1 INFO-VAX	Sun, 20 Aug 2000	Volume 2000 : Issue 464       Contents: Re: 100 MB/sec on 500au  Re: Big AlphaServer Sale RE: Big AlphaServer Sale Re: Changing address on MV3100 Re: Changing address on MV3100+ Re: Cluster transaction processing paradigm ( CRE/TER/DET challenge - hackaround found DECnet area routing  DECnet area routing $ Re: DECserver 700 documentation/help Re: OpenVMS Web Forum 0 Re: Platform migration - password extractions?!? Value of CONSTANT by name? Re: Value of CONSTANT by name? Re: Value of CONSTANT by name?8 VMS Hobbyist VAX. Problems with checksum at installation< Re: VMS Hobbyist VAX. Problems with checksum at installation Re: VMS hobbyist version( Re: Why couldn't linux read a vms drive?  F ----------------------------------------------------------------------  % Date: Sat, 19 Aug 2000 15:13:58 -0400  From: quayle@pobox.com  Subject: Re: 100 MB/sec on 500au. Message-ID: <399EA436.2316.1B32161D@localhost>  e <color><param>0100,0100,0100</param>Okay, my 500au now works at 100 MB.  I replaced the cable -- the  D first one was made from the end of the spool of CAT 5, and a friend 8 suggested that the bend radius was too short for 100 MB.    D Now it passes the autonegotiate without incident at the >>> prompt. C (After doing a SET EWA0_MODE AUTO). And both the 100 MB and duplex  E lights are "on" on the switch.  However, LANCP showed that it's only   really running simplex:          LANCP> sh dev ewa0 /char        Device Characteristics EWA0:   [...]   /                         Yes  Full duplex enable   4                          No  Full duplex operational  ,                 TwistedPair  Line media type  .                         100  Line speed (mbps)      1 You need to set full duplex operation as follows:          LANCP> set dev ewa0 /full   6     LANCP> define device ewa0/update/volatile_database    @ The second step moves it into the permanent database.  When the ; system reboots, it comes back with full duplex operational.       . Case closed.  Thanks, everyone, for your help!       <nofill> --Stan  
 ----------G Stanley F. Quayle, P.E.   N8SQ   +1 614-868-1363   Fax: +1 614 868-1671 1 8572 North Spring Ct. NW, Pickerington, OH  43147 = Preferred address:  stan@stanq.com       http://www.stanq.com    ------------------------------  % Date: Sat, 19 Aug 2000 16:07:53 -0400 ' From: "Bill Todd" <billtodd@foo.mv.com> ! Subject: Re: Big AlphaServer Sale ( Message-ID: <8nmpao$jd0$1@pyrite.mv.net>  4 Main, Kerry <Kerry.Main@compaq.com> wrote in messageD news:910612C07BCAD1119AF40000F86AF0D8052846AE@kaoexc4.kao.dec.com... > David, > J > While I can appreciate that the UNIX based programs that you are running inJ > your environment may not be running as fast as you would like them to orL > perhaps as fast as a native UNIX environment, please do not try to make itL > appear that VMS performance is behind on other programs that were designed1 > to take advantage of VMS architectual features.   K The trouble is, you are both to some degree right, and at the same time you ' are both presenting a biased viewpoint.   G David has real performance problems, but may represent a close-to-worst I case.  But to suggest that VMS performance is for most people competitive L with Unix is also incorrect:  it's just not as abysmally far behind for themL as it is for David, and for a *few* it's probably about equal - but it wouldK be difficult to find any significant population for which it's superior, so   on average it doesn't look good.   > J > As an example, many large and extremely IO intensive banks still use RMSK > databases because they can not get the IO performance that they need from  > databases like Oracle.  J Irrelevant:  it's like comparing the performance of an ocean liner to thatJ of a powerboat.  A hand-tuned RMS-style application will *always* beat theJ pants off an Oracle application, because RMS can take advantage of all theK go-fast tricks Oracle can without any of Oracle's other overhead.  (There's J a possible exception in cases where the same data is updated over and overK and Oracle can use logging to stabilize it more efficiently than writing it C back to disk each time, but in a hand-tuned RMS application you can L special-case such data as long as it's predictable and handle it essentially the same way.)  I That's why I've suggested that a new general cut at data management could H help VMS remain superior:  I suspect a lot of people might prefer a moreG comprehensive, better-performing, easier-to-use RMS-like product to the E Oracle monster, if they had the option (and if it were a cross-system K product).  But as it stands, RMS enjoys no particular performance advantage I over similar ISAM-style products on Unix (save in clustering situations - K and perhaps by now not even there) - and, unless you hand-tailor its use of H global buffers, suffers from its inability to read-share its cached dataJ efficiently among cooperating processes (does VIOC help in that respect?).   > = > >>> It doesn't include small scale technical computing, <<<  > H > While OVMS obviously does not dominate the WS industry, take a look atJ > http://www.genetek.com/ - they have an extremely competitive workstationK > offering (on OpenVMS) because it takes advantage of many of the Alpha and L > OpenVMS specific features. Floating point 64bit performance is a killer onI > the Alpha and their seismic solution takes advantage of these features.   H Why they chose Alpha seems obvious.  It would be interesting to find outL exactly why they chose VMS over Tru64 (Linux wasn't an option back when theyE developed their application), which one would assume would be equally  competitive in performance.   K By the way, please list the 'OpenVMS specific features' they take advantage I of:  I scanned their Web site and found no mention whatsoever of VMS, let H alone any of its features they need.  In fact, I would have assumed they. were using Tru64 if you hadn't said otherwise.   > K > >>> Certain aspects of network performance are also inferior, but only by  > factors of 2 or 3. <<  > K > So how does this explain the success of OpenVMS based solutions that have K > huge network and disk IO like www.northernlight.com (one of the top rated H > internet search engines), e*Trade (huge network and disk IO speaks for< > itself), ISE (Internatinal Securities Exchange) reference:8 > http://www.iseoptions.com/about/technology_compaq.html  G It doesn't.  But neither do such examples refute David's assertion of a L performance gap:  they simply make it clear that cost/performance is not theL most critically important issue for some applications, which may be happy toF scale up using more hardware than they'd otherwise need (with Unix) to( obtain VMS's stability and availability.   > L > >>> But Compaq is doing a rotten job of keeping the Alpha up to speed <<<< > H > Since the 1Ghz x86 is below the cpu integer capabilities of the 833Mhz Alpha K > and is not even close to the FP capabilities of the 667Mhz Alpha, I guess  I  > am missing something.   K Rather, I suspect David phrased his comment poorly - my guess would be that L he meant Compaq has done a rotten job of *maintaining Alpha's lead*, in thatJ Alpha has appeared not to advance much while the IA competition has surged3 ahead and significantly closed the performance gap.   H (I don't track hardware performance history and future road-maps closelyI enough to have an opinion one way or the other here:  I'm just suggesting > where David's - and others' - perceptions may be coming from.)   - bill   ------------------------------   Date: 20 Aug 2000 01:15:55 GMT2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)! Subject: RE: Big AlphaServer Sale , Message-ID: <8nnbgb$chu@gap.cco.caltech.edu>  x In article <910612C07BCAD1119AF40000F86AF0D8052846AE@kaoexc4.kao.dec.com>, "Main, Kerry" <Kerry.Main@compaq.com> writes: >David,  > L >While I can appreciate that the UNIX based programs that you are running inI >your environment may not be running as fast as you would like them to or K >perhaps as fast as a native UNIX environment, please do not try to make it K >appear that VMS performance is behind on other programs that were designed 0 >to take advantage of VMS architectual features.  ; Aargh.  Kerry, enough with this mice versus elephant logic!   I I've spent months trying to wring equivalent disk IO performance out of a K VMS DS10 and just couldn't get it to match that of a Linux DS10.  (And I've F posted these results here before).  Not even using RAMdisks on the VMSI system let it catch up, presumably because the IO overhead in RMS was too J high.   That is the arena I'm discussing - small systems, writing directlyJ to one disk at a time. Every time I bring this up you cite the performanceF of huge systems striping data across a zillion disks using specializedH hardware and man years of programmer work to optimize performance.  TheyK just aren't the same thing and cannot be compared.  That VMS runs well in a J data center is irrelevant for the guy trying to get work through a smaller	 machine.    D This is hardly a novel situation - look at NT - it works well on the) desktop but is a disaster when scaled up.      > I >As an example, many large and extremely IO intensive banks still use RMS ;                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ H == datacenter.  Mice versus elephants.  I'm driving a disk or two, theseK folks are probably spreading the IO across tens if not hundreds of devices.   J >databases because they can not get the IO performance that they need from >databases like Oracle.      > < >>>> It doesn't include small scale technical computing, <<< > G >While OVMS obviously does not dominate the WS industry, take a look at I >http://www.genetek.com/ - they have an extremely competitive workstation J >offering (on OpenVMS) because it takes advantage of many of the Alpha andK >OpenVMS specific features. Floating point 64bit performance is a killer on H >the Alpha and their seismic solution takes advantage of these features.  @ That site doesn't include the string "vms".  In any case, I saidJ specifically that for CPU bound applications VMS and Tru64 are quite closeB (as is Linux/Alpha with the ccc compiler.)  My guess is that theirJ application is very nearly CPU bound so it doesn't matter which of these 3! OS's they use in terms of speed.    J If VMS was really competitive in this end of the computing field you'd be K able to cite hundreds or thousands of such stories.  Circa 1988 that would  K have been the case.  But now you have to look _really hard_ to find VMS in   use in small systems.    > J >>>> Certain aspects of network performance are also inferior, but only by >factors of 2 or 3. << > J >So how does this explain the success of OpenVMS based solutions that haveJ >huge network and disk IO like www.northernlight.com (one of the top ratedG >internet search engines), e*Trade (huge network and disk IO speaks for < >itself), ISE (Internatinal Securities Exchange) reference: 7 >http://www.iseoptions.com/about/technology_compaq.html   G First of all, this is one more of your mice versus elephants arguments. I Secondly, even with that huge VMS machine to work with, they still chose  K not to use VMS for everything.  The web interface at northernlight does NOT  run on OpenVMS:   * $ lynx -head http://www.northernlight.com/   HTTP/1.1 200 OK # Date: Sun, 20 Aug 2000 00:05:47 GMT ) Server: Apache/1.3.9 (Unix) mod_perl/1.21   K Surely it would have been easier to go with a pure VMS solution.  That they K did not is most likely due to the realization that VMS could not handle the I webserver traffic they expected to encounter.  But it might have been due K to some other factors, such as ready access to Apache/Unix programmers, or  H perhaps the desire to save some money and not have to buy the extra VMS @ cluster licenses or machines.  But I'm guessing, if somebody at K Northernlight would care to explain this design decision I'm sure we'd all  % like to hear the reasoning behind it.    > L >>>> But Compaq is doing a rotten job of keeping the Alpha up to speed <<<<  > M >Since the 1Ghz x86 is below the cpu integer capabilities of the 833Mhz Alpha   H The 1 Ghz Athlon is Cint95 42.9, I don't have the numbers for an 833 MhzH Alpha, but for a 466 Mhz DS10 the number is 24.6.  (And yes I know that M Spec has changed their methods yet again - they do it every couple of years.) K And that 1 Ghz AMD machine will cost less than the DS10.  Just because it's H the top of the AMD/Intel line doesn't mean that it should be compared toJ the top of the Alpha line.  A machine with an 833Mhz Alpha in it will costE at least 5X more than the 1 Ghz x86 machine will.  The alpha isn't 5X F faster though, so the price/performance leadership is very strongly inI AMD/Intel's favor.   (Note also that the performance bottlenecks I see in ( VMS have nothing to do with CPU power!)   L >and is not even close to the FP capabilities of the 667Mhz Alpha, I guess I >am missing something.    G Yes, the COST of the Alpha system.  As I said - if cost is no problem,  I people buy Alphas (and Tru64).  But if you're discussing parallel systems L and building blocks, the price/performance advantage does not belong to the H Alpha.  (Has it ever?  From the very beginning it was always "twice the ' performance at three times the price.")   K >The API systems are competitive in pricing with equivalent 1Ghz WS systems 
 >as well.   K There are a couple of machines downstairs with API guts - and they've been  K nothing but trouble.  Especially the dual CPU system that couldn't be made  D to run at its rated speed and they had to "dumb it down" to 500 Mhz G processors.  At least the Digital and Compaq Alphas have been reliable.   B >Keep in mind the 1ghz x86 cpu's are WS or small 1 or 2 cpu server >offerings only right now.  I That's exactly my point - that's the part of the computing market I'm in! J We "mice" use systems like that.  The vast majority of all computers have I no more than 2 processors.  And VMS needs real work to be competitive on  H these systems.  Period.  (And I do not mean to imply that VMS's failuresG on these smaller systems in any way means that it doesn't work well on  G "elephant" systems where the computing environment is quite different.)    > G >In addition, for technical computing where FP is used quite a bit, the H >600Mhz DS10 is better than the 1Ghz x86 in FP capabilities (1Ghz=335 vs >DS10=411 vs API 833Mhz=644).   3 When did the 600 Mhz DS10 ship?  Got a part number?   G It may beat the the Intel/AMD crowd in FP, but it won't be as fast for  J integer math.  What I do is technical computing, and it's almost entirely K integer math.  Similarly, it's the very rare business program that isn't at I least 75% integer.  So file servers, web servers, most databases, and the J like, won't be as fast on that more expensive alpha as they will be on theH less expensive Intel/AMD machine.  (Caveat - memory bandwidth counts forI some applications, and the Compaq Alphas do tend to have very good memory  bandwidth's.)    > L >Reference the new standard benchmark for measuring CPU, memory and compiler
 >performance: 5 >http://www.spec.org/osg/cpu2000/results/cpu2000.html  >   H Ok, the new standard does fold in the memory bandwidth problems on some ' systems better.  Here are some numbers:   2 CINT2000                             CPU base peak2  AlphaServer DS10 6/600              1    345  3642  Intel VC820 (1.0 GHz Pentium III)   1    407  4102  Microstar K7T Pro motherboard 1GHz  1      ?    ?   CFP2000 2  AlphaServer DS10 6/600              1    377  4112  Intel VC820 (1.0 GHz Pentium III)   1    273  2842  Microstar K7T Pro motherboard 1GHz  1    290  302  J The DS10 VMS machine costs what, $6500?  I'd estimate that a machine basedI on the VC820 should be available for about half that.  (This for systems  J with similar memory, disk controllers, and disks.)   The PIII is faster atK integer math than the 600 Mhz DS10, and vice versa for floating point.  The J Intel machine (running Linux or WNT) will be (or appear to be) much faster. for disk and network IO than the VMS machine.     M >>>> so I don't know how longer they'll be able to hold onto the Linux end of  >the market. <<< > K >Well, if you think Compaq plans to ignore Linux, check out these articles: 3 >http://www.techweb.com/wire/story/TWB20000814S0018lK >http://alllinuxdevices.com/news_story.php3?ltsn=2000-07-05-001-02-NW-LF-CM0  H I didn't say they were going to ignore it.  I said that they are losing E ground in the price/performance arena and have already been surpassed I (outright) in integer performance.  Yes, the Alpha is still faster in FP s< and still costs a lot more to obtain that extra performance.   >TM >Well, how about another example - perhaps these folks are doing this becauseh$ >they have nothing better to do eh ?? >http://www.informix.com/ (see announcement on home page today)v< >http://www.informix.com/informix/press/2000/aug/linux.html J >"Informix, Compaq and SuSE Team to Deliver Alpha Linux Computing Platform >for e-Business"  I Ok, informix thinks there's money to be made there.  (They've been wrong D@ before...)  And how many Informix products do we see on OpenVMS?   >a >or these folks:5 >http://www.turbolinux.com/news/pr/alpha-release.htmliC >"TurboLinux Launches High-Performance 64-Bit OS for Alpha Systems"u  G Suse, Turbolinux, and RedHat all think there's money to be made sellinghL Linux/Alpha distributions.  Good for them.  And if Compaq and Samsung don't G deliver on faster Alpha's they'll just drop these lines and go back to c" their core, Intel based, business.   > G >>>> in the last 6 months the speed of Athlons and PIII's has more thane >doubled. << >EK >Wrong. Are you saying 6 months ago, the highest available chip was 500Mhz?   B That was a rough date.  I've got a Nov 30 1999 Microtimes with adsK indicating that 500 Mhz was the fasted PIII that was widely available then. I So 8.5 months, not 6.   Today was the first I've heard of a 600 Mhz DS10.v   > + >833Mhz Alpha API rating = 533 int; 644 FP.e >lD >Based on above, I fail to see how you feel Alpha has not kept up in >performance.   I There's that elephant again.  What does a 1 Ghz Intel or AMD system cost?eG What does an 833 Mhz ES40 cost?  (That's the only system I know of that-G has SPEC values for that processor.)  Who cares how fast an ES40 is if  , you've only got money for something smaller?   Regards,   David Mathog mathog@seqaxp.bio.caltech.edu.? Manager, sequence analysis facility, biology division, Caltech     ------------------------------    Date: 19 Aug 2000 18:54:54 -05009 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen)i' Subject: Re: Changing address on MV3100p+ Message-ID: <rPp15RNZyFum@eisner.decus.org>1  Z In article <399EA0A5.F91B1490@tsoft-inc.com>, David A Froble <davef@tsoft-inc.com> writes: > Balexi Yurguat wrote:h >> u
 >> Anyone? >>  0 >> How would I change my _dkb100: to dkb400: ??? > Q > Some details of just what dkb100: is, type of disk, how attached, etc.  This is Q > assuming you want to change the SCSI ID.  Usually involves moving a jumper from M > the 1 position to the 4 position.  The disks are addressed, for SCSI ID, in ! > binary using 3 sets of jumpers.a > O > Or, if you just have a program with "dkb400:" hard coded somewhere, you couldp1 > define a logical dkb400 with a value of dkb100.o  A But isn't the purpose of the leading underscore to defeat logicalo name rediorection ?    ------------------------------  % Date: Sat, 19 Aug 2000 18:42:28 -0400s* From: David A Froble <davef@tsoft-inc.com>' Subject: Re: Changing address on MV3100 - Message-ID: <399F0D54.2F22EB12@tsoft-inc.com>s   Larry Kilgallen wrote: > \ > In article <399EA0A5.F91B1490@tsoft-inc.com>, David A Froble <davef@tsoft-inc.com> writes: > > Balexi Yurguat wrote:h > >> > >> Anyone? > >>2 > >> How would I change my _dkb100: to dkb400: ??? > > S > > Some details of just what dkb100: is, type of disk, how attached, etc.  This isrS > > assuming you want to change the SCSI ID.  Usually involves moving a jumper from O > > the 1 position to the 4 position.  The disks are addressed, for SCSI ID, in # > > binary using 3 sets of jumpers.d > >hQ > > Or, if you just have a program with "dkb400:" hard coded somewhere, you could 3 > > define a logical dkb400 with a value of dkb100.i > C > But isn't the purpose of the leading underscore to defeat logicalp > name rediorection ?c  
 Picky! Picky!a  6 Actually, yes, the underscore does defeat translation.  ; I'm guessing the poster really wants to change the SCSI ID.t  O However, when reading the question, it appears to me that he has a disk DKB100,gL but wants to access it it using the 'name' DKB400, and therefore the logicalM name WOULD be DKB400, with a value of DKB100, and therefore the logical wouldsI still point him to the device DKB100.  There was no underscore on DKB400.a  O Damn!  I'm good!  Wonder how that happened?  I'm usually the one who misses the  details.   Dave   -- e4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com6 T-Soft, Inc.  170 Grimplin Road  Vanderbilt, PA  15486   ------------------------------  % Date: Sat, 19 Aug 2000 18:44:19 -0400   From: John Santos <JOHN@egh.com>4 Subject: Re: Cluster transaction processing paradigm6 Message-ID: <1000819181628.40860A-100000@Ives.egh.com>  $ On Wed, 16 Aug 2000, JF Mezei wrote:  M > I am trying to think of an application architecture which would not only beaO > cluster friendly (easy) but also make maximum use of clustering to distributer? > processing and provide transparent operation if a node fails.i > ; > Transactions go into an indexed file in a random fashion.' > N > Transactions are extracted from the file in an ordered fashion and sent to aO > third party (by modem) in batches of say 30 transactions at a time (or less).VO > There are various third parties involved, so the reception of 30 transactionsnP > might mean one phone call if they are all destined to the same destination, or > multiple phone calls.h > ) > FEEDING TRANSACTIONS TO THE QUEUE-FILE:s > B > To feed the transactions into the queue file, I see 3 paradigms: > 	 L > -client application itself opens the queue file and writes to it. However,I > this still requires that the client contact some server to get a unique. > transaction number.o > P > -client application uses ICC to contact a single server process on the clusterL > which provides the transaction number and it stores the transaction in theW > queue file. However, if the node on which that server runs fails, then you're cooked.e > K > -client application submits to any number of servers on the cluster. EachtN > server then provides a unique transaction number (with its node name as part- > of it) and writes to the single queue file.  > G > Question: what technologies exist right now that would allow multipledP > processes (one per node) to advertise themselves as being available to receive< > transactions from any process on any node of the cluster ?  I The VMS lock manager:  Store the current transaction ID in the lock valueaG block.  When a node boots, have it start up a server process that takes H out an "interest lock" (I forget the exact terminology) in the resource.I When it gets a request for an transaction number, it converts the lock to J exclusive mode, grabs the lock value, increments it, assigns the new valueH as the transaction number for the request, store the new lock value, and convert the lock back.  G I think you need to keep the highest transaction number around on disk tJ somewhere to restore in the event of a total cluster shutdown, and also I H think it is possible to lose the lock value block if the node mastering H the lock crashes (the lock manager will tell you if this happens.)  Two I solutions:  record the lock value in a shared file every time it updates .H (expensive in disk I/O) or search your indexed file of transactions for J the highest value used, and init the lock value block to that one.  First H accessor of the resource can tell if it didn't already exist and can do J this on a cluster boot.  First user of the lock value block who discovers J that the lock value has disappeared can be responsible if the lock master 
 node crashes.   H A low-cost variant on the 1st solution:  If you don't mind holes in yourG transaction number space, have the servers update the disk copy of the tF transaction number only every 100 allocations.  Then if the value getsG lost, use the stored (on disk) transaction ID + (100*number of nodes),  C guaranteeing a new unique value.  Only 100th the disk overhead, and  quick recovery.n  I P.S.  To prevent race conditions, update the disk copy of the transaction 0 number while the resource is exclusively locked.   >  >  > SERVER COORDINATION  > P > If I have multiple server processes on multiple nodes on the cluster, is thereI > some technology which allows these to talk to each other in a broadcastzK > fashion ? (eg: server 1 send a single message that is received by all the J > servers). And if a new server joins, then it too starts to receive thoseK > "broadcast" messages. (and ideally, when one leaves, all others receive a  > message of it leaving).y  " Again, the lock manager does this.  & > TRANSACTION CONSUMPTION/LOAD SHARING > H > Ideally, I'd like to have the multiple servers intelligently share theK > transactions. While one server has "reserved" some 30 transactions to one^M > remote partner, the other server reserves the next 30 transactions (to someeG > other or perhaps same partner) and so on. This way, the throughput isvQ > maximised, and if one node fails, then the servers on the other nodes continue.m > L > However, how can one distribute transactions amongst those servers withoutJ > having a single server coordinate this (which represents single point of > failure if that node fail?  % The lock manager comes through again.p   > N > For instance, during slow period, if a single transaction is received, whichK > server will get to pick that one up and deliver it as fast as possible ?   > P > Are there any technologies which do this type of stuff (or papers that discuss
 > this) ? ? ?s  G There is (or was) a book in the VMS doc set called "Building DependablePF Systems: The OpenVMS Approach" (AA-PV5YB-TE), that discussed all this.G This manual is listed as "archived", but still orderable in the VMS 7.2yC "Overview of OpenVMS Documentation".  Also look at "OpenVMS ClusterrI Systems" (in the doc set, may have replaced all the info in the "BuildinggE Dependable Systems" book), and the system services manual for detailsC of calling the lock services.e   -- a John Santosy Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  # Date: Sat, 19 Aug 2000 18:56:58 GMT  From: rocoto@my-deja.com1 Subject: CRE/TER/DET challenge - hackaround foundl) Message-ID: <8nml9i$bg6$1@nnrp1.deja.com>t  E A while back, I posted a query about setting the icon position from aeE CRE/TER/DETA command...result was no solution found or suggested. I'mO3 still looking for a better answer than this, but...E  F I can do this with the -xrm switches on such things as decw$clock, butE not CRE/TER.... All of the klunky variations of PIPE I could think ofgC didn't work - and writing a per-command custom resource file seemedy( pretty ugly too... So I kept looking....  D Anyway, I found a really ugly hackaround & thought I'd pass it on inG case anyone cares.... Besides, I can use this as another opportunity to,H troll around for a better solution (which I would dearly appreciate)....  G The trick I found was to overload the input to an existing command line * resource redefinition, title in this case:   $ lf[0,8]=10&   $ xpos="''lf'DECW$TERMINAL.*.iconX:"&   $ ypos="''lf'DECW$TERMINAL.*.iconY:"5   $ cre/ter/deta/win=title=".''xpos' 1154''ypos' 770"c  G Now since my login clobbers the win title first thing, a starting title ! of '.' isn't a problem for me....    ugly ugly ugly.... :)n -- Davidm rold5@tditx.come    & Sent via Deja.com http://www.deja.com/ Before you buy.i   ------------------------------  % Date: Sun, 20 Aug 2000 04:16:58 +0930>/ From: Mark Daniel <mark.daniel@wasd.vsm.com.au>A Subject: DECnet area routing/ Message-ID: <399ED622.A78F00D5@wasd.vsm.com.au>l  '        +------+    +------+    +------+l'  A/1   |      |    |      |    |      |r'        +------+    +------+    +------+z$            |-----------|-----------|                        |                     +----+!                     +----+ router                         |)      ==================|=================u                        |$            |-----------|-----------|'        +------+    +------+    +------+l'  B/2   |      |    |      |    |      |t'        +------+    +------+    +------+h    B Systems on network A (say area 1) need to be able establish DECnetC task-to-task connections for data transfer to systems on network B.aC Network A cannot allow any possibility of network connectivity fromo@ network B (say area 2).  Can I configure DECnet routing so as toD provide a route from A->B (area 1 to 2) but not from B->A (area 2 to; 1)?  I think this is a generic DECnet routinq question, not, router-vendor-specific.d   As always TIA.   --B "This electronic message and any attachments could not possibly be@ confidential otherwise it would not be sent unencrypted, via theA Internet, through countless mailing agents, gateways, and using aeA plethora of other completely insecure and extremely public media. B Even if you are not the intended recipient of this message you areC free to use and abuse any and all of the contents as everyone knows D full-well that that with zero security available the only conclusionA can be zero accountability.  Virus scanning software was not usedrC composing or mailing this message, end-use automatically nullifying C liability for infestation or subversion caused by opinions or ideasr it may contain."   ------------------------------  % Date: Sat, 19 Aug 2000 22:48:55 -0400r2 From: "Richard B. Gilbert" <DRAGON@compuserve.com> Subject: DECnet area routing7 Message-ID: <200008192249_MC2-B04E-8D03@compuserve.com>r  H         What you are asking for is seemingly a contradiction in terms. =  G DECnet task-to-task is inherently bidirectional  and therefore requireso A-->B and B-->A.  J         You could possibly configure a machine in place of the "router" t= o J allow A to initiate a transaction and B to respond, without allowing B to=  G initiate communications.  This would require something different than aDG simple router.   What you need is more like a firewall but I have neverr heard of a DECnet firewall.     < Message text written by INTERNET:mark.daniel@wasd.vsm.com.au > =   '        +------+    +------+    +------+ '  A/1   |      |    |      |    |      |y'        +------+    +------+    +------+t$            |-----------|-----------|                        |                     +----+!                     +----+ routere                        |L      =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D|=3D=3D=3D=3D=3D=$ =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D                        |$            |-----------|-----------|'        +------+    +------+    +------+o'  B/2   |      |    |      |    |      |r'        +------+    +------+    +------+0    B Systems on network A (say area 1) need to be able establish DECnetC task-to-task connections for data transfer to systems on network B.eC Network A cannot allow any possibility of network connectivity fromv@ network B (say area 2).  Can I configure DECnet routing so as toD provide a route from A->B (area 1 to 2) but not from B->A (area 2 to; 1)?  I think this is a generic DECnet routinq question, not  router-vendor-specific.<   ------------------------------  # Date: Sat, 19 Aug 2000 18:22:49 GMT ( From: Terry Kennedy <terry@gate.tmk.com>- Subject: Re: DECserver 700 documentation/helps' Message-ID: <FzJxq1.9sJ@spcuna.spc.edu>a  : In comp.os.vms David A Froble <davef@tsoft-inc.com> wrote:R > I've got a used DECserver 700, no documentation.  I've looked on both COmpaq andR > Cabletron sites for such, and no luck.  I am rather poor at finding stuff on web& > pages, so the fault's probably mine.  <   No, DEC/Compaq's web pages are incredibly badly organized.     The software manuals are at:0 http://www.networks.digital.com/dr/dnas/manuals/     There's more info at:p0 http://www.networks.digital.com/dr/hubs/servers/  K   But the DS700 manual isn't there, for some unknown reason. There's a copye at:/? http://www.networks.digital.com/dr/servers/manuals/dsrvw-mg.pdft  Q > When powered up, connected with a 10baseT connection, I get no response on portrR > 1, terminal set to 9600 baud.  The green light by the 10baseT connection blinks,P > indicating (I'd think) network activity.  The light by the power cord is off. Q > The led blinks, what I'd call an 'A' without the cross bar, or maybe an 'n', ord) > maybe it's upside down, and it's a 'U'.e  M   The green LED near the network port doesn't show anything useful if the LEDhN by the power cord is off or blinking. The fact that the second LED is blinking  says that the server is unhappy.  L   The error you have is described in the last manual I listed above, on pageO 73 of the PDF (manual page 7-9). It says "No SIMs or wrong type SIMs installed"dL so it's likely somebody mugged your DS700 for its memory. It's a generic 72-J pin SIMM - as I recall, it's FPM, parity (x36), gold pins, 80ns or faster.  - 	Terry Kennedy             http://www.tmk.coma5         terry@tmk.com             Jersey City, NJ USA    ------------------------------  % Date: Sat, 19 Aug 2000 20:25:04 +0100n( From: "Nick de Smith" <nick@desmith.net> Subject: Re: OpenVMS Web ForumA Message-ID: <966713079.10724.0.nnrp-13.9e98156e@news.demon.co.uk>   ; Richard B. Gilbert <DRAGON@compuserve.com> wrote in message 1 news:200008181854_MC2-B033-629F@compuserve.com...c [...]   L FWIW, I may be one of the few people who actually won an argument with Carl.= Not that that was ahigh point of my career, you understand...i   -- Nick nick@desmith.net   ------------------------------  % Date: Sat, 19 Aug 2000 14:37:22 -0400l- From: JF Mezei <jfmezei.spamnot@videotron.ca>o9 Subject: Re: Platform migration - password extractions?!?n, Message-ID: <399ED3DE.D2AA41CD@videotron.ca>   Ok here is another possibility:s  K Sine the VMS site is used as an ISP (correct ?), chances are that customersf. are getting at their mail through POP3 right ?  I Since POP3 supports clear text sending of username/password, (can the POP?L server force this ?), you might be able to intercept such passwords as usersF pickup their mail, and then store it for the move to the new platform.  G However, I think that in the end, what you will have to end up doing issM getting each of the customers on VMS to access some web page to confirm theirgJ details and in that web page, you capture their passwords, have it checkedH against the vMS SYSUAF for validity, and if valid, you then transfer the  informtion over to the unix box.  K And you keep sending nasty emails to customers who are still on the VMS boxa until they switch.  L And in the end, you just move the remaining users and send them a paper mail with their new password.   ------------------------------   Date: 19 Aug 2000 20:17:54 GMT6 From: DAVISM@er6.eng.ohio-state.edu (Michael T. Davis)# Subject: Value of CONSTANT by name?p9 Message-ID: <8nmq1i$fb$1@charm.magnus.acs.ohio-state.edu>e  M         Is there a means that an HLL (C, FORTRAN, etc.) routine can determineaM the value of a "constant", like the item codes for $GETDVI, $GETJPI, $GETUAI,-J etc. by name, without an internal table of name/value pairs?  For example,N could I create a routine similar to LIB$GETJPI that specified the item code by@ name, instead of by (reference to) a value?  That is, instead of  H         LIB$GETJPI ( & JPI$_ACCOUNT, 0, 0, 0, & account_desc, & a_len );  # I'd like to do something like this:h  +         char item_code [] = "JPI$_ACCOUNT";a)         $DESCRIPTOR(ITEM_CODE,item_code);o  D         my_getjpi ( & ITEM_CODE, 0, 0, 0, & account_desc, & a_len );  O Presumably, this would call for a database of some kind that would be availablepM under VMS, regardless of platform or version.  Does such a database exist?  ICI considered mapping the appropriate text library module (e.g. $JPIDEF) and L trying to parse it, but I had serious doubts that I could rely on the format, of these remaining constant in all contexts.   Thanks,h  Mikes --K              Michael T. Davis              |    Systems Specialist: ChE,MSEPN   E-mail: davism@er6.eng.ohio-state.edu    | Departmental Networking/ComputingJ            -or- DAVISM+@osu.edu            |     The Ohio State UniversityJ http://www.er6.eng.ohio-state.edu/~davism/ |     197 Watts, (614) 292-6928   ------------------------------    Date: 19 Aug 2000 20:04:44 -05009 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen)a' Subject: Re: Value of CONSTANT by name?K+ Message-ID: <CW2Wrge464Bf@eisner.decus.org>   r In article <8nmq1i$fb$1@charm.magnus.acs.ohio-state.edu>, DAVISM@er6.eng.ohio-state.edu (Michael T. Davis) writes:  M > I considered mapping the appropriate text library module (e.g. $JPIDEF) andxN > trying to parse it, but I had serious doubts that I could rely on the format. > of these remaining constant in all contexts.  C You could take the Freeware SDL sources and write a custom back end4 to emit the codes.   ------------------------------  # Date: Sun, 20 Aug 2000 02:49:03 GMTa= From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) ' Subject: Re: Value of CONSTANT by name? 0 Message-ID: <009EEDB8.22CEA4F4@SendSpamHere.ORG>  r In article <8nmq1i$fb$1@charm.magnus.acs.ohio-state.edu>, DAVISM@er6.eng.ohio-state.edu (Michael T. Davis) writes: >gN >        Is there a means that an HLL (C, FORTRAN, etc.) routine can determineN >the value of a "constant", like the item codes for $GETDVI, $GETJPI, $GETUAI,K >etc. by name, without an internal table of name/value pairs?  For example,eO >could I create a routine similar to LIB$GETJPI that specified the item code by A >name, instead of by (reference to) a value?  That is, instead ofo >cI >        LIB$GETJPI ( & JPI$_ACCOUNT, 0, 0, 0, & account_desc, & a_len );c >s$ >I'd like to do something like this: >U, >        char item_code [] = "JPI$_ACCOUNT";* >        $DESCRIPTOR(ITEM_CODE,item_code); >iE >        my_getjpi ( & ITEM_CODE, 0, 0, 0, & account_desc, & a_len );  >hP >Presumably, this would call for a database of some kind that would be availableN >under VMS, regardless of platform or version.  Does such a database exist?  IJ >considered mapping the appropriate text library module (e.g. $JPIDEF) andM >trying to parse it, but I had serious doubts that I could rely on the formatu- >of these remaining constant in all contexts.p >c >Thanks, > Mike >--mL >             Michael T. Davis              |    Systems Specialist: ChE,MSEO >  E-mail: davism@er6.eng.ohio-state.edu    | Departmental Networking/ComputingiK >           -or- DAVISM+@osu.edu            |     The Ohio State UniversityiK >http://www.er6.eng.ohio-state.edu/~davism/ |     197 Watts, (614) 292-6928o  H I've essentially done this in SYMBOL.  You pass the name of an executiveH symbol by name and SYMBOL creates a DCL symbol of the same name with theJ value of the executive symbol assigned to the DCL symbol of the same name.  \ http://www.tmesis.com/help/@sys$sysdevice:[SYMBOL]SYMBOL/SYMBOL/QUALIFIERS/SPECIAL#EXECUTIVE] http://www.tmesis.com/help/@sys$sysdevice:[SYMBOL]SYMBOL/SYMBOL/QUALIFIERS/SPECIAL/PARAMETERSm   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMw   ------------------------------  % Date: Sat, 19 Aug 2000 22:40:14 +0200g? From: Fim =?iso-8859-1?Q?W=E4stberg?= <fim.wastberg@fimator.se> A Subject: VMS Hobbyist VAX. Problems with checksum at installations) Message-ID: <399EF0AE.F41255B@fimator.se>d  @ I am installing the hobbyist VMS on a MicroVax Software Emulator% but I have problem vith the checksum:t5 ----------------------------------------------------- H    The license checksum is a 17-character verification string created by  D    the PAK issuer for each PAK.  The checksum string is presented inE    the format n-cccc-cccc-cccc-cccc, where n is an integer and c is anG    character from A through P.  A PAK presents the checksum string withcC    hyphen (-) characters for readability.  Because the LMF does not  countsG    them for authorization, you can leave them out.  Otherwise, you mustoF    enter the checksum string exactly as specified on your PAK or PAAM.  D    If a default value is displayed and you wish to use it just press=    the RETURN key.  To cancel the use of default data withoutc=    entering new  data, enter the backslash (\) character. TheaB    license checksum is a required field for the REGISTER and AMEND    options.c5                      Checksum []:1-YOUR-PAKS-CHEK-SUM   < 1-YOUR-PAKS-CHEK-SUM is not a valid license checksum string.! Press RETURN for more information 3 ---------------------------------------------------uF The checksum "1-YOUR-PAKS-CHEK-SUM" is from the OpenVMS Hobbyist GuideH but can it be correct ??. Just "character from A through P" are allowed.    E Of course, I can walk around it, but then I can only use the Console.s  / Anybody out there who had solved the problem ??d       --# Med vnliga hlsningar/Best regards 
 FIMATOR AB Fim Wstberg CPIM E ---------------------------------------------------------------------A Box 63 177 22  JRFLLA SWEDEN- Tel   : 08 580 277 00 Abroad +46 8 580 277 00eE Fax   : 08 584 116 90        +46 8 584 116 90 <=== NEW per 1-Feb-2000t, Mobil : 070 810 73 91        +46 7 810 73 91 e-mail: fim.wastberg@fimator.seo   ------------------------------  % Date: Sat, 19 Aug 2000 17:46:16 -0400t* From: David A Froble <davef@tsoft-inc.com>E Subject: Re: VMS Hobbyist VAX. Problems with checksum at installatione- Message-ID: <399F0028.88466220@tsoft-inc.com>w   Fim Wstberg wrote:l > H > The checksum "1-YOUR-PAKS-CHEK-SUM" is from the OpenVMS Hobbyist GuideJ > but can it be correct ??. Just "character from A through P" are allowed.  E That's not the actual checksum.  That is a way of telling you, in thewP documentation, to enter the unique PAK checksum that you received as part of theM hobbyist license.  You should have a license PAK, of some format, and it will N have the correct responses to set up your license.  Use your unique (and don't; post it here) PAK information when registering the license.s   Dave   -- h4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com6 T-Soft, Inc.  170 Grimplin Road  Vanderbilt, PA  15486   ------------------------------  % Date: Sat, 19 Aug 2000 23:06:32 -0400e) From: yyyc186.illegaltospam_@flashcom.netn! Subject: Re: VMS hobbyist version69 Message-ID: <399f4b4c$1$lllp186$mr2ice@news.flashcom.com>1  : In <eB_m5.135$M62.51900@typhoon.aracnet.com>, on 08/19/00 B    at 11:06 PM, "Zane H. Healy" <healyzh@shell1.aracnet.com> said:  + >yyyc186.illegaltospam_@flashcom.net wrote:eJ >> According to the link I looked at on www.decus.org the VAX distribution& >> comes on TK50 and the alpha CD rom.  9 >What?!?!  I assume you followed it to the following URL: F >http://www.montagar.com/hobbyist/media.html which specifically statesH >that the VAX and Alpha Hobbyist distro's are on CD-ROM, and that a TK50I >distro is in the works (well, it's said that for over a year I believe).o  B The order form had a checkbox and price for the VAX version on tk.   Roland   >			Zane -- n; -----------------------------------------------------------pD yyyc186@flashcom.net              To Respond delete ".illegaltospam"6                             MR/2 Internet Cruiser 1.528                             For a Microsoft free univers; -----------------------------------------------------------E   ------------------------------    Date: 19 Aug 2000 19:31:36 -0400/ From: jordan@lisa.gemair.com (Jordan Henderson) 1 Subject: Re: Why couldn't linux read a vms drive?a* Message-ID: <8nn5co$8k6$1@lisa.gemair.com>  / In article <sptfnctn87v176@corp.supernews.com>,d( John E. Malmberg <wb8tyw@qsl.net> wrote:2 ><kenn#avalon.research.wombat.ie> wrote in message6 >news:slrn8pt44n.1pu.kenn@avalon.research.wombat.ie...8 >> On Fri, 18 Aug 2000 19:14:32 -0500, David J. Dachtera) >>   <djesys.nospam@earthlink.net> wrote:2 >> >"John E. Malmberg" wrote: 
 >> >[snip]J >> >> Let see, now known for LINUX, you can get at least one DCL emulator, >DECNet-K >> >> Phase IV, ODS-2, and someone reported recently that LAT is available.  >> >K >> >I thought LAT was a licensed protocol with no public specification. has ; >> >someone finally been successful at reverse-engineering?u >>= >> Yes.  Patrick Caulfield from the Linux-DECnet team did it.  > L >The issue is not that it could be done, is that anyone that does not have aF >license from the patent holder to do so could be in legal difficulty. >a+ >Unless the patent in question has expired.-  @ Are there any patents involved really?  Or were the LAT licensesD just licenses to integrate parts of the copyrighted LAT sources into	 products?e  @ I know Meridian was a licensee for LAT and produced SuperLAT for? Solaris at one time.  I think they still have some LAT printingi< products, but I believe that they only work from Windows NT.  8 I'm pretty sure that a number of terminal server vendors: reversed engineered LAT in the 80's-90's and I don't thinkC that they held any licenses from Digital.  I believe the copyrightsn8 splashed all over the help in DECserver terminal servers> was the result of Digital trying to fend off this competition.  = Does anyone remember Digital ever suing over LAT patents?  I t= don't.  I imagine there are _some_ patents on LAT technology, ; but they might not really represent anything worth fightingc over in court.   >c >-John >wb8tyw@qsl.network  >t >e >s   -Jordan Henderson  jordan@greenapple.com,   ------------------------------   End of INFO-VAX 2000.464 ************************ions or ideasr it may contain."   ------------------------------  % Date: Sat, 19 Aug 2000 22:48:55 -0400r2 From: "Richard B. Gilbert" <DRAGON@compuserve.com> Subject: DECnet area routing7 Message-ID: <200008192249_MC2-B04E-8D03@compuserve.com>r  H         What you are asking for is seemingly a contradiction in terms. =  G DECnet task-to-task is inherently bidirectional  and therefore requireso A-->B and B-->A.  J         XImٞ[6[uPi˓+ü>vY㶵=n3lph1<œ
7o%\#ـ-^c1_|Nn=*~ dP<G1t׽	9.+:,\ή`<k)
oHYW몠C+*imumR
xoZ"dڧUc]|oNd	_I]_F:NQd[p7eL,	1<-쯼={5;.}GJx.DfD	~{ܢxxbK?g9F㻌H^l^e?;Q28"+Tȝ?L߈߂hsٓ/,ޣU9]Uw>Ni9y#oHI?cw~ێtp3oa2_ەw)۬i廇_{y=L6fM*L΍}PC8F;+{$yLr$<mNB>q&i|oqVPR)|C>wM%i;a%; j_1>HƲnFmSHci9x}3*份P_ߨ^}Kl$̜ZTo%ԻMb[׺(Ϫ{$mOT6U@:&xI[	o>J{V+{  ONqHXtއ" R6L4m>YZ^GorZ9o;e}ߙg*bD[K*ր6|A4U$m 96xO|kS7X.w67$LC1^<u@SXcgTvk?Uu~$Ν%i;s6YMsK'3UuI忷߮GO߭UUB:4yWr{=OW9juMT=vA||s
ai.׹o
t!<h9v5;aGGSƲӡ}"2}P||K'¾?*Y_#OTÜ[hg'ex8UZVL,<g'l2ssl7Z?[
2sO[mJ8:vs'Ez\lbu4`p
G_Je⽮}+;+71N~Lⷺ*@^vvќ5컙5K6!Z.w<H7ܞ!
?óɚ:w
G1odU$nWWaANFSͪ|Z =l<?U,aEx>o>q|o;$|h FtQ'/bCunm,ѷV_LxEiސw̹yC'8@vJ D'k,Q%gb&2faAUol-s0OAH|>+y%'XI]cU| jh5q(;?쑒O&4Vϲ-uBxFS%=WHZ29A<_yh1OI/SڽI?W-d{Eo~+CVázhbu쎳=O`yYR筥xyp O&8lvuVG
Nlh,;6Qgi1V'Ę]ˣ	XI]$Y!Eof
]Μ4׉a&Hw/X$NlמkZ
16|5wVlBٵOZ)">#]BPA(a[ˎw@#