1 INFO-VAX	Thu, 14 Mar 2002	Volume 2002 : Issue 143       Contents:< RE: (OT) US/Europe living costs, was: Re: VMS sys admin sala> A Hybrid Approach: Codasyl DBMS Data Warehousing on Oracle RdbB Re: A Hybrid Approach: Codasyl DBMS Data Warehousing on Oracle Rdb* Re: Another big block votes against merger* Re: Another big block votes against merger* Re: Another big block votes against merger* Re: Another big block votes against merger* Re: Another big block votes against merger* Re: Another big block votes against merger* Re: Another big block votes against merger Re: Antigen found =*.exe file  Re: COMPACTION Re: COMPACTION+ DEC C: why does exit(0) really exit with 1? / Re: DEC C: why does exit(0) really exit with 1? / Re: DEC C: why does exit(0) really exit with 1? / Re: DEC C: why does exit(0) really exit with 1? ! decus uk - confusion over joining % Re: decus uk - confusion over joining % Re: decus uk - confusion over joining 
 Re: DFO Issue 
 Re: DFO Issue  Re: DLT Degausser / Bulk Eraser  Re: DTTS on Unix Machine Re: Dynamic IP address question  HSZ10 looking for info...  Re: HSZ10 looking for info... = Re: I may have to case-mod an Alpha. Anyone been here before? = RE: I may have to case-mod an Alpha. Anyone been here before? = Re: I may have to case-mod an Alpha. Anyone been here before?  Re: Itanium troubles+ Re: Microsoft Curries Favor With Undergrads . Re: Mimer SQL for OpenVMS Itanium Announcement& Opening a SYS$OUTPUT file in VMS Basic* Re: Opening a SYS$OUTPUT file in VMS Basic* Re: Opening a SYS$OUTPUT file in VMS Basic* Re: Opening a SYS$OUTPUT file in VMS Basic  Re: OpenVMS jobs available - FYI" Oracle Context (interMedia) on VMS& Re: Oracle Context (interMedia) on VMS- Re: sending all data as out-of-band in TCP/IP @ TCPIP, sharing sockets via $QIO (programming, long and detailed)D Re: TCPIP, sharing sockets via $QIO (programming, long and detailed)@ TCPIP, sharing sockets via $QIO (programming, long and detailed)D Re: TCPIP, sharing sockets via $QIO (programming, long and detailed)D Re: TCPIP, sharing sockets via $QIO (programming, long and detailed)D Re: TCPIP, sharing sockets via $QIO (programming, long and detailed); RE: Transferring files from one OpenVMS machine to another. ; Re: Transferring files from one OpenVMS machine to another. ; Re: Transferring files from one OpenVMS machine to another. ; Re: Transferring files from one OpenVMS machine to another.  Re: VFC File Problem8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS8 Re: Viability of a Commercial ISO-9960 formatter for VMS, Re: Why not we guys buy OpenVMS from Compaq?, Re: Why not we guys buy OpenVMS from Compaq?@ Re: [OT]  NY Times Op-Ed misses point about unreliablity of PC's@ Re: [OT] US/Europe living costs, was: Re: VMS sys admin salaries  F ----------------------------------------------------------------------  % Date: Wed, 13 Mar 2002 13:59:28 -0500 * From: WILLIAM WEBB <WWEBB1@email.usps.gov>E Subject: RE: (OT) US/Europe living costs, was: Re: VMS sys admin sala - Message-ID: <0033000056254838000002L082*@MHS>   = =0AI try to avoid these threads but feel compelled to comment  once:   : Employer-provided medical benefits in the US date from the6 World War II period.  Wages and prices were frozen, so; employers provided this coverage as an additional incentive 6 since the supply of workers was constrained due to the; number of people who were in the armed forces at that time.   8 One unintended consequence of this was the separation of: the cost of a service from the person who received it.  In9 an unconstrained marketplace, the cost of a service rises ; and falls based on supply and demand, and the relative cost ? of a medical procedure versus the cost of apples or automobiles = either controls or reflects the balance of supply and demand, 8 depending on the perspective from which one looks at it.  > One way to prod the system back towards affordable health careA would be for insurance companies to provide catastrophic coverage B (which is very inexpensive) combined with medical savings accounts> and the restoration of the tax deduction for medical expenses.  B And as for government-provided health care, a major cause of these; types of prices in the US is because the rates at which the @ government reimburses providers are invariably years out of dateB and frequently below the costs involved in providing the services.  % The term for this is "cost shifting".    WWWebb   -----Original Message-----/ From: Info-VAX-Request@Mvb.Saic.Com at INTERNET ' Sent: Wednesday, March 13, 2002 1:05 PM B To: Webb, William W Raleigh, NC; Info-VAX@Mvb.Saic.Com at INTERNETE Subject: RE: [OT] US/Europe living costs, was: Re: VMS sys admin sala     3 "David J. Dachtera" <djesys.nospam@fsi.net> writes:   C > When I was in the hospital 2 days prior to Fall '99 DECUS (kidney D > stone), a room and care (blood/urine tests, MRI, Iv saline and oneC > shot of morphine) for roughly 48 hours came to somewhere close to 
 > $6,000 U.S.   D > This past week, a blood draw and battery of blood tests (trying toA > track down a hormone imbalance) came to some $240 + $40 for the  > office visit.   E Hum, I was in hospital for 24 days. 4 hrs OR, 10 days ICU, 14 days on ) ward. Total cost, $130 for the Ambulance.   F A large number of people with no health care is a disaster in waiting.; The disease your taxes pay for, is the one you don't get...    --< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.G EPIC, The Architecture of the future, always has been, always will be.=    ------------------------------  # Date: Thu, 14 Mar 2002 02:57:54 GMT $ From: "Tim Peer" <peert@envysys.com>G Subject: A Hybrid Approach: Codasyl DBMS Data Warehousing on Oracle Rdb C Message-ID: <SkUj8.10240$vN5.2582517928@newssvr14.news.prodigy.com>   I > What I'm really wondering about is the performance benefits (if any) of L > using a single mixed storage area for similarly constructed tables, ratherH > than one or more uniform storage areas. The application I'm working onI > requires moving records among such tables and I'm looking to reduce the  I/O - > as much as possible during those processes.   D Why not try a hybrid approach to take advantage of the excellent I/OE performance from MIXED format storage areas with placement via hashed D indexes, and store data in UNIFORM storage areas? We have found thisE approach to be among the best performing with very little maintenance I required. We found the implementation described below to work best.  Many L sites where this approach is deployed have little technical expertise and/orL may not have a DBA on staff--the Corporate Controller sometimes performs the system administration tasks.   For perspective:  L We needed to create an Rdb data warehouse using the Codasyl DBMS schema as aH model. Since Codasyl DBMS is not a relational model, it was necessary toK "reverse-engineer" the Codasyl DBMS schema and create a relational database K that preserved Codasyl DBMS Owner and Member relationships.  In our Codasyl E DBMS database, PRTREC is stored in PRTAREA.  We preserved the orginal J Codasyl structure in Rdb.  Companion storage areas and tables use the sameL name as the Codasyl DBMS counterpart. Storage maps were created to place theG row in the PRTAREA_NV.RDB storage area. Records are stored in Rdb using F ordinals from the Codasyl DBMS DBKEY (rowid). We broke apart the DBKEYI "1:22:5" , formatted as: area=1, page = 22, line = 5.  The Rdb schema was G designed with a MIXED format storage area (MFSA) to store "unique" keys L (since we are using the DBMS DBKEY, there is one and only one DBMS DBKEY forE each record). A HASHed index was created for the composite DBKEY with G placement in the MFSA. Data are stored to tables via a storage map in a F UNIFORM format-storage area (not RDB$SYSTEM).  When inserted, rows areJ appended to the table, unique key lookups occur via the composite - HASHed index.  H This is put to test as we read the Codasyl DBMS After Image Journal fileI (AIJ) and obtain changes, insertions and deletions made to Codasyl. These J AIJ rows are inserted, changed and deleted in Rdb using the Codasyl DBKEY. Performance is excellent.   L Create SORTed indexes when selecting data using a RANGE of key values. StoreI the sorted indexes in the storage with the data or in a dedicated UNIFORM C storage area. A dedicated storage for sorted indexes may work best.   F We retrieve data via ODBC using AREA,PAGE,LINE - our application moves8 copies data to a database servers running Oracle 8i, SQLB Server, and SYBASE.  SORTed index keys take the form: NVTIMESTAMP,K PAGE,AREA,LINE. The sorted index compares the timestamp on the Rdb row to a 	 saved VMS L timestamp (when the table was copied last to the PC/UNIX database server theF time the transfer started is stored in a transfer table).  Only new orI changed rows are ever moved to the database server. Deletions are handled K using a different method.  There was a bug in Oracle's Rdb ODBC driver when ? selecting records via a timestamp key, "WHERE timestamp-value > K other-timestamp-value."  The bug walked the table and did not use an index. F For larger tables, performance was unacceptable, adding additional key7 segments provided a reasonable performance work-around.   1 Here is the Rdb schema information for reference:    create storage area PRTAREA  filename 'PRTAREA_NV.RDA'  -- read write storage area locking is row level page format is UNIFORM page size is 2 blocks  allocation is 2 pages : extent is (minimum 2000, maximum 20000, percent growth 20)   create storage area NVINDEX . filename 'NV$DISK:[ENVY.002]MANDB_NVINDEX.RDA' -- read write storage area locking is row level page format is MIXED page size is 3 blocks  allocation is 251386 pages7 extent is (minimum 98, maximum 9998, percent growth 20) 7 snapshot filename 'NV$DISK:[ENVY.002]MANDB_NVINDEX.SNP' " snapshot allocation is 32464 pages@ snapshot extent is (minimum 98, maximum 9998, percent growth 20) interval is 216;>     CREATE STORAGE MAP PRTREC_MAP FOR PRTREC STORE IN PRTAREA;!     CREATE INDEX NV_PRTREC_HSHIDX $         ON PRTREC (PAGE, AREA, LINE)9         TYPE IS HASH ENABLE COMPRESSION STORE IN NVINDEX;   J Our observations revealed this to be a simpler alternative to using hashedI insertions and row clustering--data warehousing on Rdb without a Database C Administrator on staff. Send me an e-mail and I will send you a DCL : procedure that you can use to calculate allocation values.   -- Timothy E. Peer  eNVy Systems Inc.  4960 Almaden Exp. #330 San Jose, CA 95118 Voice: (408) 363-8896  Fax:    (408) 363-8897   http://envysys.com  7 "Frank Sapienza" <sapienza@noesys.com> wrote in message & news:a6jf3v01bfm@enews3.newsguy.com...I > I know this will be a shock, if anyone is still reading this newsgroup,  but 0 > here is an actual Oracle RDB related question. >  > Given the following:G > -  Two tables, with identical structure (that is, field names, types,  > ordering -- metadata) ' > -  One mixed page format storage area K > -  Both tables have storage maps putting them in that single storage area  > -  Consider the following SQL  >     INSERT INTO TABLE2 >     SELECT * FROM TABLE1 > 
 > Question 1: L > -  Is RDB smart enough to realize that there is no need to physically move4 > the record, but simply change its table ownership? > H > My guess is no, for a variety of reasons.  Primarily because I haven'tL > studied much on RDB internals and I'm just thinking, nah, that wouldn't be > possible.  :-) > 
 > Question 2: J > -  Let's say the record being SELECTed from TABLE1 in the above query isG > stored on, say, page 1422.  Further, assume page 1422 contains enough  space J > to hold another copy of that record.  Will RDB place the INSERTed record onL > the same page, or will it search for free space starting at the top of the > storage area?  > I > What I'm really wondering about is the performance benefits (if any) of L > using a single mixed storage area for similarly constructed tables, ratherI > than one or more uniform storage areas.  The application I'm working on I > requires moving records among such tables and I'm looking to reduce the  I/O - > as much as possible during those processes.  >  >  > Thanks in advance, > Frank  >  >  >    ------------------------------  % Date: Thu, 14 Mar 2002 03:38:24 +0000 & From: Alan  Greig <a.greig@virgin.net>K Subject: Re: A Hybrid Approach: Codasyl DBMS Data Warehousing on Oracle Rdb 8 Message-ID: <t9609u831fhk79dhfhappfrue4l1e97ti3@4ax.com>  @ On Thu, 14 Mar 2002 02:57:54 GMT, "Tim Peer" <peert@envysys.com> wrote:    M >We needed to create an Rdb data warehouse using the Codasyl DBMS schema as a I >model. Since Codasyl DBMS is not a relational model, it was necessary to L >"reverse-engineer" the Codasyl DBMS schema and create a relational databaseL >that preserved Codasyl DBMS Owner and Member relationships.  In our CodasylF >DBMS database, PRTREC is stored in PRTAREA.  We preserved the orginal  @ MANMAN eh? Did you evaluate using ISG Navigator/Attunity to justC automatically map the DBMS database to a relational  model. Or even E use RDB/DBMS transparent gateway - although I wouldn't recommend that A method as we found it slow and buggy during testing. But Attunity > works fine and lets us do whatever we need with ODBC accessingC underlying MANMAN databases. However I'm not sure if that  would be / the case if we allowed updates via this method,   K >Codasyl structure in Rdb.  Companion storage areas and tables use the same M >name as the Codasyl DBMS counterpart. Storage maps were created to place the H >row in the PRTAREA_NV.RDB storage area. Records are stored in Rdb usingG >ordinals from the Codasyl DBMS DBKEY (rowid). We broke apart the DBKEY J >"1:22:5" , formatted as: area=1, page = 22, line = 5.  The Rdb schema wasH >designed with a MIXED format storage area (MFSA) to store "unique" keysM >(since we are using the DBMS DBKEY, there is one and only one DBMS DBKEY for F >each record). A HASHed index was created for the composite DBKEY withH >placement in the MFSA. Data are stored to tables via a storage map in aG >UNIFORM format-storage area (not RDB$SYSTEM).  When inserted, rows are K >appended to the table, unique key lookups occur via the composite - HASHed  >index.  > I >This is put to test as we read the Codasyl DBMS After Image Journal file J >(AIJ) and obtain changes, insertions and deletions made to Codasyl. TheseK >AIJ rows are inserted, changed and deleted in Rdb using the Codasyl DBKEY.  >Performance is excellent. > M >Create SORTed indexes when selecting data using a RANGE of key values. Store J >the sorted indexes in the storage with the data or in a dedicated UNIFORMD >storage area. A dedicated storage for sorted indexes may work best. > G >We retrieve data via ODBC using AREA,PAGE,LINE - our application moves 9 >copies data to a database servers running Oracle 8i, SQL C >Server, and SYBASE.  SORTed index keys take the form: NVTIMESTAMP, L >PAGE,AREA,LINE. The sorted index compares the timestamp on the Rdb row to a
 >saved VMSM >timestamp (when the table was copied last to the PC/UNIX database server the G >time the transfer started is stored in a transfer table).  Only new or J >changed rows are ever moved to the database server. Deletions are handledL >using a different method.  There was a bug in Oracle's Rdb ODBC driver when@ >selecting records via a timestamp key, "WHERE timestamp-value >L >other-timestamp-value."  The bug walked the table and did not use an index.G >For larger tables, performance was unacceptable, adding additional keyD8 >segments provided a reasonable performance work-around. >e2 >Here is the Rdb schema information for reference: >e >create storage area PRTAREA >filename 'PRTAREA_NV.RDA' >-- read write storage areae >locking is row levell >page format is UNIFORM  >page size is 2 blocks >allocation is 2 pages; >extent is (minimum 2000, maximum 20000, percent growth 20)e >f >create storage area NVINDEX/ >filename 'NV$DISK:[ENVY.002]MANDB_NVINDEX.RDA'1 >-- read write storage area0 >locking is row level  >page format is MIXEDt >page size is 3 blocks >allocation is 251386 pages 8 >extent is (minimum 98, maximum 9998, percent growth 20)8 >snapshot filename 'NV$DISK:[ENVY.002]MANDB_NVINDEX.SNP'# >snapshot allocation is 32464 pages A >snapshot extent is (minimum 98, maximum 9998, percent growth 20)M >interval is 216; ? >    CREATE STORAGE MAP PRTREC_MAP FOR PRTREC STORE IN PRTAREA;n" >    CREATE INDEX NV_PRTREC_HSHIDX% >        ON PRTREC (PAGE, AREA, LINE)e: >        TYPE IS HASH ENABLE COMPRESSION STORE IN NVINDEX; >eK >Our observations revealed this to be a simpler alternative to using hasheduJ >insertions and row clustering--data warehousing on Rdb without a DatabaseD >Administrator on staff. Send me an e-mail and I will send you a DCL; >procedure that you can use to calculate allocation values.$   ------------------------------  # Date: Wed, 13 Mar 2002 19:05:41 GMT " From: bugs@pu.net (Mark Hittinger)3 Subject: Re: Another big block votes against mergere: Message-ID: <9qNj8.25$1m.1492@bin5.nnrp.aus1.giganews.com>  H Bank of America announced today that they are following Wells Fargo and  voting their stock no.  M I wonder of either or both of these large banks have a significant investmenti in Tandems and/or Alphas :-)   Lateri   Mark Hittinger bugs@pu.netg   ------------------------------  # Date: Wed, 13 Mar 2002 20:16:43 GMTo0 From: prune@ZAnkh-Morpork.mv.com (Paul Winalski)3 Subject: Re: Another big block votes against mergerg8 Message-ID: <3c8fb350.160833326@proxy.news.easynews.com>  < On 12 Mar 2002 18:39:11 GMT, bill@triangle.cs.uofs.edu (Bill Gunshannon) wrote:  H >Early comments were that by SEC rules none votes counted as "No" votes.H >It was also said that more than a simple majority was needed.  based onG >that, I saw little chance of the merger going through once Hewlett andVH >Packard made their intentions to voe "No" clear.  Based on what seem toE >the rules now, the deck seems to be stacked against the stockholdersc	 >bigtime.0  @ As I understand it, the HP and Compaq stockholders are voting on) different questions with different rules:-  < On the HP side, the vote technically is whether to authorize< management to issue enough shares to accomplish the purchaseE of Compaq.  This requires (1) that at least 50% of outstanding shares C be voted (i.e., that there be a quorum), and (2) that a majority (ItD think a simple majority) of the shares that are voted be in favor of? the proposal.  In this context, management has to get a greaterSC than 50% 'yes' of the shares voted.  Not voting at all only affectsa whether they get a quorum.  A On the Compaq side, passing the proposal requires a 'yes' vote byCA 50% of all outstanding shares, not just those that vote.  In this C case not voting at all is equivalent to a 'no' vote on management'ss	 proposal.-    
 ---------- Remove 'Z' to reply by email.-   ------------------------------  # Date: Wed, 13 Mar 2002 20:19:59 GMTE0 From: prune@ZAnkh-Morpork.mv.com (Paul Winalski)3 Subject: Re: Another big block votes against mergers8 Message-ID: <3c8fb4aa.161179684@proxy.news.easynews.com>  D On 12 Mar 2002 11:46:56 -0600, young_r@encompasserve.org (Rob Young) wrote:  O >THIS PROXY WILL BE VOTED AS SPECIFIED. IF NO SPECIFICATION IS MADE, THIS PROXYrP >WILL BE VOTED "FOR" THE PROPOSAL TO APPROVE AND ADOPT THE AGREEMENT AND PLAN OFM >REORGANIZATION AMONG HEWLETT-PACKARD COMPANY, HELOISE MERGER CORPORATION ANDt4 >COMPAQ AND APPROVE THE MERGER CONTEMPLATED THEREBY.  F This means that if you sign your proxy card but don't indicate how youB want your shares voted, management will vote your shares 'for' theE proposal.  They will NOT vote your shares at all if you don't send inc' a properly signed and dated proxy card.o  @ As a shareholder you and you alone get to decide how your sharesB are voted, unless you nominate someone else (a proxy) to vote yourB shares for you.  If you don't nominate a proxy, your shares aren't
 voted at all.o  
 ---------- Remove 'Z' to reply by email.a   ------------------------------   Date: 13 Mar 2002 22:02:15 GMT) From: leslie@clio.rice.edu (Jerry Leslie)t3 Subject: Re: Another big block votes against mergerg' Message-ID: <a6oi97$i5g$1@joe.rice.edu>a  # Mark Hittinger (bugs@pu.net) wrote: J : Bank of America announced today that they are following Wells Fargo and  : voting their stock no. : E : I wonder of either or both of these large banks have a significant e) : investment in Tandems and/or Alphas :-)   H There was an article on the Risks Digest last year about a major outage ' of an Hitachi mainframe at Wells Fargo:P  4    http://catless.ncl.ac.uk/Risks/21.15.html#subj1.1&    Wells Fargo computer network outage  *   "Peter G. Neumann" <neumann@csl.sri.com>     "Sun, 3 Dec 2000 21:12:02 PSTo  J    On 1 Dec 2000, the nationwide Wells Fargo computer network crashed for H    a few hours, three days after WF had finished merging their computer J    networks with those of Norwest (which bought WF in 1998).  One of four M    Hitachi Tritium 400 mainframes in the Minneapolis data center shut itself eN    down, apparently after detecting some sort of anomaly.  The result stopped J    all banking operations that depend on real-time interaction.  [Source: M    Article by Sam Zuckerman, *San Francisco Chronicle*, 2 Dec 2000, PGN-ed] "r    4 --Jerry Leslie     (my opinions are strictly my own)   ------------------------------  # Date: Wed, 13 Mar 2002 22:28:22 GMTR0 From: prune@ZAnkh-Morpork.mv.com (Paul Winalski)3 Subject: Re: Another big block votes against mergerl8 Message-ID: <3c8fd1ab.168603889@proxy.news.easynews.com>  D On 12 Mar 2002 12:01:41 -0600, young_r@encompasserve.org (Rob Young) wrote:  Q >> WHEN PROPERLY EXECUTED, THIS PROXY WILL BE VOTED IN THE MANNER DIRECTED HEREIN-M >> BY THE UNDERSIGNED SHAREOWNER. IF NO DIRECTION IS MADE, THIS PROXY WILL BE.L >> VOTED FOR THE ABOVE-DESCRIBED PROPOSAL AND IN THE DISCRETION OF THE PROXYK >> HOLDERS UPON SUCH OTHER BUSINESS AS MAY PROPERLY COME BEFORE THE SPECIALh6 >> MEETING OR ANY ADJOURNMENT OR POSTPONEMENT THEREOF. >>  A >> 	Hard to tell if that means all unreturned proxies are assumedc >> 	to be "for".  C It means that shares that aren't represented at the meeting, either B directly by the shareholder attending or by proxy, aren't voted at> all.  What that means for the merger question depends on whichD company is involved.  On the HP side, shares that aren't representedD don't matter except for quorum purposes.  If 50% of the shares vote,E it doesn't matter what happens to the shares that don't vote.  On theh5 Compaq side, not voting is equivalent to voting "no".l  < >	There is an amended version out there...  seems to suggestB >	that unreturned ballots count as yes, look at the added line(s): >aK >http://www.sec.gov/Archives/edgar/data/47217/000101287002000387/dex991.txta >CO >WHEN PROPERLY EXECUTED, THIS PROXY WILL BE VOTED IN THE MANNER DIRECTED HEREIN:K >BY THE UNDERSIGNED SHAREOWNER. IF NO DIRECTION IS MADE, THIS PROXY WILL BEsN >VOTED FOR THE ABOVE-DESCRIBED PROPOSAL AND, WHETHER OR NOT DIRECTION IS MADE,N >                                            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^N >WILL BE VOTED IN THE DISCRETION OF THE PROXY HOLDERS UPON SUCH OTHER BUSINESS5 >^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ F >AS MAY PROPERLY COME BEFORE THE SPECIAL MEETING OR ANY ADJOURNMENT ORL >POSTPONEMENT THEREOF. THE UNDERSIGNED HEREBY REVOKES ALL PROXIES PREVIOUSLYN >GIVEN BY THE UNDERSIGNED TO VOTE AT THE SPECIAL MEETING OR ANY ADJOURNMENT OR >POSTPONEMENT THEREOF.  E The section you underlined only applys to "such other business as mayt> properly come before the special meeting or any adjournment orC postponement thereof".  On the question of the merger, which is thesB one question on the proxy form, the proxy holders WILL vote as youF directed them to, if you directed them to vote in a particular fashion at all.l  D If you sign, date, and return the proxy form but don't indicate how @ you want the proxy holder to vote, they will vote FOR the merger	 proposal.d  C If you don't return the proxy form, or you don't sign and date it, nC the solicitors of the proxy can't vote your shares at all.  Period.o  
 ---------- Remove 'Z' to reply by email.o   ------------------------------  # Date: Wed, 13 Mar 2002 22:28:00 GMTa1 From: "Terry C. Shannon" <terryshannon@attbi.com>a3 Subject: Re: Another big block votes against mergerc9 Message-ID: <QnQj8.21300$44.4839145@typhoon.ne.ipsvc.net>d  6 "Jerry Leslie" <leslie@clio.rice.edu> wrote in message! news:a6oi97$i5g$1@joe.rice.edu...g >oI > There was an article on the Risks Digest last year about a major outageG) > of an Hitachi mainframe at Wells Fargo:  >a6 >    http://catless.ncl.ac.uk/Risks/21.15.html#subj1.1( >    Wells Fargo computer network outage >E, >   "Peter G. Neumann" <neumann@csl.sri.com> > ! >   "Sun, 3 Dec 2000 21:12:02 PST  >eK >    On 1 Dec 2000, the nationwide Wells Fargo computer network crashed forvI >    a few hours, three days after WF had finished merging their computer K >    networks with those of Norwest (which bought WF in 1998).  One of four G >    Hitachi Tritium 400 mainframes in the Minneapolis data center shutw itselfG >    down, apparently after detecting some sort of anomaly.  The resulte stoppeduK >    all banking operations that depend on real-time interaction.  [Source:pE >    Article by Sam Zuckerman, *San Francisco Chronicle*, 2 Dec 2000, 	 PGN-ed] "p  E I would beware of any mainframe whose name was identical to that of an+ radioisotope with a fairly brief half-life.e   ------------------------------  % Date: Wed, 13 Mar 2002 18:18:21 -0500f1 From: Michael Austin <maustin@firstdbasource.com>n3 Subject: Re: Another big block votes against merger 2 Message-ID: <3C8FDE3D.7CF526DF@firstdbasource.com>   "Terry C. Shannon" wrote:A ><snip>E > G > I would beware of any mainframe whose name was identical to that of aD- > radioisotope with a fairly brief half-life.p  + Thanks Terry,  I needed a laugh today :)   e    -- still looking for work... E  -- Oracle (Rdb and Classic) DBA on OpenVMS, and UNIX (doesn't matterg the flavor)e4  -- OpenVMS Sys Admin, Applications Production Admin  -- OpenVMS Apache Admin  -- and a whole lot more...O  G I got really bent out of shape the last few days as a major Billion $$$PG company wanted an out-of-work contractor to pay $1000-$1500 to fly fromNF NC to NYC for a one hour "courtesy" face-to-face interview ... for a 2@ month contract. In the words of John Stossel of ABC.. "GIVE ME A
 BREAK!!"    H --name with-held because I might get place at another one of their sites in the future :(     -- u Regards,  7 Michael Austin            Registered Linux User #261163y7 First DBA Source, Inc.    http://www.firstdbasource.com8 Sr. Consultant 704-947-1089 (Office)  704-236-4377 (Mobile)s   ------------------------------   Date: 13 Mar 2002 18:11 CST ' From: carl@gerg.tamu.edu (Carl Perkins) & Subject: Re: Antigen found =*.exe file- Message-ID: <13MAR200218112152@gerg.tamu.edu>e  - P.Young@unsw.EDU.AU (Patrick Young) writes... G }Seriously though - it would appear we cannot email OpenVMS executables D }around anymore - all because of Window(tm). Some sites setup "anti-: }virus" software on their email gateways - anyone have anyE }interesting stories on what this may/has done to OpenVMS .exe files?    Zip them and send the zip file?E+ Use backup on them up and send the saveset?   F As far as I know, it is very uncommon to block .ZIP files and probablyF even less common to block the typical saveset extensions (.SAV or .BCK are pretty common for that).  ? You can, of course, call the executable file anything you want.uD FOO.EXE could just as easily be called FOO.EXECUTABLE or FOO.RUN_ME.D This won't stop you from being able to extract it the same as a .EXEF file and run it via the RUN command. Doing this for Windows .EXE files> will tend to defeat the "autmatically run the program when theE attatchment is clicked on" behavior that you tend to get with WindowsfA (a side effect which could be considered a very good thing) - thehG recipient would have to save the attachment to disk and then change the ? extender back to .EXE to get the typcial behavior for the file.s   --- Carl   ------------------------------    Date: 13 Mar 2002 12:34:23 -0800. From: SPAMSINK2001@YAHOO.COM (Alan E. Feldman) Subject: Re: COMPACTION = Message-ID: <343f30ae.0203131234.4bd2240f@posting.google.com>    rdeininger@mindspring.com (Robert Deininger) wrote in message news:<rdeininger-1303020836130001@1cust64.tnt2.nashua.nh.da.uu.net>...@ > In article <a6nb7d$nmq$1@odysseus.uci.kun.nl>, "Hamid Bourchi" > <hbourchi@hotmail.com> wrote:  > 	 > >Hello,D > >64 > >I have a problem making a COMPACTed IMAGE BACKUP.G > >When I INITIALIZE and MOUNT the tapedrive with COPMACTION qualifier,e > >everything is O.K.aL > >But when I start the IMAGE BACKUP then SHOW DEVICE MKA500/FULL shows that > >COMACTION is disabled.t   [...]s  F > The tape has to be written with compaction from the beginning.  SomeG > drives foolishly turn off compaction if you give them a chance to use.E > built-in defaults when they are at the beginning of tape.  To avoide6 > problems with overly clever drives, it's best to useJ > /MEDIA_FORMAT=COMPACTION with INITIALIZE, MOUNT (if used), and BACKUP.      E I can't test it on V7.2-1 but I would guess that it works the same asnA on my V6.2 systems. BACKUP /MEDIA_FORMAT=COMPACTION only works iftD /REWIND is specified. So use it on both INIT and MOUNT  *or*  use it with BACKUP ... /REWIND.     Disclaimer: JMHO Alan E. Feldmanf. afeldman qiasdfasdfeubqei gfigroup asdfasd com   ------------------------------    Date: 13 Mar 2002 12:38:10 -0800. From: SPAMSINK2001@YAHOO.COM (Alan E. Feldman) Subject: Re: COMPACTIONa= Message-ID: <343f30ae.0203131238.22be7ad4@posting.google.com>s  b "Hamid Bourchi" <hbourchi@hotmail.com> wrote in message news:<a6nb7d$nmq$1@odysseus.uci.kun.nl>... > Hello, > 3 > I have a problem making a COMPACTed IMAGE BACKUP. F > When I INITIALIZE and MOUNT the tapedrive with COPMACTION qualifier, > everything is O.K.K > But when I start the IMAGE BACKUP then SHOW DEVICE MKA500/FULL shows thatE > COMACTION is disabled. > / > $ INIT/MEDIA_FORMAT=COMPACTION MKA500: imgbkp  > $ SHOW DEVICE MKA500/FULLa > N > Magtape AZNA02$MKA500:, device type TZ88, is online, record-oriented device,I >     file-oriented device, error logging is enabled, controller supportss@ >     compaction (compaction enabled), device supports fastskip. > , > $ MOUNT/FOR/MEDIA_FORMAT=COMPACTION MKA500 > $ SHOW DEVICE MKA500/FULLh > L > Magtape AZNA02$MKA500:, device type TZ88, is online, allocated, deallocate > onN >     dismount, mounted foreign, record-oriented device, file-oriented device,J >     error logging is enabled, controller supports compaction (compaction( >     nabled), device supports fastskip. > B > $ BACKUP/IMAGE/RECORD/REWIND/IGNORE=(LABEL,NOBACKUP,INTERLOCK) - >                 DKB0: -bB >                 MKA500:dkb0/MEDIA_FORMAT=COMPACTION/SAVE_SET/LOG > $ SHOW DEVICE MKA500/FULLt   [...]a   [Nitpick alert!]  F /REWIND is an input and output qualifier. It probably doesn't fix thisD problem to put it in the right place, but BACKUP is very fussy aboutB the positioning of at least some of its qualifiers, so I recommendF that you put it in the right place after the tape drive. Also, /LOG isD a command qualifier and should be places before the input specifier,5 though I think it still works when placed at the end.i   Disclaimer: JMHO Alan E. Feldmanu afeldman zxcv gfigroup zxcv comn   ------------------------------  % Date: Thu, 14 Mar 2002 00:51:38 -0000t/ From: Michael Zarlenga <zarlenga@conan.ids.net>d4 Subject: DEC C: why does exit(0) really exit with 1?/ Message-ID: <u8vt0q2t48icc7@corp.supernews.com>   ; What the heck is going on ... why is exit 0 exiting with 1?e     Contents of file makeit.com:   $ set verify $ create exitzero.cA #include <stdlib.h>N main() { 
     exit (0);a }a
 $ cc exitzeroi $ link exitzerog $ run exitzero $ write sys$output $STATUS $ set noverify      	 $ @makeit  $ create exitzero.ci #include <stdlib.h>o main() {f
     exit (0);g }  a
 $ cc exitzeron $ link exitzeroY $ run exitzero $ write sys$output $STATUS
 %X00000001     Why?!      --   -- Mike Zarlenga   ------------------------------   Date: 13 Mar 2002 19:35 CSTt' From: carl@gerg.tamu.edu (Carl Perkins) 8 Subject: Re: DEC C: why does exit(0) really exit with 1?- Message-ID: <13MAR200219350711@gerg.tamu.edu>   3 Michael Zarlenga <zarlenga@conan.ids.net> writes...t< }What the heck is going on ... why is exit 0 exiting with 1? }Contents of file makeit.com:R } 
 }$ set verifyN }$ create exitzero.c }#include <stdlib.h> }main()  }{ }    exit (0); }} }$ cc exitzero }$ link exitzero }$ run exitzeror }$ write sys$output $STATUSo }$ set noverifye } 
 }$ @makeit }$ create exitzero.c }#include <stdlib.h> }main()u }{ }    exit (0); }}   }$ cc exitzero }$ link exitzero }$ run exitzerod }$ write sys$output $STATUSe }%X00000001  }  }  }Why?! }--  }-- Mike Zarlenga   G Because at the DCL level a status of 0 is not a success, it is a statusr with the "warning" severity.  E Thus it gets translated to a 1, which is a "success" severity status.N  D Since your C program is doing what C programs tend to do to indicate@ success, it makes a lot more sense to report a status value that translates to the messageo  . %SYSTEM-S-NORMAL, normal successful completion  < than it does to keep it a 0, which translates to the message  ( %NONAME-W-NOMSG, Message number 00000000  & This is explained in the on-line help, $ help cc run-time exito        [...]J        The status argument corresponds with an errno value if exiting fromF        a child process using vfork and/or an exec function, or with anH        OpenVMS condition value if exiting from a process invoked by DCL.D        The errno values are defined in the <errno.h> header file.  AE        status value of 0 or EXIT_SUCCESS is translated to the OpenVMStG        SS$_NORMAL status code to return the OpenVMS success value.  AnyPJ        other status value is left the same.  The status value is passed toH        the parent process.  A status value of EXIT_FAILURE is translated%        to an error-level exit status.C   --- Carl   ------------------------------  # Date: Thu, 14 Mar 2002 02:03:32 GMTe4 From: Tim Llewellyn <tim.llewellyn@blueyonder.co.uk>8 Subject: Re: DEC C: why does exit(0) really exit with 1?0 Message-ID: <3C9003BB.4F642E39@blueyonder.co.uk>   Michael Zarlenga wrote:  > = > What the heck is going on ... why is exit 0 exiting with 1?t  1 $STATUS is set to the exit status of the program.e  # $write sys$output F$message(status)s. %SYSTEM-S-NORMAL, normal successful completion  < Presumably the C exit call does not directly set the program exit status.  B Actually, I just checked and my initial suspicions are configrmed.E Its explained in the C Runtime manual, under the description of exit..  C http://www.openvms.compaq.com/commercial/c/5763p025.htm#index_x_683g  F Hey, maybe I CAN do C consultancy :-) However, you need to try to read the manuals more.-   Regards      -- , tim.llewellyn@blueyonder.co.uk    F * tim.llewellyn@cableinet.co.uk address will cease to work June 2002 *   ------------------------------  # Date: Thu, 14 Mar 2002 03:23:54 GMTw1 From: "David J. Dachtera" <djesys.nospam@fsi.net>c8 Subject: Re: DEC C: why does exit(0) really exit with 1?' Message-ID: <3C90192C.59190C7A@fsi.net>S   Michael Zarlenga wrote:E > = > What the heck is going on ... why is exit 0 exiting with 1?A >  > Contents of file makeit.com: >  > $ set verify > $ create exitzero.c  > #include <stdlib.h>V > main() > {N >     exit (0);. > }h > $ cc exitzeroy > $ link exitzeroc > $ run exitzero > $ write sys$output $STATUS > $ set noverify >  > $ @makeith > $ create exitzero.cl > #include <stdlib.h>o > main() > {l >     exit (0);d > }  > $ cc exitzero  > $ link exitzerod > $ run exitzero > $ write sys$output $STATUS > %X00000001 >  > Why?!   C Well, I'm not a C-coder, but I understand that in UN*X-land, statusf' zero(0) means success. Notice, however:o   DJAS01::DDACHTERA$ exit 0l( %NONAME-W-NOMSG, Message number 00000000  - Zero(0) implies a "warning" severity on VMS.     In VMS-land:  2 DJAS01::DDACHTERA$ write sys$output f$message( 1 ). %SYSTEM-S-NORMAL, normal successful completion  E There was a recent thread about $STATUS and severities. Try searching. for it.e   -- a David J. Dachteran dba DJE Systemse http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------   Date: 14 Mar 2002 01:00:08 GMT( From: joechip31@hotmail.com (Herb Asher)* Subject: decus uk - confusion over joining8 Message-ID: <3c8ff46f.6483823@news1.uncensored-news.com>  
 Hi people,  B Either Decus UK, or CUO or what ever they are called at the moment4 have a very confusing website or I am just dull. :O(  E I want to join DECUS and have been to 3 or 4 different pages and seenmD prices ranging from ?35 to a USA site quoting $135 for internationalD members (ouch!). I'm not even sure what the CUO one is all about. ItA does not mention DECUS or having DECUS membership included in thea price.  F All I want to do is join Decus so I can buy the Alpha Hobby Kit CD andF i'm not too keen on paying $135 to join and then more for the kit as IF am in-between jobs (unemployed if you don't know what that means)  :O)   HELP before I go insane !i   Thanks for any advice.   Herb        F ______________________________________________________________________R Posted Via Uncensored-News.Com - Still Only $9.95 - http://www.uncensored-news.comP    With NINE Servers In California And Texas - The Worlds Uncensored News Source      ------------------------------  # Date: Thu, 14 Mar 2002 01:57:50 GMTy* From: 2damncommon <2damncommon@nospam.com>. Subject: Re: decus uk - confusion over joining< Message-ID: <20020313165852.7fa848cb.2damncommon@nospam.com>  ] I joined in the US with the free Associate Membership to participate in the Hobbyist program.AA I found the site not so much confusing as lacking in information.S After going through all pages to complete the registration for the Associate Membership I never ran across information stating what, if anything, I was waiting on. Email, snailmail, nothing??? At least in the US - Free Associate Membership is good for the Hobbyist Program (you will, of course, still need to acquire software)  I was sent an Encompass ID about 3 weeks or so after my application. Apparently you also need to wait a few more days before ordering a Hobbyist CD (if needed.)' Fun after getting through the red tape.v9 Expect to RTFM if you are not already well versed in VMS.'
 Good Luck.   On 14 Mar 2002 01:00:08 GMTs) joechip31@hotmail.com (Herb Asher) wrote:a   >  >  > Hi people, > D > Either Decus UK, or CUO or what ever they are called at the moment6 > have a very confusing website or I am just dull. :O( > G > I want to join DECUS and have been to 3 or 4 different pages and seenIF > prices ranging from ?35 to a USA site quoting $135 for internationalF > members (ouch!). I'm not even sure what the CUO one is all about. ItC > does not mention DECUS or having DECUS membership included in the  > price. > H > All I want to do is join Decus so I can buy the Alpha Hobby Kit CD andH > i'm not too keen on paying $135 to join and then more for the kit as IH > am in-between jobs (unemployed if you don't know what that means)  :O) >  > HELP before I go insane !  >  > Thanks for any advice. >  > Herb >  >  >  > H > ______________________________________________________________________T > Posted Via Uncensored-News.Com - Still Only $9.95 - http://www.uncensored-news.comR >    With NINE Servers In California And Texas - The Worlds Uncensored News Source >      ------------------------------  # Date: Thu, 14 Mar 2002 04:33:12 GMTe4 From: Tim Llewellyn <tim.llewellyn@blueyonder.co.uk>. Subject: Re: decus uk - confusion over joining0 Message-ID: <3C9026CE.FF9E2B2C@blueyonder.co.uk>   Herb  D I renewed with CUO-UK for 25 quid recently, thats private membership7 rather than corporate, the latter being more expensive.l   regardsh   Herb Asher wrote:  >  > Hi people, > D > Either Decus UK, or CUO or what ever they are called at the moment6 > have a very confusing website or I am just dull. :O( > G > I want to join DECUS and have been to 3 or 4 different pages and seenaF > prices ranging from ?35 to a USA site quoting $135 for internationalF > members (ouch!). I'm not even sure what the CUO one is all about. ItC > does not mention DECUS or having DECUS membership included in thes > price. > H > All I want to do is join Decus so I can buy the Alpha Hobby Kit CD andH > i'm not too keen on paying $135 to join and then more for the kit as IH > am in-between jobs (unemployed if you don't know what that means)  :O) >  > HELP before I go insane !  >  > Thanks for any advice. >  > Herb > H > ______________________________________________________________________T > Posted Via Uncensored-News.Com - Still Only $9.95 - http://www.uncensored-news.comR >    With NINE Servers In California And Texas - The Worlds Uncensored News Source >    -- i tim.llewellyn@blueyonder.co.uk    F * tim.llewellyn@cableinet.co.uk address will cease to work June 2002 *   ------------------------------    Date: 13 Mar 2002 14:21:14 -0600+ From: kuhrt@encompasserve.org (Marty Kuhrt)e Subject: Re: DFO Issue3 Message-ID: <W7hi73vN$Apa@eisner.encompasserve.org>e  n In article <a5mea4$423$1@newslocal.mitre.org>, lewis@luminascrewsupspammers.mitre.org (Keith A. Lewis) writes:z > "Barry Treahy, Jr." <Treahy@mmaz.com> writes in article <3C7D46B0.80807@mmaz.com> dated Wed, 27 Feb 2002 13:50:56 -0700:G >>a recommendation.  Contact Raxco and ask for a demo version of their iI >>PerfectDisk product.  Shutdown DFO and try it, you may find that it is  K >>worth changing loyalties with regard to disk defraging because I suspect oI >>it will do a much better job...  We've had a long time experience with c> >>Raxco and they are good folks that put out a good product... > I > While we're giving recommendations, I'll tell you why I moved the other.) > direction -- from Perfectdisk to DFO.  n > M > Raxco uses a different (non-PAK) licensing scheme.  To get the license, yourN > must install the software and pay for it, run a utility that reads some infoA > from your system, send the info to Raxco, and wait for the key.  > L > I rebuilt my system disk a couple of times in the early 1990s.  Both timesK > the key stopped working.  The first time they were nice and re-did it fortL > free.  The second time they noticed I had upgraded the CPUs in my cabinet. > H > "So you upgraded from an Alphaserver 2100 4/250 to an Alphaserver 2100% > 5/300.  You owe us a transfer fee."  >  > "But it's the same machine!" > 
 > No luck. > K > I bought DFO instead and have not had any problems putting the license oniK > whatever damn server I wish.  Last year I replaced the 2100 with a DS20E,rL > rebuilding the system disk from scratch, and my PAK transferred just fine. >   > Most licenses are not transferable between machines, legally.    ------------------------------  % Date: Wed, 13 Mar 2002 22:24:33 -0500u2 From: rdeininger@mindspring.com (Robert Deininger) Subject: Re: DFO IssueK Message-ID: <rdeininger-1303022224330001@1cust130.tnt3.nashua.nh.da.uu.net>e  3 In article <W7hi73vN$Apa@eisner.encompasserve.org>,i, kuhrt@encompasserve.org (Marty Kuhrt) wrote:    ? >Most licenses are not transferable between machines, legally. s  E Depending on the vintage, many layered product licenses are legal forb transfer between machines.   ------------------------------  # Date: Thu, 14 Mar 2002 02:56:10 GMTh1 From: "David J. Dachtera" <djesys.nospam@fsi.net>e( Subject: Re: DLT Degausser / Bulk Eraser' Message-ID: <3C9012B7.2B8ACC1C@fsi.net>R   Todd Nelson wrote: >  > Greetings Everyone.t > J > I am looking into purchasing a Degausser / Bulk Eraser that'll work with > DLT, SDLT, and 8mm media.  > I > Does anyone have experience with "low volume" (cheaper) degaussers that" > would care to share? > L > Manufacturer information; Reseller Information; etc; would all be helpful.  H Radio Shack's VHS tape eraser works well on DLTs, 9-tracks and such, but, has a very short duty cycle (less than 10%).  G It is, however, quite powerful. I once used one to help pull the fridgeaB out away from the wall. It didn't like it, and it heated up *VERY*0 quickly, but it stuck like glue while energized.   Dunno if they still sell it...   -- g David J. Dachterag dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/t   ------------------------------  % Date: Wed, 13 Mar 2002 22:17:39 -0500h2 From: rdeininger@mindspring.com (Robert Deininger)! Subject: Re: DTTS on Unix MachineTK Message-ID: <rdeininger-1303022217390001@1cust130.tnt3.nashua.nh.da.uu.net>N  H In article <bc8f06ddcce967ce5f3b4dfb273dc434.77540@mygate.mailgate.org>,1 "Max Mauceri" <mmauceri@rome.sema.slb.com> wrote:N  	 >Hi Guys t< >i have a trouble on DS10 machine running 4.0f Digital Unix.= >The machine is dtss clerk (Digital Distributed Time Service)i> >But unfortunately if i check the status of dtts on NCL prompt >i have a following message :  >v% >migwy1# ncl sync dtss set clock truev > 
 >Node 0 DTSS e' >AT 2002-03-11-16:41:39.338+00:00I-----I > ! >FAILED IN DIRECTIVE: Synchronize ' >DUE TO: No such Entity Instance existss >h >i1 >I suppose a dtss clerk process doesn't start....E    That's what it looks like to me.    F >which is the manner (on Digital Unix System ) to start this porcess ?  H Well, that's a bit hard to answer, since I've only used DTSS on VMS.  WeG have one or more *dtss*.ncl files (in the usual places for ncl files). eI But there is also a command procedure, which starts up all the components)J in the right order, including calling the NCL script(s).  The NCL might beI the same between VMS and Unix, but the command procedures surely aren't. e> I wouldn't know where to look for corresponding files on unix.   >o >Thank for your attectionG >A >P >-- 9 >Posted via Mailgate.ORG Server - http://www.Mailgate.ORGu   ------------------------------  % Date: Wed, 13 Mar 2002 19:44:55 -0500O  From: John Santos <JOHN@egh.com>( Subject: Re: Dynamic IP address question- Message-ID: <1020313193022.351C@Ives.egh.com>D  & On 13 Mar 2002, Bill Gunshannon wrote:  ) > In article <3C8F5631.69E465CE@aaa.com>,r> >  Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com> writes:. > |> Isn't this what name servers should fix ? >  > No.a >  > |> c4 > |> That is, the DHCP server on your network should9 > |> update your name server(s) whenever it assignes yourp7 > |> laptop a different IP address (then it had before)  > A > Many places consider the security risks of DDNS to far outweighc@ > the advantages of having names that map to constantly changing > addresses. > G > |> And then SET DISPLAY to your host name instead of your IP address.P > |> RH > |> Or, ask for a looong "lease time" for your MAC address, so the DHCPF > |> server never will put your IP address back into the free pool (or > |> whatever it's called).  > = > Which has problems in itself, like defeating the purpose ofi> > running DHCP in the first place.  Might as well assign fixed@ > addresses.  If DHCP is used so that the clients can roam, then> > the solution works, but if DHCP is used primarily because ofE > the potential for there to be more possible clients than addresses, $ > well, I think you get the picture. > > > Seems to me that it should be trivial for this to be done inA > someone's login.com, thus negating the need to reduce security._ >  > bill  < DDNS seems designed so that the client after receiving an IP= address from the DHCP server then informs the DNS server whatr= its new IP address and name are.  This seems to me, to put ite: mildly, insane.  If you actually enable DDNS, can you ever trust your DNS server again?  : Shouldn't the DHCP server be responsible for informing the< DNS server what IP addresses it has doled out (and also when; they cease to be valid?)  The DNS server should only accepti< dynamic updates from the DHCP server, and only for addresses0 and host names it knows are in the dynamic pool.  < BTW, Windows 2000 (and apparently XP) have DDNS turned on by9 default, even for clients with static IP addresses.  This ? fills our OPERATOR.LOG with "named: unapproved update" messagesC; (about 350 of them in the last 4 days on a relatively smalls: network from the two PCs that haven't been reconfigured to: turn this off.)  A VAX running TCPware V5.5 is the primary? nameserver, with and Alpha running TCPIP V5.1 as the secondary.i   -- d John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  # Date: Wed, 13 Mar 2002 21:03:07 GMTi# From: "Dan" <io_crater@hotmail.com>m" Subject: HSZ10 looking for info...; Message-ID: <f8Pj8.138271$kb.7590940@news1.calgary.shaw.ca>n  G I recently picked up what I believe is an HSZ10 at an auction that I am_L hoping to attach to a VMS hobbyist  system (Alpha 4100 or 1000).    As usualJ for auctions the HSZ10 came with no manuals or software. A model number on$ the back of the unit shows SWXRA-01.    @ It does not have the PC Card that I have seen in newer model HSZH controllers.  It has an MMJ jack on the front along with 2 68 pin female SCSI connectors-    J Any pointers on documentation and software requirements for this unit?   I3 am not having much luck with the internet searches.-   Dan Henigman   .n   ------------------------------  % Date: Wed, 13 Mar 2002 18:40:06 -0500w, From: "Frank Sapienza" <sapienza@noesys.com>& Subject: Re: HSZ10 looking for info..., Message-ID: <a6oo5j02fng@enews3.newsguy.com>  F The SWXRA part number you gave is for the cabinet, not the controller.  I You may want to remove the controller from the cabinet and check the part L number just to make sure it's not an HSD10, which wouldn't be very useful toJ you.  The other option is to get (or make) a DEConnect cable and plug in a; VT (or emulator) into that MMJ port and power up the thing.n  K The single-channel controllers (HSD10, HSZ10) did not use a PCMCIA card forr
 the firmware.      Franko  . "Dan" <io_crater@hotmail.com> wrote in message5 news:f8Pj8.138271$kb.7590940@news1.calgary.shaw.ca...AI > I recently picked up what I believe is an HSZ10 at an auction that I am H > hoping to attach to a VMS hobbyist  system (Alpha 4100 or 1000).    As usualiL > for auctions the HSZ10 came with no manuals or software. A model number on& > the back of the unit shows SWXRA-01. >  > B > It does not have the PC Card that I have seen in newer model HSZJ > controllers.  It has an MMJ jack on the front along with 2 68 pin female > SCSI connectors  >r >dL > Any pointers on documentation and software requirements for this unit?   I5 > am not having much luck with the internet searches.  >  > Dan Henigman >o   ------------------------------  % Date: Wed, 13 Mar 2002 20:29:03 -0000h5 From: "Malcolm" <malcolmix@neverness.freeserve.co.uk>0F Subject: Re: I may have to case-mod an Alpha. Anyone been here before?. Message-ID: <a6ocl3$rtf$1@news8.svr.pol.co.uk>  1 "Shane Smith" <ssmith@icius.com> wrote in message-* news:01C1C9DC.4EB7EFF0@sulfer.icius.com...J > My employer wants a small, easily carted-around Alpha for demos. We haveJ > an old AlphaStation 200, but it's too heavy for his tastes. He knows theC > Tadpole AlphaBook 1 is defunct, and he knows one of my hobbies iseH > remanufacturing and customizing PC cases. He just put 2 and 2 togetherH > and asked me to look into taking a larger Alpha and rebuilding it in a > smaller package. >oJ > In other words, he just gave the kid the keys to the candy shop..... ;-)   Is money no object? ;-),  K Well, there are some nice possibilities. Someone else mentioned a DS10; you 2 want a DS10L 1U server. Then you need a display...  H On our PC servers at work, we have a Compaq switching box. It's a 19" 1UI rackmount device, which basically consists of a standard PC and trackballwH with buttons. If this display unit has no "intelligence" and can operateJ without the switching box (Not sure on this point.... then you put the twoK together and you have a nice loggable Alpha, 2U high, which is probably the J easiset way to get what you're looking for. And with a nice Compaq logo on the display unit  L I've now checked the Compaq website, and the thing I am talking about is theI TFT5600 RKM Rackmount Keyboard Display. I think - the description doesn'ttL mention a trackball, but I'm sure it's the same one. Part number 221546-001. Cost $2,500 (gulp!)   K The two of these together would make a nice luggable Alpha system. *If* thelL two are compatible. You might need some small tweaks, but I'm sure it'd workD OK. You could integrate the two in a nice moulded case, if the brief included that.  K You'll need a graphics card for the DS10, though, which means you can't addrJ any more cards - only one PCI slot. (Or does it come with video? Does come? with keyboard and mouse connectors, but doesn't mention VGA...)M  J A cheaper option, but a bit of a cheat: get an PC laptop with network cardL and decent screen and video subsystem (You could just buy a new one... stillK cheaper than the TFT5600 RKM), plug it and the DS10L together, run a serial L line to the laptop as well, then use the laptop (with operating system and XF server of your choice) as the serial console while booting and as an XK terminal while running. This could also make it very nice for debugging, ifnJ you run M$ Windoze-based development/debugging tools with your application (note I said "if" :-)    Hope that helps,	 -Malcolm.r >Don >tI > Has anyone done anything like this before? I know about the VAXbar, buteH > I'm interested in swapping cautionary tales with anyone who's taken anD > angle grinder to an Alpha, or tried building them from parts in PCJ > cases, etc Advice on good donor machines to hunt down would be good too.G > I'm looking for >100mhz CPU, preferably ATX format motherboards, etc.U >, > Shanec >aG > P.S. Sorry if Outlook (hawk, spit) sends this in HTML. I seem to haveo& > mislaid the "send in format" option. >  >  #####	 > #-O-O-#-	 > #  L  #  >  #===# >   ###r >    ------------------------------  % Date: Wed, 13 Mar 2002 12:56:57 -08001$ From: Shane Smith <ssmith@icius.com>F Subject: RE: I may have to case-mod an Alpha. Anyone been here before?0 Message-ID: <01C1CA8E.93E63930@sulfer.icius.com>  E There's some good ideas flying around here, and turning up in privateoG e-mail. Thanks guys. Let's consolidate answers to some of the questions> I'm getting.  F Does it have to be an Alpha. Yes. We don't normally build VAX versionsA of our product any more, and performance is a factor. This is forh% presentations to potential customers.   A Can it be a Multia. Possibly. I'll put it on the list of options,r thanks.r  F Why ATX. Well, it's not exactly a requirement. Another format would beF OK, I was thinking of the availability of the power supplies. The caseH itself I'm thinking of building from scratch, and taking the motherboardE tray from a donor case that has the card mount on the tray. I'd brusheE the dust off my metalworking skills for the rest. I figure I can packtE things in pretty tightly since once built the config won't change. AsrG long as I make sure there's cool airflow to all the bits that need it I  should be fine.   , What I'm aiming at is an Alpha equivalent of3 http://www.shuttleonline.com/spec.php3?model=ss40. e   Shane    ------------------------------  % Date: Wed, 13 Mar 2002 23:10:58 -0000e5 From: "Malcolm" <malcolmix@neverness.freeserve.co.uk>tF Subject: Re: I may have to case-mod an Alpha. Anyone been here before?. Message-ID: <a6om4l$2t0$1@news8.svr.pol.co.uk>  1 "Shane Smith" <ssmith@icius.com> wrote in message2* news:01C1CA8E.93E63930@sulfer.icius.com...G > There's some good ideas flying around here, and turning up in private I > e-mail. Thanks guys. Let's consolidate answers to some of the questionsi > I'm getting. >iH > Does it have to be an Alpha. Yes. We don't normally build VAX versions7 > of our product any more, and performance is a factor.   < A DS10L would certainly give you *very* good performance ;-)   -Malcolm   -- Malcolm MacArthuro  , Subtract nine for e-mail (anti-spam measure)   ------------------------------  % Date: Wed, 13 Mar 2002 16:05:02 -0800i  From: Jim Hull <hull@cup.hp.com> Subject: Re: Itanium troubles0- Message-ID: <MPG.16f9890c95b24e329896dc@news>-  ' Stuecheli <jeffas@us.ibm.com> wrote ...r  G > It is interesting that the 701 peak result for the hp rx4610 has beenCG > revised to be quite a bit less.  I assume this is due to not shippinggG > that system?  So they posted the result, yet could not really deleverWJ > the result, thus had to remove it.  That is a bit worse than just having+ > higher results posted from other vendors.   M What makes you think that the rx4610's SPECfp peak of 701 has been revised?  oL When I go to the spec.org website and seach for rx4610, I find these single- cpu results:   CFP2000$F Company                     System           Result Baseline PublishedE Hewlett Packard Corporation hp server rx4610 653    545      Aug-2001oE Hewlett Packard Corporation hp server rx4610 701    701      Jun-2001 E Hewlett Packard Corporation hp server rx4610 623    623      Jun-2001u  M Granted, if that's all you look at, you might conclude that 701 went down to uK 653, and you might wonder what would explain the 701 vs. 623.  However, if dL you configure the search to include additional items, in particular the CPU ) frequency, OS, and compiler, you'll find:E  / Result Baseline MHz Op. Sys. Compiler Published . 653    545      800 HP-UX 11 HP C/ANS Aug-2001. 701    701      800 Windows  Intel C+ Jun-2001/ 623    623      733 Windows  Intel C+ Jun-2001 p  J which should make it clear that the Aug 2001 results are not revised, but K instead are results from a different OS and compiler; if you use the Intel eH compiler & Windows, an 800MHz Itanium gets you 701 peak, while the same K hardware running HP-UX & HP's compiler gets you 653.  If you choose to run >L Linux and compile with gcc, or to run Windows and use Microsoft's compiler, # you'll get still different results.c  M As far as not shipping the system, or being unable to deliver these results,  M that's not true either.  HP will be very happy to sell you an rx4610, and it vI will deliver the results shown above, depending on which compiler/OS you   choose to run.    -- Jim-     HP IA-64 Architect   ------------------------------  % Date: Thu, 14 Mar 2002 03:10:10 +0000n& From: Alan  Greig <a.greig@virgin.net>4 Subject: Re: Microsoft Curries Favor With Undergrads8 Message-ID: <gq409uo2d383n8aide9of7cgl9jf3fc7bm@4ax.com>  A On Wed, 13 Mar 2002 16:47:36 GMT, jlsue <jlsuexxxz@insightbb.com>A wrote:  4 >On Tue, 12 Mar 2002 15:20:31 -0500, "Brian Tillman"- ><tillman_brian@notnoone.notnohow.com> wrote:l >cC >>> What would you have us do here -- what additional steps can we tD >>>  provide here that we could afford to provide and that we could @ >>>  profit (indirectly, obviously) from -- that is not already  >>>  available?p >>= >>Put your computers back in schools the way they used to be.2 >t >Ah, okay.  A shotgun wedding.  C Are you totally nuts or just living in a fantasy world? We lost VMSoC from the university where I was the systems manager because Digital.E went over my head to senior university management and said NT was the-E way forward not VMS. How many times do I have to say this? Go talk tomF Rich Marcello who admitted to myself and others that he was aware this? had gone on during  the "affinity" tragedy. Contact Clair WellsfD formerly of VMS project management in Livingston, who tried to get aB retraction from Digital management and couldn't. It makes my bloodE boil when Digital/Compaq personnel prefer to stick their heads in the ' sand rather than fix the real problems.     2 >Not speaking for anyone, certainly not DEC/Compaq. >(get rid of the xxxz in my address to e-mail)   ------------------------------  % Date: Thu, 14 Mar 2002 03:24:39 +0000 & From: Alan  Greig <a.greig@virgin.net>7 Subject: Re: Mimer SQL for OpenVMS Itanium Announcement$8 Message-ID: <lt509u061eglbcu2uc5opp01fsoo3trlo9@4ax.com>  1 On Wed, 13 Mar 2002 08:47:58 -0500, "Main, Kerry"u <Kerry.Main@Compaq.com> wrote:   >All - >  >As a fyi -C. >http://www.mimer.com/news/engnews_mar02_2.htm >0C >"Compaq's OpenVMS Systems Software Group has signed a co-operation:G >agreement with Upright Database Technology for the database management C >system Mimer SQL. The agreement entails that Upright guarantees toUF >continue developing Mimer SQL for OpenVMS. The next development phase0 >includes development for the Itanium platform."   	A8 Always nice to see some positive news. Keep posting 'em.    + >[snip - see url for rest of press release]4 >r	 >Regards,: >G >Kerry Mainc >Senior Consultant >Compaq Canada Corp. >Professional Services >Voice: 613-592-4660 >Fax  :  819-772-7036. >Email: Kerry.Main@Compaq.coml   ------------------------------  % Date: Wed, 13 Mar 2002 15:26:47 -0700a4 From: "Michael D. Ober" <mdo.@.wakeassoc.com.nospam>/ Subject: Opening a SYS$OUTPUT file in VMS Basice1 Message-ID: <LmQj8.870$kA.110197@news.uswest.net>U  I In the following code: (i've replace variables with literals for brevity)a  / redirect = "sys$dsk:[testarea.err]MYOUTPUT.LIS"C6 retval = lib$spawn("RUN MYPROGRAM",,redirect,,,,,,,,,) if retval = SS$_NORMAL   then3      open redirect for input as file #92, [options]  ...o   end if  H The "Open Redirect" fails with a record type mismatch.  For [options], I
 have tried   <no options> organization Sequential stream  organization sequential variable# organization unkown, recordtype any_   among others to no avail._  ? Dir/full reports the file is sequential with a record type VFC.9  5 How can I open and read the text in my redirect file.s   -- Thanks,A
 Mike Ober.   ------------------------------  % Date: Wed, 13 Mar 2002 20:12:52 -0500a  From: John Santos <JOHN@egh.com>3 Subject: Re: Opening a SYS$OUTPUT file in VMS Basic:4 Message-ID: <1020313200949.351E-100000@Ives.egh.com>  + On Wed, 13 Mar 2002, Michael D. Ober wrote:   K > In the following code: (i've replace variables with literals for brevity)o > 1 > redirect = "sys$dsk:[testarea.err]MYOUTPUT.LIS"g8 > retval = lib$spawn("RUN MYPROGRAM",,redirect,,,,,,,,,) > if retval = SS$_NORMAL >   then5 >      open redirect for input as file #92, [options]s > ...w >  > end if > J > The "Open Redirect" fails with a record type mismatch.  For [options], I > have tried >  > <no options>  > organization Sequential stream" > organization sequential variable% > organization unkown, recordtype anyw >  > among others to no avail.A > A > Dir/full reports the file is sequential with a record type VFC.i > 7 > How can I open and read the text in my redirect file.   A Did you try "organization sequential [variable], recordtype list".  A I believe that's what DCL creates for output files.  I'm not surew@ if you need to include "variable" in the organization clause, or& if the "recordtype list" implies that.   > --	 > Thanks,s > Mike Ober.   -- u John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  # Date: Thu, 14 Mar 2002 03:17:17 GMT 1 From: "David J. Dachtera" <djesys.nospam@fsi.net>e3 Subject: Re: Opening a SYS$OUTPUT file in VMS Basict' Message-ID: <3C90183E.EE8C9C48@fsi.net>    "Michael D. Ober" wrote: > K > In the following code: (i've replace variables with literals for brevity)' > 1 > redirect = "sys$dsk:[testarea.err]MYOUTPUT.LIS" 8 > retval = lib$spawn("RUN MYPROGRAM",,redirect,,,,,,,,,) > if retval = SS$_NORMAL >   then5 >      open redirect for input as file #92, [options]. > ...  >  > end if > J > The "Open Redirect" fails with a record type mismatch.  For [options], I > have tried >  > <no options>  > organization Sequential stream" > organization sequential variable% > organization unkown, recordtype anyh >  > among others to no avail.y > A > Dir/full reports the file is sequential with a record type VFC.r > 7 > How can I open and read the text in my redirect file.f  . Try "organization sequential, recordtype any".   --   David J. Dachteray dba DJE Systemsd http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------    Date: 13 Mar 2002 20:11:39 -0800) From: P.Young@unsw.EDU.AU (Patrick Young) 3 Subject: Re: Opening a SYS$OUTPUT file in VMS BasicT= Message-ID: <55f85d77.0203132011.10ac8f29@posting.google.com>s  m "Michael D. Ober" <mdo.@.wakeassoc.com.nospam> wrote in message news:<LmQj8.870$kA.110197@news.uswest.net>...r  5 >      open redirect for input as file #92, [options].  3 open redirect for input as file #92, recordtype anyP   ------------------------------  # Date: Thu, 14 Mar 2002 02:44:50 GMT 1 From: "David J. Dachtera" <djesys.nospam@fsi.net>-) Subject: Re: OpenVMS jobs available - FYI ' Message-ID: <3C90100D.980939E1@fsi.net>m   Tony Cheung wrote: > b > "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message news:<3C898E1C.DC2DF3E4@fsi.net>... > > John Smith wrote:  > > >u > > > www.directemployers.comaR > > > This site is a co-op that takes you directly to the employer's own web site. > >mI > > Thanks for that, John! Found something for myself (maybe) that as yetsK > > hasn't even hit HeadHunter.Net. Not on the major job boards yet, eitherH' > > (Monster, ComputerJobs, Dice, ...).h > F > I am just interested to know what does the pay of OpenVMS developers+ > compare to Unix or Windows NT developers?s  G Hhmmm... Tough call. OVMS job postings are generally looking for senioriH people (since that's about all there is nowadays, anyway) while UN*X andE NT/W2K postings usually want cert.'s, but also want relatively junior:E people to hold down on payroll costs (since the need ratio (headcount>5 vs. server count) of VMS Vs. NT/W2K is so disparate).n  D Most OVMS developer jobs these days are Oracle-related or C-related,; anyway while most UN*X and NT/W2K developer jobs are VisualnG (fill-in-the-blank) related, from what I've seen in the Chicago market.t   -- k David J. Dachterae dba DJE SystemsT http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/H   ------------------------------  % Date: Wed, 13 Mar 2002 21:06:32 +0100i5 From: "Robert A.M. van Lopik" <lopik@mail.telepac.pt>7+ Subject: Oracle Context (interMedia) on VMSe$ Message-ID: <3c8fb1c6$1@news.wau.nl>  J Did anyone ever use Oracle's full text retrieval software (called Context,L Intermedia of just Text, depending on the season) on VMS. And if so, did youG want to index multiple document types (like ASCII, Word, HTML, PDF) for.J which you need the so called INSO filters, that are not available for VMS.$ Finally, did you find a work-around?  L This is suddenly very important to me. Any suggestions or ideas are welcome.   regardse
 rob van lopikt   ------------------------------  % Date: Wed, 13 Mar 2002 22:23:19 -0500t2 From: rdeininger@mindspring.com (Robert Deininger)/ Subject: Re: Oracle Context (interMedia) on VMS2K Message-ID: <rdeininger-1303022223190001@1cust130.tnt3.nashua.nh.da.uu.net>c  < In article <3c8fb1c6$1@news.wau.nl>, "Robert A.M. van Lopik" <lopik@mail.telepac.pt> wrote:  K >Did anyone ever use Oracle's full text retrieval software (called Context,CM >Intermedia of just Text, depending on the season) on VMS. And if so, did youaH >want to index multiple document types (like ASCII, Word, HTML, PDF) forK >which you need the so called INSO filters, that are not available for VMS.o% >Finally, did you find a work-around?r   I know nothing from Oracle...j  F But Northern Light (www.northernlight.com) specializes in indexing andG searching documents in any and all formats.  They used to have a publicsH search web site, but they no longer offer that service.  They still sellG similar services to companies with private data they want to search and & index.  And their software run on VMS.  J You might want to contact them and see if your offerings match your needs.   ------------------------------    Date: 13 Mar 2002 22:34:58 -0800, From: tony.cheung@asiayeah.com (Tony Cheung)6 Subject: Re: sending all data as out-of-band in TCP/IP= Message-ID: <f9dc0a5a.0203132234.7719c3fa@posting.google.com>p   Hi John,  = It is due to the client's request. We are helping a client tooE re-architect an OpenVMS application. In the client's past experience, D he has experienced message lost when sending data from OpenVMS to NT? clients without using OOB. Somehow, he seems to have solved the # problem by sending all data in OOB.i  ? I also agree we should avoid OOB and only use it for compelling.F reasons. At the same time, I am interested in understanding more aboutC OOB and if it really could have any impact if one sends all data int OOB.  = Anyway, meanwhile, we have persuaded the client to avoid OOB.    Thank you very much.   Tony Cheunge  n john@ossc.net (John Gemignani, Jr.) wrote in message news:<35b06b78.0203121828.23582cc5@posting.google.com>...- > Why would you want to send ALL data as OOB?s >  > -John  > z > "Matt Muggeridge" <Matt.Muggeridge@compaq.com> wrote in message news:<wvij8.9420$mp.41309@news-server.bigpond.net.au>...I > > > Rather than sending all data normally, if one transmits all data asuH > > > out-of-band, will there be any performance or behavior difference? > > 8 > > Yes.  It won't work... at least as you might expect. > >  > > > Ie= > > > won't be sending any normal data, but out-of-band data.  > > : > > Defining normal and OOB data is a good place to start. > > N > > OOB data is a single character that can be queued at the receiver ahead ofO > > other 'normal' data.  By setting the OOBINLINE flag, it will be queued witha > > the 'normal' data. > > M > > The OOB data byte can and is overwritten by subsequent OOB data.  So your / > > receiver has to keep up with it or lose it.a > > K > > OOB seemed like a good idea back when it was being standardised, but ins4 > > practice it suffers from a lot of tricky detail. > > M > > My advice is... try to avoid it.  If you absolutely cannot, then take thetJ > > time to learn all of its limitations.  I've seen normally well behavedO > > programs fall apart when their receiver could not keep up with the OOB data#, > > when their system became heavily loaded. > > 	 > > Matt.- > >  > > --A > > -------------------------------------------------------------t > > OpenVMS TCP/IP Engineering > > Enterprise Computing Group > > Compaq Computer Corporations > > Gold Coast, AUSTRALIAsA > > -------------------------------------------------------------  > >  > > = > > "Tony Cheung" <tony.cheung@asiayeah.com> wrote in messageo; > > news:f9dc0a5a.0203112030.7364a382@posting.google.com...l
 > > > Hi All,n > > >aA > > > I have looked into the TCP/IP's out-of-band implementation.  > > >rI > > > Rather than sending all data normally, if one transmits all data asaJ > > > out-of-band, will there be any performance or behavior difference? I= > > > won't be sending any normal data, but out-of-band data.  > > >nI > > > Are out-of-band data subject to the same flow control mechanisms inn
 > > > TCP/IP?r > > >t > > > Thank you very much. > > >y > > > Tony Cheungt   ------------------------------  % Date: Thu, 14 Mar 2002 06:15:55 +1030g/ From: Mark Daniel <Mark.Daniel@wasd.vsm.com.au>sI Subject: TCPIP, sharing sockets via $QIO (programming, long and detailed) . Message-ID: <3C8FAC73.7080108@wasd.vsm.com.au>  8 Compaq TCP/IP Services for OpenVMS Alpha Version V5.1-15, on a DEC 3000 Model 300 running OpenVMS V7.3  A I have an application where I would like two (and more) detached -H processes to share a socket, in particular both to listen and accept on E that socket and share the incoming connections.  Note: this is *not* a9 parent-child, it's peer.  Both processes are independent..  F The TCP/IP socket sharing bit is not particularly well documented and E more implies that it can be done rather than stating it.  Anyway, by gH experiment I established that it was possible and the incoming requests H are shared in a round-robin fashion.  Works fine - except under extreme E load, when eventually the backlog queue length grows and connections rC stop being accepted (established using a TCPIP> SHOW DEVICE bgnnnn t
 /CONTINUOUS).*  G Now of course this could be a programming logic issue but I've reduced tC it to the barest minimum (no *real* processing, just accepting the -G connection, quick response and finished).  It could be a process quota o@ limitation but no error status is reported by any of the system B services, and process monitoring, SDA and AMDS all atest that the H *processes* are not resource starved.  It could be the TCP/IP "kernel", E but I've checked the characteristics covered in the "Troubleshooting   Guide" and it looks OK.t  I I guess for starters I need a definitive answer on whether the BG driver eH is designed for such behaviour (we have CSC support) and advice on what  else I can do to tackle this.u  G The C code is non-trivial so fully posting it here (or to CSC for that  ? matter) is not really an option.  I hope this precis is enough.-    * On the first of the multiple processes ...* ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~  * 1) $ASSIGN() a channel to the INET device.  =     /* no pre-existing socket device available, assign one */s?     status = sys$assign (&InetDeviceDsc, &ServerChannel, 0, 0);o  # 2) $QIO the socket characteristics.2  <     /* make the channel a TCP, connection-oriented socket */;     status = sys$qiow (EfnWait, ServerChannel, IO$_SETMODE,eE                        &svptr->ServerIOsb, 0, 0, &TcpSocket, 0, 0, 0,l8                        &ServerSocketReuseAddrOption, 0);:     if (VMSok (status)) status = svptr->ServerIOsb.Status;  + 3) Bind the socket to the address and port.U  F     /* prepare to bind the server socket to the IP address and port */I     svptr->ServerSocketNameItem.Length = sizeof(svptr->ServerSocketName);rC     svptr->ServerSocketNameItem.Address = &svptr->ServerSocketName; 1     svptr->ServerSocketName.sin_family = AF_INET;t8     svptr->ServerSocketName.sin_addr.s_addr = IpAddress;5     svptr->ServerSocketName.sin_port = htons(IpPort);p;     status = sys$qiow (EfnWait, ServerChannel, IO$_SETMODE, 0                        &svptr->ServerIOsb, 0, 0,:                        0, 0, &svptr->ServerSocketNameItem,3                        svptr->ListenBacklog, 0, 0);h:     if (VMSok (status)) status = svptr->ServerIOsb.Status;  F 4) I then get the socket BG device name using $GETDVIW.  This is also % stored for access by other processes.h  I 5) $ASSIGN another (autonomous) channel to the BG device returned above. -8   This becomes just another channel to the above socket.  ? 6) Important step, make this channel shareable (TCPIP$C_SHARE).W  E     /* make the socket shareable (seems to work only if done here) */ B     status = sys$qiow (EfnWait, svptr->ServerChannel, IO$_SETMODE,0                        &svptr->ServerIOsb, 0, 0,@                        0, 0, 0, 0, &ServerSocketShareOption, 0);:     if (VMSok (status)) status = svptr->ServerIOsb.Status;  = 7) Then $DASSGN the original channel.  The BG device remains  ( instantiated by the second one assigned.     In subsequent processes ...e ~~~~~~~~~~~~~~~~~~~~~~~~~~~>  & 1) Retrieve the stored BG device name.  H 2) $ASSIGN a channel to the BG device returned above.  This becomes yet $ another channel to the above socket.  / 3) Make this channel shareable (TCPIP$C_SHARE).y  E     /* make the socket shareable (seems to work only if done here) */nB     status = sys$qiow (EfnWait, svptr->ServerChannel, IO$_SETMODE,0                        &svptr->ServerIOsb, 0, 0,@                        0, 0, 0, 0, &ServerSocketShareOption, 0);:     if (VMSok (status)) status = svptr->ServerIOsb.Status;    1 In all processes (after socket instantiation) ... 1 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~u  F 1) $ASSIGN a channel to the INET device.  This will become the client  socket.h  :     /* assign a channel to the internet template device */F     status = sys$assign (&InetDeviceDsc, &svptr->ClientChannel, 0, 0);  6 2) Queue an accept() I/O to the then listening socket.  I     svptr->ClientSocketNameItem.Length = sizeof(svptr->ClientSocketName);.C     svptr->ClientSocketNameItem.Address = &svptr->ClientSocketName;g+     svptr->ClientSocketNameItem.LengthPtr =.'         &svptr->ClientSocketNameLength;0  6     status = sys$qio (EfnNoWait, svptr->ServerChannel,/                       IO$_ACCESS | IO$M_ACCEPT,g?                       &svptr->ServerIOsb, &NetAcceptAst, svptr,a9                       0, 0, &svptr->ClientSocketNameItem,T,                       &svptr->ClientChannel,.                       &ClientSocketOption, 0);  I 3) Then accept AST function creates a data structure and uses the client eE channel and socket name to begin autonomously, AST-driven processing s (i.e. does not block).  G 4) Once the client data is retrieved and initiated go straight back to 4 step 1 in this block.K    H NOTE that after the socket is set up and begins processing ALL activity 2 will be happening at user-mode AST delivery level.  H To emphasize again.  All this works mighty well until lots of multiple, G indeterminate connections occur, sometime after that the backlog queue sI builds up and processing stops.  I know, I know, it sounds horribly like 8G the code fails to queue another listen (accept) on the channel but I'm rA 99.9% sure that's not the case.  Why?  Well if I STOP one of the .F processes (and get the right one) everything springs into life again. H It's as if the AST was not being delivered (and there's plenty of ASTLM D left).  Also I've reduced the AST routine processing down to almost % nothing and the problem still recurs.a  C I've checked for relevant TCP/IP ECOs but there doesn't seem to be s anything recent.    As always, thanks for your time.  F +--------------------------------------------------------------------+E   Mark Daniel                         http://wasd.vsm.com.au/adelaidedF   mailto:Mark.Daniel@wasd.vsm.com.au (Mark.Daniel@dsto.defence.gov.au)F +--------------------------------------------------------------------+   ------------------------------  % Date: Thu, 14 Mar 2002 08:40:51 +10301/ From: Mark Daniel <Mark.Daniel@wasd.vsm.com.au> M Subject: Re: TCPIP, sharing sockets via $QIO (programming, long and detailed)v. Message-ID: <3C8FCE6B.3000100@wasd.vsm.com.au>  D Perhaps I should add that when only the one process is operating it I works fine regardless of the load.  Add the second and sooner or later - o9 kaboom (or perhaps more accurately the sound of silence).    Mark Daniel wrote:: > Compaq TCP/IP Services for OpenVMS Alpha Version V5.1-15. > on a DEC 3000 Model 300 running OpenVMS V7.3 > C > I have an application where I would like two (and more) detached aJ > processes to share a socket, in particular both to listen and accept on G > that socket and share the incoming connections.  Note: this is *not*  ; > parent-child, it's peer.  Both processes are independent.  > H > The TCP/IP socket sharing bit is not particularly well documented and G > more implies that it can be done rather than stating it.  Anyway, by aJ > experiment I established that it was possible and the incoming requests J > are shared in a round-robin fashion.  Works fine - except under extreme G > load, when eventually the backlog queue length grows and connections aE > stop being accepted (established using a TCPIP> SHOW DEVICE bgnnnn o > /CONTINUOUS).u
 8< snip 8<   ------------------------------  % Date: Wed, 13 Mar 2002 15:54:15 -0800n$ From: Shane Smith <ssmith@icius.com>I Subject: TCPIP, sharing sockets via $QIO (programming, long and detailed)'0 Message-ID: <01C1CAA7.6063D8B0@sulfer.icius.com>  : > Compaq TCP/IP Services for OpenVMS Alpha Version V5.1-15. > on a DEC 3000 Model 300 running OpenVMS V7.3 > B >I have an application where I would like two (and more) detached I >processes to share a socket, in particular both to listen and accept on IF >that socket and share the incoming connections.  Note: this is *not* : >parent-child, it's peer.  Both processes are independent. >tG >The TCP/IP socket sharing bit is not particularly well documented and lF >more implies that it can be done rather than stating it.  Anyway, by I >experiment I established that it was possible and the incoming requests aI >are shared in a round-robin fashion.  Works fine - except under extreme  F >load, when eventually the backlog queue length grows and connections D >stop being accepted (established using a TCPIP> SHOW DEVICE bgnnnn  >/CONTINUOUS). >@H >Now of course this could be a programming logic issue but I've reduced D >it to the barest minimum (no *real* processing, just accepting the H >connection, quick response and finished).  It could be a process quota A >limitation but no error status is reported by any of the system fC >services, and process monitoring, SDA and AMDS all atest that the 9I >*processes* are not resource starved.  It could be the TCP/IP "kernel", 0F >but I've checked the characteristics covered in the "Troubleshooting  >Guide" and it looks OK. >a [snip]  D Mark, try monitoring the other quotas of the processes, not just theC ASTLM. I've seen this behaviour before in something very different.3G After a couple of late nights I tracked it down to running out of TQEs,iE and the network stack waiting for one to come free instead of blowingtA with an out of quota. I wasn't using TQEs in my code, so I didn'tuH initially bother to check. You said "if I get the right one", maybe thatF one's hung in the network stack waiting for something and tying up theE socket.  It may be a total red herring, but I thought I'd mention it.y  F If I might make another suggestion, instead of ASTs might you considerE multi threading? The central thread posts an accept(), when something H comes in it fires off a thread to deal with it and goes straight back toC accept() again. It simplifies the main loop, and keeps the accept():C posted for a higher percentage of the time. Posix threads are a lotn& easier than they look at first glance.   Shanep   ------------------------------  % Date: Thu, 14 Mar 2002 12:15:30 +1030a/ From: Mark Daniel <Mark.Daniel@wasd.vsm.com.au> M Subject: Re: TCPIP, sharing sockets via $QIO (programming, long and detailed)g. Message-ID: <3C9000BA.1060102@wasd.vsm.com.au>   Shane Smith wrote:
 8< snip 8<F > Mark, try monitoring the other quotas of the processes, not just theE > ASTLM. I've seen this behaviour before in something very different.3I > After a couple of late nights I tracked it down to running out of TQEs,OG > and the network stack waiting for one to come free instead of blowing C > with an out of quota. I wasn't using TQEs in my code, so I didn'toJ > initially bother to check. You said "if I get the right one", maybe that  E I'm using SDA and/or AMDS to monitor process quotas.  TQE I have 196 P remaining out of 200.d  H > one's hung in the network stack waiting for something and tying up theG > socket.  It may be a total red herring, but I thought I'd mention it.e  G My thoughts as well.  I'm not sure how this could be pinpointed though.-  H > If I might make another suggestion, instead of ASTs might you considerG > multi threading? The central thread posts an accept(), when something)  C Not possible in the context of rest of the application environment..  J > comes in it fires off a thread to deal with it and goes straight back toE > accept() again. It simplifies the main loop, and keeps the accept()tE > posted for a higher percentage of the time. Posix threads are a lote   So does my code.  ( > easier than they look at first glance.  D Yes it is, and allows a nice main-line of code to be developed (and F anyway, couldn't be any more difficult than a large application using  ASTs ;^)   Thanks for your input Shane.   ------------------------------    Date: 13 Mar 2002 19:06:48 -0800) From: john@ossc.net (John Gemignani, Jr.)uM Subject: Re: TCPIP, sharing sockets via $QIO (programming, long and detailed)t= Message-ID: <35b06b78.0203131906.5a0f85fa@posting.google.com>e  \ Shane Smith <ssmith@icius.com> wrote in message news:<01C1CAA7.6063D8B0@sulfer.icius.com>...< > > Compaq TCP/IP Services for OpenVMS Alpha Version V5.1-150 > > on a DEC 3000 Model 300 running OpenVMS V7.3 > >-D > >I have an application where I would like two (and more) detached K > >processes to share a socket, in particular both to listen and accept on -H > >that socket and share the incoming connections.  Note: this is *not* < > >parent-child, it's peer.  Both processes are independent. > >iI > >The TCP/IP socket sharing bit is not particularly well documented and rH > >more implies that it can be done rather than stating it.  Anyway, by K > >experiment I established that it was possible and the incoming requests  K > >are shared in a round-robin fashion.  Works fine - except under extreme  H > >load, when eventually the backlog queue length grows and connections F > >stop being accepted (established using a TCPIP> SHOW DEVICE bgnnnn  > >/CONTINUOUS). > >wJ > >Now of course this could be a programming logic issue but I've reduced F > >it to the barest minimum (no *real* processing, just accepting the J > >connection, quick response and finished).  It could be a process quota C > >limitation but no error status is reported by any of the system gE > >services, and process monitoring, SDA and AMDS all atest that the eK > >*processes* are not resource starved.  It could be the TCP/IP "kernel", nH > >but I've checked the characteristics covered in the "Troubleshooting  > >Guide" and it looks OK. > >. > [snip] > F > Mark, try monitoring the other quotas of the processes, not just theE > ASTLM. I've seen this behaviour before in something very different.nI > After a couple of late nights I tracked it down to running out of TQEs,tG > and the network stack waiting for one to come free instead of blowing C > with an out of quota. I wasn't using TQEs in my code, so I didn'taJ > initially bother to check. You said "if I get the right one", maybe thatH > one's hung in the network stack waiting for something and tying up theG > socket.  It may be a total red herring, but I thought I'd mention it.t > H > If I might make another suggestion, instead of ASTs might you considerG > multi threading? The central thread posts an accept(), when somethingiJ > comes in it fires off a thread to deal with it and goes straight back toE > accept() again. It simplifies the main loop, and keeps the accept()tE > posted for a higher percentage of the time. Posix threads are a lot ( > easier than they look at first glance. >  > Shane-  A I really don't understand why this is so complex.  I have writtenCE servers to do this and, in fact, Apache works this way on OpenVMS.  IA? especially don't see why there is the "extra" $ASSIGN to change-	 channels.    In the main process:       1) $ASSIGN TCPIP$DEVICEo2     2) $QIOW IO$_SETMODE to do the bind and listen-     3) Use $GETDVIW to get the BG device namea4     4) Define a logical name pointing to this device&     5) Fire up the other processes ...   In the "other processes":i  "     1) $ASSIGN to the logical name)     2) $ASSIGN a new TCPIP$DEVICE channel 6     3) $QIOW IO$_ACCESS|IO$M_ACCEPT using the channels     4) ... do your work ... #     5) $DASSGN the accepted channels     6) Goto step 2  D If you use the SHARE facilities, you are changing characteristics on> the BG UCB to allow multiple assignments.  Note that it is notB necessary to use SHARE if you have ample privileges to $ASSIGN the	 listener.w  F The servers that I have written do not use threads within them -- eachE process is basically a thread and VMS does the scheduling.  I did noto2 use ASTs at all, maing the code incredibly simple.   -JohnD   ------------------------------  % Date: Thu, 14 Mar 2002 15:13:22 +1030l/ From: Mark Daniel <Mark.Daniel@wasd.vsm.com.au>aM Subject: Re: TCPIP, sharing sockets via $QIO (programming, long and detailed) . Message-ID: <3C902A6A.9020702@wasd.vsm.com.au>   John Gemignani, Jr. wrote:
 8< snip 8<C > I really don't understand why this is so complex.  I have writtenl > servers to do this and, in fact, Apache works this way on OpenVMS.> especially don't see why there is the "extra" $ASSIGN to changes > channels.   H I added it as I began to bushrange looking for some (less than reasoned H through cause/effect) solution/workaround when I ran into this problem. E   That extra $ASSIGN is actually neither here-nor-there and could be VF removed.  What it does is resuse some of the code.  If the socket has E not been created (first process) a code section creates it, gets the rF device name, then using the same code section as subsequent processes B assigns a channel using the device name to the socket.  Same-same  effectively.  H My scenario seems slightly (perhaps significantly) different to the one  described here though.   > In the main process: >  >     1) $ASSIGN TCPIP$DEVICEu4 >     2) $QIOW IO$_SETMODE to do the bind and listen/ >     3) Use $GETDVIW to get the BG device names  	 Yep, yep.s  6 >     4) Define a logical name pointing to this device  H I use a lock value block to contain the device name (for other reasons, # a logical name certainly suffices).o  ( >     5) Fire up the other processes ...  F Mine are independent peers.  Again neither here-nor-there essentially.   > In the "other processes":o > $ >     1) $ASSIGN to the logical name+ >     2) $ASSIGN a new TCPIP$DEVICE channel)8 >     3) $QIOW IO$_ACCESS|IO$M_ACCEPT using the channels  H Slight difference.  My code creates a data structure which is given the F channel number and it begins an independent "thread" (not Pthread) of G AST-driven activity.  Before the first AST of that thread is delivered =D however the original "accept()" AST effectively goes to your step 2 F above (i.e queues the next accept() while still in the AST of the one G just delivered - before it's suggested, yes I've tried decoupling that pB new accept() queuing by using an AST to deliver after the initial  completes - same result).d   >     4) ... do your work ...e% >     5) $DASSGN the accepted channelN >     6) Goto step 2  G My created thread of execution described above eventually $DASSGNs the pE channel of it's own accord (and of course does not do step 2 because TG thats already long accomplished probably with lots of other AST-driven S/ threads circulating through the image context).t  F > If you use the SHARE facilities, you are changing characteristics on@ > the BG UCB to allow multiple assignments.  Note that it is notD > necessary to use SHARE if you have ample privileges to $ASSIGN the > listener.  > H > The servers that I have written do not use threads within them -- eachG > process is basically a thread and VMS does the scheduling.  I did noto4 > use ASTs at all, maing the code incredibly simple.  G Having a per-process thread of execution is certainly the simplest for eD development.  Can't beat a straight main-line!  The cost is the O/S F maintainance of all those processes, as well as a "cluttered" system, H particularly if a lot of concurrent requests need to be serviced.  Each H one of my processes routinely will be handling 100 to 150 requests (and I do now as a single process instance), each of which may vary in duration sD from milliseconds to many tens of seconds.  All in the one process. F This scales quite nicely even over two processors but I would like to F scale it further by allowing a single process per CPU (for instance). H It would be easy to concieve of eight server processes supporting 1,000 F concurrent requests (for example, and all things being equal), indeed G this scale has been suggested for one deployment.  The U**x concept of mG one process per activity scales less well on VMS (processes are fairly sF costly, even without creation), and even under U**x it is acknowleged D there are more efficient ways than per-process, hence toolkits like 	 Pthreads.k   Thanks for your input John.o   ------------------------------   Date: 13 Mar 2002 13:44:45 CDT= From: wayne@tachysoft.xxx.564897.killspam.00c7 (Wayne Sewell)~D Subject: RE: Transferring files from one OpenVMS machine to another.. Message-ID: <9vmFspHev0gX@tachxxsoftxxconsult>  _ In article <CIEJLCMNHNNDLLOOGNJIMEHNEFAA.tom@kednos.com>, "Tom Linden" <tom@kednos.com> writes:- > From the freeware; > ( > $! RESET_BACKUP_SAVESET_ATTRIBUTES.COM > $!3 > $! P1  is the specification of the BACKUP savesetz > $!   [cmd proc deleted]    G This is needed if the saveset is passed through ftp directly.  If it is I embedded within a zip file, as the original poster indicates, it is *not*eG necessary, because the original file attributes will be restored if the   original zip was done with "-V".  H My take is that either the zip did not use "-V", or it was not done withK info-zip from the vms machine at all, or there was something wrong with the  saveset to begin with.    O Also, the poster never said it was a saveset anyway.  We are just assuming thate  because of the extension (.bck).     >  >> -----Original Message-----n, >> From: Arjun [mailto:arjunram@hotmail.com]* >> Sent: Wednesday, March 13, 2002 7:30 AM >> To: Info-VAX@Mvb.Saic.ComC >> Subject: Transferring files from one OpenVMS machine to another.o >> i >>   >> A VMS newbie here.g >> aG >> We have a bunch of files including .bck files. We need to ship theseiC >> to our customers. I created a zip file and then ftp'ed it to theeE >> customer machine (Windows NT) using binary mode. The customer thenlI >> ftp'ed the files in binary mode to VMS. He is able to unzip the files. F >> But when he tries to restore the .bck files, he runs into problems. >> u   [stuff deleted]a   -- yO ===============================================================================nM Wayne Sewell, Tachyon Software Consulting  (281)812-0738  wayne@tachysoft.xxxl: http://www.tachysoft.xxx/www/tachyon.html and wayne.html  K change .xxx to .com in addresses above, assuming you are not a spambot  :-)kO ===============================================================================s= Society Lady:  Are you familiar with the Great Wall of China?-5        Curly:  No, but I know a big fence in Chicago!    ------------------------------  # Date: Thu, 14 Mar 2002 03:13:50 GMTa1 From: "David J. Dachtera" <djesys.nospam@fsi.net>aD Subject: Re: Transferring files from one OpenVMS machine to another.' Message-ID: <3C901762.35D7EB46@fsi.net>    Arjun wrote: >  > A VMS newbie here. > F > We have a bunch of files including .bck files. We need to ship theseB > to our customers. I created a zip file and then ftp'ed it to the2 > customer machine (Windows NT) using binary mode.  
 Nota bene:   > The customer theno) > ftp'ed the files in binary mode to VMS.l  H If this means he's UNZIPping the archives on NT, then yes: these are the problems you'll have.c    > He is able to unzip the files.. > But when he tries to restore the .bck files,   "Restore" - do you mean UNZIP?   > he runs into problems. >  > this is what he gets >   > $ set def sys$sysdevice:[test] > $ dir  >   > Directory SYS$SYSDEVICE:[test] >  > .;2                 .;1t >  > Total of 2 files.o > # > Any help on this would be great!!   B The key here, as others have noted, is to properly create the .ZIPE files. UNZIP cannot "restore" was not "saved"; that is, if you expect C files to have their original ODS and RMS attributes when you unpacka" them, you must pack them properly.   These links may help:o  4 http://www.djesys.com/vms/support/zipunzip/index.htm http://www.djesys.com/zip.html  http://www.djesys.com/unzip.html   This one is a "How-to":d0 http://www.djesys.com/vms/hobbyist/zipunzip.html   -- a David J. Dachtera  dba DJE SystemsI http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/m   ------------------------------  % Date: Wed, 13 Mar 2002 22:18:51 -0500(1 From: Michael Austin <maustin@firstdbasource.com>CD Subject: Re: Transferring files from one OpenVMS machine to another.2 Message-ID: <3C90169B.4F73EC24@firstdbasource.com>  G or if you must use ZIP, backup the files to a .BCK and ZIP/ftp the .ZIPmG file Unzip leaves you with a .BCK file you can use BACKUP on to RESTOREn1 the files.  I have used this approach many times.n       "David J. Dachtera" wrote: >  > Arjun wrote: > >e > > A VMS newbie here. > >eH > > We have a bunch of files including .bck files. We need to ship theseD > > to our customers. I created a zip file and then ftp'ed it to the4 > > customer machine (Windows NT) using binary mode. >  > Nota bene: >  > > The customer thene+ > > ftp'ed the files in binary mode to VMS.e > J > If this means he's UNZIPping the archives on NT, then yes: these are the > problems you'll have.c > " > > He is able to unzip the files.0 > > But when he tries to restore the .bck files, >   > "Restore" - do you mean UNZIP? >  > > he runs into problems. > >s > > this is what he gets > >," > > $ set def sys$sysdevice:[test]	 > > $ dirs > > " > > Directory SYS$SYSDEVICE:[test] > >  > > .;2                 .;1u > >r > > Total of 2 files.y > >f% > > Any help on this would be great!!A > D > The key here, as others have noted, is to properly create the .ZIPG > files. UNZIP cannot "restore" was not "saved"; that is, if you expectoE > files to have their original ODS and RMS attributes when you unpack $ > them, you must pack them properly. >  > These links may help:e > 6 > http://www.djesys.com/vms/support/zipunzip/index.htm  > http://www.djesys.com/zip.html" > http://www.djesys.com/unzip.html >  > This one is a "How-to": 2 > http://www.djesys.com/vms/hobbyist/zipunzip.html >  > -- > David J. Dachteran > dba DJE Systems  > http://www.djesys.com/ > * > Unofficial Affordable OpenVMS Home Page:! > http://www.djesys.com/vms/soho/h   -- r Regards,  7 Michael Austin            Registered Linux User #261163t7 First DBA Source, Inc.    http://www.firstdbasource.comm Sr. Consultant 704-947-1089 (Office)f 704-236-4377 (Mobile)r   ------------------------------  # Date: Thu, 14 Mar 2002 03:45:34 GMT 1 From: "David J. Dachtera" <djesys.nospam@fsi.net>gD Subject: Re: Transferring files from one OpenVMS machine to another.' Message-ID: <3C901E48.A77AC877@fsi.net>    Michael Austin wrote:o > I > or if you must use ZIP, backup the files to a .BCK and ZIP/ftp the .ZIP-I > file Unzip leaves you with a .BCK file you can use BACKUP on to RESTORE 3 > the files.  I have used this approach many times.R  F Watch that the saveset size is not significantly larger than the totalH (EOF) size of the files you're processing. Backup adds redundancy groupsD (/GROUP=0 is not recommended, though /NOCRC is usually o.k. for diskH savesets) and other packaging. This also adds BACKUP's idiosyncrasies to those of ZIP and UNZIP.a   -- t David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/o   ------------------------------  # Date: Wed, 13 Mar 2002 19:44:58 GMTh. From: "Barry in Indy" <bleadhurwitz@uindy.edu> Subject: Re: VFC File Problem ? Message-ID: <__Nj8.10142$k5.3792745@newssvr28.news.prodigy.com>0  5 "Paul Sture" <paul.sture@bluewin.ch> wrote in messageh# news:3C8EFE8F.7080708@bluewin.ch...a >eE > What version of VMS are you running, and is it VAX or Alpha? I havewJ > confirmed that TECO translates the VFC line feed codes into the relevantJ > number of blank lines in the output file correctly - Alpha VMS V7.1 with > the Y2K ECO.  I I had to search for the TECO commands, but it appears to work! I read the L file in using EDIT/TECO filename, then saved the file with the EX command. ID was then able to use EVE to make the desired changes, and it printed
 correctly.   Thanks to all!   ------------------------------  % Date: Wed, 13 Mar 2002 18:46:08 -0500e  From: John Santos <JOHN@egh.com>A Subject: Re: Viability of a Commercial ISO-9960 formatter for VMS-4 Message-ID: <1020313182922.351B-100000@Ives.egh.com>   Much snippage...  - On Wed, 13 Mar 2002, David J. Dachtera wrote:n > Carl Perkins wrote:. > >  [...]-H > > Normally you also get a package of other licenses too, NAS or EIS or
 > > whatever.n > 9 > Yes - if you buy them, you get them. If not, you don't.a > 1 > > }Want networking? Sorry, that's extra (or was3 > > }up until recently). > > H > > This is only the case if by "recently" you mean "for the last decade) > > or more". And that's only talking IP,  > H > I don't recall UCX being included in the OVMS base license much before  > about 1998 or so, maybe later. >  i > > DECnet has been there a lotd > > longer.e > I > I don't recall either DECnet end-node or Routing-IV ever being included-F > in the base OVMS license. You always had to buy them separately, andG > choose one or the other. Of course, DECnet-IV for OVMS-Alpha does not E > provide the routing functionality. DVNETEND is the best you can do.r >   E When we got our first Alpha, a 3000-300 in late 1993, it came bundled G with NAS-250, which includes DECWindows (don't remember if it was Motif E yet, but that eventually was included in an update under the originaltE license), UCX, DECnet end-node, and VMSCluster.  (The cluster licensenE may have been somewhat restricted, but I think we could do everythingE we needed to do with it.)   E > > (Of course a lot of people, like us, still use 3rd party softwaree > > for the IP stuff r > @ > I still prefer Multinet, 2nd choice: TCPware, 3rd choice: UCX.   [...]s > J > In a cluster, yes. DECnet provides some limited functionality, also, but; > is not recommended for any serious production work, IMHO.o   Why not?   [...]a  J > I don't believe "Mozilla" (CSWB) is currently "shipping" at all. I thinkE > you still must register for a download, and think it's still T1.0 -o' > pre-general-release (pre-Gamma-Test).     It's now officially a Beta test.   [...]      -- o John Santosr Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  # Date: Thu, 14 Mar 2002 02:06:56 GMTp1 From: "David J. Dachtera" <djesys.nospam@fsi.net>oA Subject: Re: Viability of a Commercial ISO-9960 formatter for VMSe& Message-ID: <3C9007C0.69C1031@fsi.net>   Larry Kilgallen wrote: > ] > In article <3C8EC6A0.D6BE73AE@fsi.net>, "David J. Dachtera" <djesys.nospam@fsi.net> writes:n > > Larry Kilgallen wrote: > >>` > >> In article <3C8D7A18.25D755BD@fsi.net>, "David J. Dachtera" <djesys.nospam@fsi.net> writes: > >>M > >> > You buy VMS, you get VMS. Want networking? Sorry, that's extra (or wasdM > >> > up until recently). Want file/print sharing for PCs? That almost costs K > >> > more than VMS (Advanced Server). Want a GUI? That was extra, too, upc > >> > until recently. > >>F > >> What is this "until recently" business ?  Has something changed ? > >>D > >> So far as I know, DECwindows and networking protocols are still& > >> not bundled into the VMS license. > > C > > Hhmmm... Don't have the doc.'s from the LP SPL handy. I thoughtfD > > DecW/Motif was included in one or more of the NAS licenses since3 > > recently (within the last five to eight years).  > C > But before that it was in a package with a different acronym, andT5 > before that it was just bundled with most machines.g > H > > Major Hint - Marketing Opportunity missed: even without ads or otherF > > fanfare, if you could get even one of the major the computer storeG > > chains to carry OpenVMS and the LP SPL off-the-shelf, at least thatnL > > would be more exposure than it currently gets. It doesn't cost a fortuneL > > either, just the current packaging costs, times a few ten-thousands moreJ > > units, plus expenses to have sales critters schmooze(sp?) the big wigsL > > from the store chains to get them to carry it, a couple of units of each > > per store. > D > No, you actually have to pay for shelf space, just like in grocery
 > stores.   < Would probably still be cheaper than a national ad campaign.  4 > There would be no short-term return to judge it by  - The time for short-term returns is long past.-   > and-F > it would end up like Posix, a program abandoned due to lack of proof5 > that the ongoing expense was producing any results.     From, "The Empire Strikes Back":   Yoda:		The boy has no patience!   E > How many people here would trust their local computer store to have ! > the latest version of VMS C++ ?   & How many people here can read a label?  E Anyway, all the more reason to put LP's on-line for download. Buy thel> full distro. off the shelf, download the update to the newest.   <switch shift=paradigm>e	 </switch>r   -- o David J. Dachtera  dba DJE Systemsh http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/l   ------------------------------  # Date: Thu, 14 Mar 2002 02:14:02 GMTe1 From: "David J. Dachtera" <djesys.nospam@fsi.net>eA Subject: Re: Viability of a Commercial ISO-9960 formatter for VMSn' Message-ID: <3C900964.6E5A7CCB@fsi.net>s   Tim Llewellyn wrote: >  > "David J. Dachtera" wrote: > > >Carl Perkins wrote: > > >fJ > > > This is only the case if by "recently" you mean "for the last decade+ > > > or more". And that's only talking IP,e > > J > > I don't recall UCX being included in the OVMS base license much before" > > about 1998 or so, maybe later. > 8 > Sorry, David, NAS has been around for longer than that   NAS is not OVMS-BASE. Sorry.   >, maybe the turbochannelkM > alphas didn't have it (can't remember) but all alphas with PCI buses that IrC > can recall having anything to do with came with NAS of some sort.e   ...if you bought it that way.i   > Enough toh$ > run IP and DECNET end node anyway.  F UCX was not included in OVMS-BASE until recently, within the last four+ or five years. DVNETEND still isn't, AFAIK.A  ! > > > DECnet has been there a lot 
 > > > longer.e > > K > > I don't recall either DECnet end-node or Routing-IV ever being includedeH > > in the base OVMS license. You always had to buy them separately, andI > > choose one or the other. Of course, DECnet-IV for OVMS-Alpha does nottG > > provide the routing functionality. DVNETEND is the best you can do.p >  > See above.   Yes. See above.    > [snip] > H > yeah, I have a W98 and a W98SE box networked, no NT domain controller,I > file and print sharing works under control of some parameter I have yetlD > to determine, and rarely when I need it most. Ah well, W98 box has > recently been linuxed.  F I have no problem sharing my LaserJet 1100A with my wife's machine (orB any other that may one day be on the home LAN), likewise my wife'sE OfficeJet K Series. No NT domain here, either - which is o.k. because 9 when I enabled sharing, I wasn't asked for a domain name.)   --   David J. DachteraE dba DJE Systemsa http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/e   ------------------------------  # Date: Thu, 14 Mar 2002 02:15:00 GMTn1 From: "David J. Dachtera" <djesys.nospam@fsi.net>uA Subject: Re: Viability of a Commercial ISO-9960 formatter for VMS ' Message-ID: <3C9009A4.88C56D9C@fsi.net>s   Carl Perkins wrote:e > * > tim.llewellyn@blueyonder.co.uk writes...Q > }Sorry, David, NAS has been around for longer than that, maybe the turbochannelrN > }alphas didn't have it (can't remember) but all alphas with PCI buses that IN > }can recall having anything to do with came with NAS of some sort. Enough to% > }run IP and DECNET end node anyway.c  > }Tim.Llewellyn@cableinet.co.uk > L > Our DEC 3000m600 systems (turbochannel, bought in 1994) came with NAS-250.N > This high-end "client" license covered quite a bit. According to the utility > that tells you these things: > ? > The NET-APP-SUP-250 license enables the following components:  > [snip]  ! NET-APP-SUP-250 is not OVMS-BASE.)   -- w David J. Dachterac dba DJE Systemsv http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  # Date: Thu, 14 Mar 2002 02:15:38 GMTm1 From: "David J. Dachtera" <djesys.nospam@fsi.net>(A Subject: Re: Viability of a Commercial ISO-9960 formatter for VMSd' Message-ID: <3C9009CA.C417D85C@fsi.net>e   Robert Deininger wrote:o > = > In article <3C8ECD82.237E3BE9@fsi.net>, "David J. Dachtera"n  > <djesys.nospam@fsi.net> wrote: >  > >> }You buy VMS, you get VMS.  > >>I > >> Normally you also get a package of other licenses too, NAS or EIS ors > >> whatever. > >e: > >Yes - if you buy them, you get them. If not, you don't. > >t2 > >> }Want networking? Sorry, that's extra (or was > >> }up until recently).n > >>I > >> This is only the case if by "recently" you mean "for the last decadeE* > >> or more". And that's only talking IP, > >tI > >I don't recall UCX being included in the OVMS base license much beforei! > >about 1998 or so, maybe later.  > >t  > >> DECnet has been there a lot > >> longer. > >iJ > >I don't recall either DECnet end-node or Routing-IV ever being includedG > >in the base OVMS license. You always had to buy them separately, andyH > >choose one or the other. Of course, DECnet-IV for OVMS-Alpha does notF > >provide the routing functionality. DVNETEND is the best you can do. > K > NAS packages have been included with all the systems I dealt with for theaJ > last decade or more, and included UCX, DECnet (end-node), and DECwindowsB > Motif.  The names of the NAS packages have changed occasionally. > G > I don't think any of these products has been included in the VMS baseCH > license.  But the (separate) license has been bundled with new machine > purchases.   ...if you bought them that way.a   -- n David J. Dachteran dba DJE Systemsn http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  # Date: Thu, 14 Mar 2002 02:32:04 GMT 1 From: "David J. Dachtera" <djesys.nospam@fsi.net>eA Subject: Re: Viability of a Commercial ISO-9960 formatter for VMSn' Message-ID: <3C900D06.5B06E648@fsi.net>    "John E. Malmberg" wrote:= >  > David J. Dachtera wrote: > > Carl Perkins wrote:  > > 9 > >>"David J. Dachtera" <djesys.nospam@fsi.net> writes...a > >>}Carl Perkins wrote: > >>}> > <SNIP> >  > >>} J > >>}You buy W/9x, W/NT, W2K, you also get file sharing (no extra charge),% > >>}printer sharing (no extra cost),= > 8 > Did you miss the Microsoft URLs that I posted earlier?   No.   G > File and print sharing is free only to Workstations, and then you areb' > limited by license to 10 connections.-    What constitutes a "connection"?  D > And the only way to go to the 11th connection is to upgrade to theD > Server license, + buy 5 connections.  (5 per server file and printI > licenses are (were?) bundled with the server product.  Quite a bit of a  > jump.b > H > File and print sharing from a Microsoft Server costs extra.  It is not > free from Microsoft.  G Between "workstations" (PCs), I doubt any family or office small enoughd7 to survive without a "server" will exceed those limits.r   > [snip]J > The operating system license for the client does not include the license4 > to access the file and print services on a server.  H ...but does not prevent you from sharing local resources with other PCs.   > [snip]F > Building a computer by ordering the individual boards will also cost  > more than buying it assembled.  E Really? My experience was the reverse: I bought parts for circa. $640]H (case, mobo, CPU, memory, NIC, hard drive, CD-ROM, ...) and built for myG wife a machine that would have retailed (with WhineBloze) for in excessrF of $1,200, based on what my step-daughter paid for her W/98 box circa.1 the same time (prices have come down, of course).l   > [snip] > Of course the    (Advanced Server)   ; > package includes a terminal emulator that actually works,w  F I still prefer Reflection/2 or -/4 to PowerTerm, but PowerTerm will do in a pinch.d   > I > LAT protocol, DECNET, and an X-11 emulator that is one of the best thati > I have tested. > C > The X-11 emulator is lower cost than most of the competition too.n  : Too bad they don't sell it (X11 aka eXcursion) separately.    > So it is quite a good package. > J > Even affordable to the serious hobbyist that wants to use a PC to access1 > their VMSCLUSTER that is stuck in the basement.i > <SNIP> > E > The summary is that there are too many differences to do a straightc > comparisons. > G > And the many companies that sell PC software are finding that the lowpH > mass market price is not covering the long term support costs, and are= > either changing the licensing conditions or raising prices.e  G ...or making a separate business of support and charging separately forh* it so each business unit pays its own way.   -- t David J. Dachterao dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  # Date: Thu, 14 Mar 2002 02:39:37 GMTy1 From: "David J. Dachtera" <djesys.nospam@fsi.net>tA Subject: Re: Viability of a Commercial ISO-9960 formatter for VMSc' Message-ID: <3C900ED5.67368BF2@fsi.net>    John Santos wrote: >  > Much snippage... > / > On Wed, 13 Mar 2002, David J. Dachtera wrote:r > > Carl Perkins wrote:  > > >s > [...]hJ > > > Normally you also get a package of other licenses too, NAS or EIS or > > > whatever.C > >s; > > Yes - if you buy them, you get them. If not, you don't.f > >y3 > > > }Want networking? Sorry, that's extra (or was  > > > }up until recently). > > > J > > > This is only the case if by "recently" you mean "for the last decade+ > > > or more". And that's only talking IP,  > > J > > I don't recall UCX being included in the OVMS base license much before" > > about 1998 or so, maybe later. > > ! > > > DECnet has been there a lotp
 > > > longer.d > >mK > > I don't recall either DECnet end-node or Routing-IV ever being included H > > in the base OVMS license. You always had to buy them separately, andI > > choose one or the other. Of course, DECnet-IV for OVMS-Alpha does not.G > > provide the routing functionality. DVNETEND is the best you can do.. > >m > G > When we got our first Alpha, a 3000-300 in late 1993, it came bundled  > with NAS-250,    NAS-250 is not OVMS-BASE.f  ; > which includes DECWindows (don't remember if it was Motif-G > yet, but that eventually was included in an update under the original G > license), UCX, DECnet end-node, and VMSCluster.  (The cluster licenseuG > may have been somewhat restricted, but I think we could do everythingg > we needed to do with it.)n > G > > > (Of course a lot of people, like us, still use 3rd party softwares > > > for the IP stuff > >IB > > I still prefer Multinet, 2nd choice: TCPware, 3rd choice: UCX. >  > [...]  > >iL > > In a cluster, yes. DECnet provides some limited functionality, also, but= > > is not recommended for any serious production work, IMHO.e > 
 > Why not?  D Try it. You're likely to find that even 100Mbit E-net runs out steam before SCSI-2 or better.    G Not that you *CAN'T* use it for serious production work - Mark Levy and H I did that with twin 750s long prior to VAXCLUSTER's appearance (VAX/VMSC V3.4 or so, circa. 1983). I just wouldn't recommend it if there's a ( viable, affordable (???!!!) alternative.   > [...]- > L > > I don't believe "Mozilla" (CSWB) is currently "shipping" at all. I thinkG > > you still must register for a download, and think it's still T1.0 -1) > > pre-general-release (pre-Gamma-Test).  > " > It's now officially a Beta test.  A We'll see how it goes in what I call "Gamma Test" (others call itrB anything from "General Release" to "General Availability", "StableA Release", etc.). Last I heard, it still had fairly nasty resourcet6 requirement problems (takes *TONS* of RAM, mucho CPU).   -- V David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/e   ------------------------------   Date: 13 Mar 2002 21:04 CSTe' From: carl@gerg.tamu.edu (Carl Perkins)pA Subject: Re: Viability of a Commercial ISO-9960 formatter for VMS-- Message-ID: <13MAR200221045908@gerg.tamu.edu>r  5 "David J. Dachtera" <djesys.nospam@fsi.net> writes...  }Carl Perkins wrote:+ }> tim.llewellyn@blueyonder.co.uk writes...2R }> }Sorry, David, NAS has been around for longer than that, maybe the turbochannelO }> }alphas didn't have it (can't remember) but all alphas with PCI buses that IaO }> }can recall having anything to do with came with NAS of some sort. Enough toy& }> }run IP and DECNET end node anyway.! }> }Tim.Llewellyn@cableinet.co.ukt }>  M }> Our DEC 3000m600 systems (turbochannel, bought in 1994) came with NAS-250. O }> This high-end "client" license covered quite a bit. According to the utilityP }> that tells you these things:g }> u@ }> The NET-APP-SUP-250 license enables the following components:	 }> [snip]  } " }NET-APP-SUP-250 is not OVMS-BASE. }  }--  }David J. Dachtera  C And what is the name of the Windows NT license? What is the name ofbB the File Sharing license? Are they the same? (The answer is below,
 actually.)  > When you buy the system it doesn't come with just the base VMS? license unless you work realy really hard to make them sell you B one like that (you will find it even harder than convincing Compaq- to sell you a VMS system in the first place).d  C Our DEC 3000m600 systems CAME WITH the NAS-250 license. It was partnC of the package. It was not separate. They were together. Which partM. of this concept are you failing to understand?  C You'll find that there are a variety of client access licenses thatdB you need to actually use a lot of NT's stuff once you go beyond 10D or so systems in a workgroup up to an NT Server system. They are notB the same license as the Windows NT license. They are separate. ButB NT Workstation includes a small number of them for that are usefulD for small network purposes. Once you get past that, you'll find thatF the situation is essentially the same as with VMS - different licensesA with different names that you have to load to do things (assumingnA you want to do them legally, for the most part it doesn't seem tot> enforce the licensing very stictly). This is functionally veryB similar to the way VMS comes with a NAS license to let you do some/ stuff, but requires you to get more to do more.6   --- Carl   ------------------------------  % Date: Wed, 13 Mar 2002 22:27:00 -0500 2 From: rdeininger@mindspring.com (Robert Deininger)A Subject: Re: Viability of a Commercial ISO-9960 formatter for VMStK Message-ID: <rdeininger-1303022227000001@1cust130.tnt3.nashua.nh.da.uu.net>   ; In article <3C9009A4.88C56D9C@fsi.net>, "David J. Dachtera"8 <djesys.nospam@fsi.net> wrote:    " >NET-APP-SUP-250 is not OVMS-BASE.  O So what?  If they bundle OVMS-BASE and NAS with the system purchase, who cares?o   ------------------------------  % Date: Wed, 13 Mar 2002 22:30:17 -0500m2 From: rdeininger@mindspring.com (Robert Deininger)A Subject: Re: Viability of a Commercial ISO-9960 formatter for VMS K Message-ID: <rdeininger-1303022230170001@1cust130.tnt3.nashua.nh.da.uu.net>   ; In article <3C9009CA.C417D85C@fsi.net>, "David J. Dachtera"s <djesys.nospam@fsi.net> wrote:    H >> I don't think any of these products has been included in the VMS baseI >> license.  But the (separate) license has been bundled with new machines
 >> purchases.t >(  >...if you bought them that way.  G Were VMS alpha systems ever offered without the whole bundle?  I'm surenI they were sometimes, but every price quote on an alpha that I was offerediI in the past decade included bundled NAS licenses of one form or another. iC I would not have been able to by a VMS system without NAS, at leasttJ without begging for special permission.  Certainly the default has been to4 bundle NAS whenever a VMS base license was included.   ------------------------------  # Date: Thu, 14 Mar 2002 03:47:53 GMTh1 From: "David J. Dachtera" <djesys.nospam@fsi.net>sA Subject: Re: Viability of a Commercial ISO-9960 formatter for VMSg' Message-ID: <3C901ED7.28A2B919@fsi.net>    Robert Deininger wrote:  > = > In article <3C9009A4.88C56D9C@fsi.net>, "David J. Dachtera"   > <djesys.nospam@fsi.net> wrote: > $ > >NET-APP-SUP-250 is not OVMS-BASE. > Q > So what?  If they bundle OVMS-BASE and NAS with the system purchase, who cares?n  F When you find yourself with a corrupted/deleted .LDB and an incomplete0 set of hard-copy PAKs, you'll care a great deal.  ? ...and that's "If they bundle OVMS-BASE and NAS with the system ' purchase". Don't count your chickens...o   -- p David J. Dachterau dba DJE Systemsm http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  # Date: Thu, 14 Mar 2002 03:49:16 GMTu1 From: "David J. Dachtera" <djesys.nospam@fsi.net>eA Subject: Re: Viability of a Commercial ISO-9960 formatter for VMSD' Message-ID: <3C901F29.80BB0A62@fsi.net>e   Robert Deininger wrote:r > = > In article <3C9009CA.C417D85C@fsi.net>, "David J. Dachtera"-  > <djesys.nospam@fsi.net> wrote: > J > >> I don't think any of these products has been included in the VMS baseK > >> license.  But the (separate) license has been bundled with new machineg > >> purchases.E > >B" > >...if you bought them that way. > I > Were VMS alpha systems ever offered without the whole bundle?  I'm sureiK > they were sometimes, but every price quote on an alpha that I was offered"J > in the past decade included bundled NAS licenses of one form or another.E > I would not have been able to by a VMS system without NAS, at least.L > without begging for special permission.  Certainly the default has been to6 > bundle NAS whenever a VMS base license was included.  F I've seen situ's where someone got the bright idea to say, "Oh, we canG save a few bux here: we already have a Multinet license - we don't needi NAS".o   --   David J. Dachterau dba DJE Systemsa http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  % Date: Wed, 13 Mar 2002 23:50:17 -0500   From: John Santos <JOHN@egh.com>A Subject: Re: Viability of a Commercial ISO-9960 formatter for VMSu6 Message-ID: <1020313234120.18492A-100000@Ives.egh.com>  - On Thu, 14 Mar 2002, David J. Dachtera wrote:e   > John Santos wrote: > >  > > Much snippage... > > 1 > > On Wed, 13 Mar 2002, David J. Dachtera wrote:  > > > Carl Perkins wrote:v > > > >.	 > > [...] L > > > > Normally you also get a package of other licenses too, NAS or EIS or > > > > whatever.e > > >i= > > > Yes - if you buy them, you get them. If not, you don't.  > > >r5 > > > > }Want networking? Sorry, that's extra (or wash > > > > }up until recently). > > > >nL > > > > This is only the case if by "recently" you mean "for the last decade- > > > > or more". And that's only talking IP,i > > >fL > > > I don't recall UCX being included in the OVMS base license much before$ > > > about 1998 or so, maybe later. > > >a# > > > > DECnet has been there a lot  > > > > longer.p > > >dM > > > I don't recall either DECnet end-node or Routing-IV ever being includedmJ > > > in the base OVMS license. You always had to buy them separately, andK > > > choose one or the other. Of course, DECnet-IV for OVMS-Alpha does notiI > > > provide the routing functionality. DVNETEND is the best you can do.5 > > >i > > I > > When we got our first Alpha, a 3000-300 in late 1993, it came bundled  > > with NAS-250,  >  > NAS-250 is not OVMS-BASE.a  < I didn't say it was.  I said it was *BUNDLED* with VMS.  Who: cares if it is a separate PAK if there is no extra charge?   > = > > which includes DECWindows (don't remember if it was MotifnI > > yet, but that eventually was included in an update under the originaluI > > license), UCX, DECnet end-node, and VMSCluster.  (The cluster license I > > may have been somewhat restricted, but I think we could do everythingy > > we needed to do with it.): > > I > > > > (Of course a lot of people, like us, still use 3rd party software  > > > > for the IP stuff > > >ED > > > I still prefer Multinet, 2nd choice: TCPware, 3rd choice: UCX. > > 	 > > [...]A > > >lN > > > In a cluster, yes. DECnet provides some limited functionality, also, but? > > > is not recommended for any serious production work, IMHO.  > >  > > Why not? > F > Try it. You're likely to find that even 100Mbit E-net runs out steam > before SCSI-2 or better. . >   > I was comparing it to IP (NFS), which is used for a great deal of serious production work.d  I > Not that you *CAN'T* use it for serious production work - Mark Levy andgJ > I did that with twin 750s long prior to VAXCLUSTER's appearance (VAX/VMSE > V3.4 or so, circa. 1983). I just wouldn't recommend it if there's a * > viable, affordable (???!!!) alternative. > 	 > > [...]y > > N > > > I don't believe "Mozilla" (CSWB) is currently "shipping" at all. I thinkI > > > you still must register for a download, and think it's still T1.0 -n+ > > > pre-general-release (pre-Gamma-Test).e > > $ > > It's now officially a Beta test. > C > We'll see how it goes in what I call "Gamma Test" (others call iteD > anything from "General Release" to "General Availability", "StableC > Release", etc.). Last I heard, it still had fairly nasty resourcea8 > requirement problems (takes *TONS* of RAM, mucho CPU).  E It is very slow on an AS200 4/100, though much faster with 256MB thanuC it was with 80MB, which was unusable.  Netscape 4.73 on DEC Unix ongC the same hardware is fine (not screamingly fast but fast enough nota to be annoying.)   -- c John SantosC Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  # Date: Thu, 14 Mar 2002 04:52:02 GMTf4 From: Tim Llewellyn <tim.llewellyn@blueyonder.co.uk>A Subject: Re: Viability of a Commercial ISO-9960 formatter for VMSi0 Message-ID: <3C902AD8.C14BD8C1@blueyonder.co.uk>   "David J. Dachtera" wrote: >  > Tim Llewellyn wrote: > >e > > "David J. Dachtera" wrote: > > > >Carl Perkins wrote: > > > >yL > > > > This is only the case if by "recently" you mean "for the last decade- > > > > or more". And that's only talking IP,. > > >fL > > > I don't recall UCX being included in the OVMS base license much before$ > > > about 1998 or so, maybe later. > > : > > Sorry, David, NAS has been around for longer than that >  > NAS is not OVMS-BASE. Sorry.  @ Well, if you want to quibble. Do others apart from islandco sellD alphas without licences for VMS? Obviously, hobbiest use is not your concern.   >  > >, maybe the turbochanneleO > > alphas didn't have it (can't remember) but all alphas with PCI buses that I E > > can recall having anything to do with came with NAS of some sort.  >  > ...if you bought it that way.   E I personally wouldn't propose any other way for serious business use.?   > 
 > > Enough toy& > > run IP and DECNET end node anyway. > H > UCX was not included in OVMS-BASE until recently, within the last four- > or five years. DVNETEND still isn't, AFAIK.v >   P You mean you can run IP services now without an appropriate NAS/EIP/UCX licence? Things are changing if correct.     f > 
 > > [snip] > >wJ > > yeah, I have a W98 and a W98SE box networked, no NT domain controller,K > > file and print sharing works under control of some parameter I have yetnF > > to determine, and rarely when I need it most. Ah well, W98 box has > > recently been linuxed. > H > I have no problem sharing my LaserJet 1100A with my wife's machine (orD > any other that may one day be on the home LAN), likewise my wife'sG > OfficeJet K Series. No NT domain here, either - which is o.k. becausey; > when I enabled sharing, I wasn't asked for a domain name.   P Well, at least its good to be reassured that because its M$, its inconsistent:-)D I do have a DSL connection, maybe I am seeing DOS attacks on my LAN,- even though I have a router with a firewall. rO Somedays I can map drives, print remotely, other days, I just sit there waiting D for WIndows to timeout. Anyway, enough of that, this is comp.os.vms.   regards     c -- y tim.llewellyn@blueyonder.co.uk t  F * tim.llewellyn@cableinet.co.uk address will cease to work June 2002 *   ------------------------------  # Date: Thu, 14 Mar 2002 04:57:32 GMT 4 From: Tim Llewellyn <tim.llewellyn@blueyonder.co.uk>A Subject: Re: Viability of a Commercial ISO-9960 formatter for VMS10 Message-ID: <3C902C7E.842DC8D4@blueyonder.co.uk>   "David J. Dachtera" wrote:  a > H > I've seen situ's where someone got the bright idea to say, "Oh, we canI > save a few bux here: we already have a Multinet license - we don't needr > NAS".   yB NAS 150 was cheaper than UCX, when I was in that situation (1997).   -- s tim.llewellyn@blueyonder.co.uk g  F * tim.llewellyn@cableinet.co.uk address will cease to work June 2002 *   ------------------------------   Date: 14 Mar 2002 00:19:59 GMT' From: dashw459@aol.comeatspam (Doug W.)i5 Subject: Re: Why not we guys buy OpenVMS from Compaq?h9 Message-ID: <20020313191959.14633.00001814@mb-fc.aol.com>o  N << I hope someone in Spitbrook elects to address your statement above, since II just don't find it credible.  In particular, something is very wrong if a J GS160 can't give you at least the performance of a GS140 (if it can't, theL server team should be shot - which, given their apparent contribution to theL Alphacide might not be a bad idea anyway) - with no application modification
 whatsoever >>   O I'll try to establish credibility by giving more details, history, what we havesN looked into, our experiences and to some extent what they have meant and where
 we are going.h  M We were running with 8 CPU GS140s on what today are rather slow CPUs.  Only ayG single application was run on the units consisting of many simultaneouseH processes communicating extensively through global memory.  Think of theL application as a medium sized transaction box.   We tried increasing the CPUM count, but diminishing returns made 8 CPUs appear to be a good place to stop.   M Our team did go to Nashua to speak with VMS engineering at a high level.  OurdN primary concerns were performance and scaling up the transaction rate.  At theF time we believed in consolidating almost everything into single units.N Offloading and frontends were avoided in favor of administrative simplicity, a1 low box count and minimal floorspace consumption.   M At Nashua we explained our application in detail to engineering.  EngineeringlN suggested the GS NUMA architecture.  However, they were very candid about NUMAJ performance problems.  Based on their experience with other customers withI similar applications, engineering advised us to consider reorganizing ourDF application to deal with NUMA.  Engineering was exceedingly helpful inN providing specific reorganization suggestions and even diagrams.  Their candorN and willingness to talk openly about NUMA performance impressed me. They couldO not predict how much of an affect NUMA would have, but they were pretty sure weaJ were going to see it.  I met people who I formerly had only read about. OhK yeah, I managed to get upstairs to see what Stephen Hoffman looked like andrM collect some trinkets from Sue S.  I met Mr. Colorwheel himself, was suitablyhM intimidated by Clair Grant (my fault, I still can't figure out why) and met au1 lot of good engineers doing interesting VMS work.L  D So we did not go into a GS160 investigation blind or expecting greatK performance.  The NUMA situation on the current GSs is known, openly spoken M about and with a little reaching out you can find others who have experiencedv it.z  I I spent 3 days in the Compaq benchmark test center in Toronto performancesK testing our application on an 8 CPU GS 160.  Our management chose the GS160eL configuration.  Basically they wanted to compare the SMP performance of an 8N CPU GS140 with the NUMA performance of an 8 CPU GS160 using much CPUs.  ApplesF to apples, no code changes.  For us, the performance results were very disappointing. s  N Yes, unmodified code on a GS160 can run slower than on a GS140.  I don't thinkO you need anyone from Spitbrook to verify it.  By the way, I experimented moving N memory boards around the 160.  Performance varied by up to 28% but never quiteO reached the GS140 level.  The Toronto people were not surprised at our results. N  They had seen it before.  They said the code would fly if it were reorganized	 for NUMA.1  O I did try to sell management on a GS160.  It was a disaster.  You want a lot ofCG money?  It runs slower than what you have?  You want even more money torO reorganize legacy code?  You can get it to run faster but you can't tell me howlA much faster?  Have you heard of our policy of no new work on VMS?t  L So off we went to the brave new world of too many ES40s and PC frontending. N While the VMS world is waiting to exhale, things have changed here.  There areN now a lot of ES40s with nothing running on them (hint).  So while I would likeJ to try the ES45, management wants to find some kind of use for the surplus ES40s.     t  N Future hardware always looks very promising. Something similar enough to 8 wayN SMP will be available in some form or another for VMS.  But I needed somethingM real last December.  Besides, management is being very frank about VMS use in 
 this company.e  O I am pleased with ES40 performance running an application that was designed for)K 8 CPUs.  But the old 8 way SMP GS140s with slower chips push a lot more TPSnO through our application.  I can only imagine what a new  8 way SMP design usingiK the latest CPUs would do.  Sadly, such a product is not currently availables
 from Compaq. u  M So what is incredible?  The GS architecture knocked the legs out from under aeN small but loyal number of legacy SMP users requiring more than 4 CPUs.  As theI VMS user base probably runs legacy applications, its surprising that codecM changes are required to use the latest hardware efficiently.  Management hereyK does not want to modify legacy code.  Also incredible is the shabby way VMS O users have been treated over the past year and their willingness to put up withnK it.  The lack of certainty over the future of VMS and what kind of hardwaresM will be used is ridiculous.  I am sure Compaq management knows this is no waysJ to run a business.  And last, but not least, theres the knock on my door. C After 20 years of VMS, I am being transferred to a Windows project.s   ------------------------------  % Date: Wed, 13 Mar 2002 22:56:54 -0500n2 From: rdeininger@mindspring.com (Robert Deininger)5 Subject: Re: Why not we guys buy OpenVMS from Compaq? K Message-ID: <rdeininger-1303022256550001@1cust130.tnt3.nashua.nh.da.uu.net>e  9 In article <20020313191959.14633.00001814@mb-fc.aol.com>,n( dashw459@aol.comeatspam (Doug W.) wrote:    P >I am pleased with ES40 performance running an application that was designed forL >8 CPUs.  But the old 8 way SMP GS140s with slower chips push a lot more TPSP >through our application.  I can only imagine what a new  8 way SMP design usingL >the latest CPUs would do.  Sadly, such a product is not currently available >from Compaq.    You don't have to imagine.  G There's a relatively new VMS "test-drive" lab at Spit Brook.  CustomersnB can schedule time to go in and try their real applications on realI hardware, with dedicated support folks to help them set up and test.  TheuG lab is secure, so you can protect any proprietary data and software you>J might have.  There are a variety of systems in the lab, including at least" one prototype EV7 system, I think.  J You might want to look into another visit to Spit Brook to try out various configurations.s  H I didn't notice what version of VMS you are running.  If it's not 7.3 orA later, you may see significant performance gains, on your currentoC hardware, if you upgrade.  Again, the test drive lab could help youe	 evaluate.e  N >So what is incredible?  The GS architecture knocked the legs out from under aG >small but loyal number of legacy SMP users requiring more than 4 CPUs.   J That was unplanned.  The high-P scaling of the big GS boxes was worse thanC expected, and by the time folks realized the extent of the problem,,J several years of hardware roadmap was pretty well frozen.  I'm pretty sureI the experience of GS performance was considered very carefully by the EV7s system designers.8  G Some applications seem to scale much better than what you describe, andC0 many customers seem happy with their GS systems.   >As thetJ >VMS user base probably runs legacy applications, its surprising that codeN >changes are required to use the latest hardware efficiently.  Management hereL >does not want to modify legacy code.  Also incredible is the shabby way VMSP >users have been treated over the past year and their willingness to put up with >it.    4 That treatment hasn't originated from the engineers.  G >The lack of certainty over the future of VMS and what kind of hardwarel >will be used is ridiculous.    D You really, really need to work with a VMS ambassador.  Once you getI connected to the right people, there are VMS folks in Nashua dedicated tomH helping folks like you solve your problems.  The situation seems to have. improved significantly in the last year or so.  < The newsgroup is NOT the best way to work toward a solution.   ------------------------------  % Date: Wed, 13 Mar 2002 16:34:36 -0500I; From: "Brian Tillman" <tillman_brian@notnoone.notnohow.com>MI Subject: Re: [OT]  NY Times Op-Ed misses point about unreliablity of PC's $ Message-ID: <3c8fc641$1@news.si.com>   >> What's the point?  Use a VT!  >eJ >because users and some management think vts are old fashioned and windoze; >is the future ... so give them the best of both worlds ...a  J I was in my doctor's office today and while I was waiting, I looked at theF PCs the people behind the desk were using.  All were fairly modern PCsJ running an up-to-date Windows.  What were the applications they were using7 on those fancy PCs?  DOS-based character cell programs!i --A Brian Tillman                   Internet: tillman_brian at si.com"A Smiths Aerospace                          tillman at swdev.si.com = 3290 Patterson Ave. SE, MS      Addresses modified to preventy< Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@"8        This opinion doesn't represent that of my company   ------------------------------  # Date: Thu, 14 Mar 2002 02:48:56 GMTo1 From: "David J. Dachtera" <djesys.nospam@fsi.net> I Subject: Re: [OT] US/Europe living costs, was: Re: VMS sys admin salaries-' Message-ID: <3C901105.D9F399FC@fsi.net>0   Paul Repacholi wrote:e > 5 > "David J. Dachtera" <djesys.nospam@fsi.net> writes:  > E > > When I was in the hospital 2 days prior to Fall '99 DECUS (kidneyoF > > stone), a room and care (blood/urine tests, MRI, Iv saline and oneE > > shot of morphine) for roughly 48 hours came to somewhere close to  > > $6,000 U.S., > F > > This past week, a blood draw and battery of blood tests (trying toC > > track down a hormone imbalance) came to some $240 + $40 for the  > > office visit.c > G > Hum, I was in hospital for 24 days. 4 hrs OR, 10 days ICU, 14 days ont+ > ward. Total cost, $130 for the Ambulance.e  E Ah, yes. My out-of-pocket on that experience was in the $1,100 range. 9 Not bad considering my doctor was not in the health plan.h  H > A large number of people with no health care is a disaster in waiting.= > The disease your taxes pay for, is the one you don't get...S   -- e David J. Dachtera  dba DJE Systemsl http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/h   ------------------------------   End of INFO-VAX 2002.143 ************************