring Ct., Pickerington, OH  43147  USA 0 stan-at-stanq-dot-com       http://www.stanq.com) "OpenVMS, when downtime is not an option"    ------------------------------  % Date: Tue, 31 Jan 2006 17:48:03 +0000 ! From: Baldrick <none@[127.0.0.1]> C Subject: Re: p2c (dave gillespie's pas to c converter) & VMS pascal ' Message-ID: <dro7sk$dqh$1@lore.csc.com>    Stanley F. Quayle wrote:  - > On 31 Jan 2006 at 10:28, Dave Froble wrote:  > G >>But if a "time-critical real-time application" on a "linux webserver" + >>is involved, I'm guessing anything goes.   >  > D > Emulating a VAX is a full-time job.  Running CHARON-VAX on a host B > system (other than Alpha) doing anything else is an unsupported B > configuration.  Usually works fine, sometimes not.  Sounds like & > suicide for a real-time application. > ) > I have customers doing it, of course...   = How, can you run real-time on a host which is dependent on a  = non-premptive OS host? (i.e. Windows, which is NOT real time)   H I suspect it is not _real_ time but just happens to be fast enough, and I that if push came to shove it would fail. Like, when 5ms must and has to  F be, without exception, 5ms. If windows does what Windows does best at ! that critical time, you're toast.    Any engineers here?    --  E Regards, Nic Clews a.k.a. Mr. Car Park Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------    Date: 31 Jan 2006 12:39:38 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) C Subject: Re: p2c (dave gillespie's pas to c converter) & VMS pascal 3 Message-ID: <1kvlNy2uc0m6@eisner.encompasserve.org>   Y In article <W4KdnciUOcqRH0LeRVn-jQ@libcom.com>, Dave Froble <davef@tsoft-inc.com> writes:  > K > Fixing something that ain't broke, such as a re-write in (hawk, spit) C,  D > seems to me to be a totally worthless redundant effort.  But if a B > "time-critical real-time application" on a "linux webserver" is ' > involved, I'm guessing anything goes.   B    Time-critical may not mean the same thing to all of us, just as    real-time doesn't.   E    When I say real-time, I mean "hard real-time" as in high interrupt .    rates and I/O devices that will not buffer.  E    I have interviewed folks from the nuclear power industry for which H    real-time means "if it take 24 hours to melt down in real life, then C    it takes 24 hours to melt down in the simulation", and I can see <    someone considering that to be a time-critical simulator.   ------------------------------    Date: 31 Jan 2006 07:49:56 +01006 From: peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER)- Subject: Re: Problems with DECnetPlus-over-IP , Message-ID: <43df16a4$1@news.langstoeger.at>  ^ In article <1138658295.599945.95010@z14g2000cwz.googlegroups.com>, bob@instantwhip.com writes:2 >if you can not ask Process techs, why not go back >to the stable 5.6-2 version?   < 1) If you could read you would know that I'm back to V5.6-2.= 2) V5.6-2 is not supported on OpenVMS V8.2 (and doesn't run). 8 3) I wanted to upgrade TCPware first (and OpenVMS later)- 	a) to have TCPware on V5.7-2 on all my nodes 5 	b) to not upgrade too much software at the same time C 4) I already asked Process techs via vmsnet.networks.tcp-ip.tcpware    btw. What is your plan ?H Do you want to upgrade or stay forever on your current system/versions ?   --   Peter "EPLAN" LANGSTOEGER % Network and OpenVMS system specialist  E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------    Date: 31 Jan 2006 07:53:02 -0600 From: briggs@encompasserve.org& Subject: Re: Splitting serial cables ?3 Message-ID: <F7jiQDz4lkLG@eisner.encompasserve.org>   e In article <v%vDf.2975$Iw6.196541@news20.bellglobal.com>, "Neil Rieck" <n.rieck@sympatico.ca> writes:  > . > <briggs@encompasserve.org> wrote in message / > news:oCgSttFou3DR@eisner.encompasserve.org... 9 >> In article <43DD3DBC.6E32BF1E@teksavvy.com>, JF Mezei  ) >> <jfmezei.spamnot@teksavvy.com> writes:  > [...snip...] >>F >> Transmit data is on one of those wires.  You have two stations bothD >> wired into transmit data.  Both of them will be holding that leadF >> in the space condition (or mark -- I can never remember) when idle.? >> This is not tri-state logic where the lead floats when idle.  >>G >> If you've ever put a breakout box on an RS232 connector, you'll have * >> noticed that TD and RD are always live. >> > [...snip...] >>F >> You can have one source sending data to multiple receivers.  That'sJ >> no problem since the receivers aren't trying to coerce the signal line. >>E >> You can't have multiple sources sending data to a single receiver. D >> That's a problem since multiple senders are each trying to coerce >> the signal line differently.  >>L > Umm, if you read my original post you'll notice that I said 2 diodes were M > required to isolate the two transmit lines from each other. Only one phase  J > of the signal (negative if memory serves) is required to send data from G > transmitter to receiver. Without the diodes one transmitter would be  D > positive while the other is negative and you would have a problem.  G I posted before seeing the reports that diodes were a possible solution - that had actually been deployed successfully.   B What I recall of RS232 in action is that transmit and receive dataJ use positive and negative voltages to signal mark and space.  Zero voltage is never used.  I Checking online I see that the signalling range is technically -3 to -25V D denoting a logic 1 (mark) and +3 to +25V denoting a logic 0 (space).  B Actual implementations are more likely to keep the signals down to7 8 to 14 volts according to the first reference I found.   D No matter.  A diode can still work in any case.  It just needs to beB capable of dealing with 50 volts end to end without breaking down.  G And it needs to leak enough current so that even when both transmitters A are idle and, consequently, both diodes are blocking current flow @ that enough signal leaks through to keep the receiver seeing the mark condition.   B One poster mentioned a pull-up resister tied to DSR, presumably to& address with this leakage requirement.  B Mind you, I'm speaking from theory, not from practice.  Looking atC LEDs on a breakout box is about as far as I've ever actually mucked ! about with the relevant hardware.    ------------------------------   Date: 31 Jan 2006 14:34:07 GMT( From: bill@cs.uofs.edu (Bill Gunshannon)& Subject: Re: Splitting serial cables ?+ Message-ID: <449aqvF11n9iU1@individual.net>   3 In article <F7jiQDz4lkLG@eisner.encompasserve.org>, ! 	briggs@encompasserve.org writes: g > In article <v%vDf.2975$Iw6.196541@news20.bellglobal.com>, "Neil Rieck" <n.rieck@sympatico.ca> writes:  >>  / >> <briggs@encompasserve.org> wrote in message  0 >> news:oCgSttFou3DR@eisner.encompasserve.org...: >>> In article <43DD3DBC.6E32BF1E@teksavvy.com>, JF Mezei * >>> <jfmezei.spamnot@teksavvy.com> writes: >> [...snip...]  >>> G >>> Transmit data is on one of those wires.  You have two stations both E >>> wired into transmit data.  Both of them will be holding that lead G >>> in the space condition (or mark -- I can never remember) when idle. @ >>> This is not tri-state logic where the lead floats when idle. >>> H >>> If you've ever put a breakout box on an RS232 connector, you'll have+ >>> noticed that TD and RD are always live.  >>>  >> [...snip...]  >>> G >>> You can have one source sending data to multiple receivers.  That's K >>> no problem since the receivers aren't trying to coerce the signal line.  >>> F >>> You can't have multiple sources sending data to a single receiver.E >>> That's a problem since multiple senders are each trying to coerce   >>> the signal line differently. >>> M >> Umm, if you read my original post you'll notice that I said 2 diodes were  N >> required to isolate the two transmit lines from each other. Only one phase K >> of the signal (negative if memory serves) is required to send data from  H >> transmitter to receiver. Without the diodes one transmitter would be E >> positive while the other is negative and you would have a problem.  > I > I posted before seeing the reports that diodes were a possible solution / > that had actually been deployed successfully.  > D > What I recall of RS232 in action is that transmit and receive dataL > use positive and negative voltages to signal mark and space.  Zero voltage > is never used. > K > Checking online I see that the signalling range is technically -3 to -25V F > denoting a logic 1 (mark) and +3 to +25V denoting a logic 0 (space). > D > Actual implementations are more likely to keep the signals down to9 > 8 to 14 volts according to the first reference I found.  > F > No matter.  A diode can still work in any case.  It just needs to beD > capable of dealing with 50 volts end to end without breaking down. > I > And it needs to leak enough current so that even when both transmitters C > are idle and, consequently, both diodes are blocking current flow B > that enough signal leaks through to keep the receiver seeing the > mark condition.  > D > One poster mentioned a pull-up resister tied to DSR, presumably to( > address with this leakage requirement. > D > Mind you, I'm speaking from theory, not from practice.  Looking atE > LEDs on a breakout box is about as far as I've ever actually mucked # > about with the relevant hardware.   G While many people have written to say this is a sure thing and that the F diodes are the answer, I can assure you that as long as 25 years ago IF had to trouble shoot a terminal<->host problem that turned out to be aG host that output only TTL on it's serial port (+5v mark - 0v space) and C a terminal (Wyse, I think) that followed the standard correctly and D refused to work as it never saw a space.  +3 to -3 is defined as notC valid data.  I would expect devices today to be much more likely to E stick to standards as it avoids a whole bunch of finger pointing when  things don't work.  Be aware!!   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, 31 Jan 2006 15:48:10 +0000 (UTC) 7 From: moroney@world.std.spaamtrap.com (Michael Moroney) & Subject: Re: Splitting serial cables ?( Message-ID: <dro0rq$6qp$1@pcls4.std.com>  * bill@cs.uofs.edu (Bill Gunshannon) writes:  9 >  I would expect devices today to be much more likely to F >stick to standards as it avoids a whole bunch of finger pointing when >things don't work.  Be aware!!   G Not necessarily.  Consumer UPS's from a few years ago used serial ports C speaking RS232 to control them, but they used a DB-9 using a pinout A entirely different from the standard (if it is a standard) pinout E widely used today.  In fact, if you used a standard cable to a PC and ? tried to transmit, many would power off, taking the PC with it.   1 Today they use USB, who knows if that's standard.    ------------------------------  % Date: Tue, 31 Jan 2006 11:26:18 -0500 C From: "David Turner, Island Computers US Corp" <dbturner@icusc.com> & Subject: Re: Splitting serial cables ?3 Message-ID: <F4MDf.20$_i2.7@bignews3.bellsouth.net>   J You can buy a DECSERVER 300 for under $200 from any dealer in US or Canada  % It'll cost that+more in time to build   , Shop for one on ebay and maybe less than $10   DT   --     David B Turner Island Computers US Corp 2700 Gregory St, Suite 180 Savannah GA 31404  Tel: 912 447 6622 X201 Cell: 912 447 6622 X252  Fax: 912 201 0402  Email: dbturner@icusc.com  Web: http://www.islandco.com% ===================================== < All orders are subject to the following terms and conditions. of sale. These should be read before ordering.% http://www.islandco.com/warranty.html   ; "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message ' news:43DD2C88.CAB68891@vaxination.ca... H > Is it possible to split a serial cable coming from an OPA0: to go to 2G > separate serial ports ? (not a switch, I want hardwired connections).  > I > Goal: have a node's OPA0: go to serial ports on 2 different machines so E > that no matter which node is up, I can always SET HOST/DTE from the " > running node to access the OPA0: >  >  > F > And while I am at it, how about a 3 way ? Have OPA0: go to 2 serialsJ > ports as well as a true blue VT220 (actually, they are now yellowish :-)   ------------------------------  % Date: Tue, 31 Jan 2006 07:57:13 -0500 ) From: "Neil Rieck" <n.rieck@sympatico.ca> , Subject: The Pentium Chronicles (2006.01.31)9 Message-ID: <F8JDf.3494$Iw6.318697@news20.bellglobal.com>    Folks,  G I'm very close to finishing Robert Colwell's 2006 IEEE book titled "The F Pentium Chronicles". This book provides an inside look at the problemsL technical people encounter when working in large teams. Although it is aboutF making chips (mostly about developing the Pentium-Pro with some lesserK information about the P4) I would say it also applies to software projects. 7 At least it seems relevant where I'm currently working.   K I was surprised to learn that the beginnings of modern RISC technology were C begun in the early 1980's by analyzing VAX execution traces and VAX  micro-code. (Forward, p. xii)   	     #####   K Chapter 7 has some interesting essays but I'll only mention 5 of the titles ? which may be of some interest to some people in this newsgroup.   L 1. What was Intel thinking with that chip ID tag, which caused such a public uproar? G 2. Was the P6 (PentiumPro) project affected by Pentium's floating point  divider bug?H 3. Why did Pentium have a flawed floating point divider (FDIV), when its predecessor, the i486, did not? J 4. How would you respond to the claim that the P6 is build on ideas stolen from Digital Equipment Corp.? E 5. What did the P6 team think about Intel's Itanium Processor family?    8. Why did you leave Intel?   
 Neil Rieck Kitchener/Waterloo/Cambridge,  Ontario, Canada.8 http://www3.sympatico.ca/n.rieck/links/cool_openvms.html   ------------------------------   End of INFO-VAX 2006.062 ************************                                                                                                                                                                                                                          !˿,+ߏ-5tUw"qĝ'^#1Y,-'%ߗ-xKjEپE1A2^sᙞYNϸ'>*q|9o䪣x|)ߎ,nobof(x	<4pK%Ȱ#qǚ/g0+Jt4~/)<^j~_gBLĿ%SϠ06\H'/դc˟z	%%s?g.O/Ǯ(9Q]S1QR?1\ejZk8E߹,\Wu\WTST,+88kX/8[t/,NQqX*]8[L#{T3x^J?k2Q