1 INFO-VAX	Fri, 22 Nov 2002	Volume 2002 : Issue 646       Contents:+ Re: Advanced server 7.3 on OVMS 7.3 problem + Re: Advanced server 7.3 on OVMS 7.3 problem , Re: ANN: libxml2 2.4.27 XML parser available, Re: ANN: libxml2 2.4.27 XML parser available, Re: ANN: libxml2 2.4.27 XML parser available1 Anybody with a HDS 7700 need some DK308-90 disks?  Changing vms passwords via web1 Clustering, CSWS/Apache, and TCP/IP cluster alias 5 Re: Clustering, CSWS/Apache, and TCP/IP cluster alias 5 Re: Clustering, CSWS/Apache, and TCP/IP cluster alias 5 Re: Clustering, CSWS/Apache, and TCP/IP cluster alias 5 Re: Clustering, CSWS/Apache, and TCP/IP cluster alias  Re: Database for VMS Re: Database for VMS" Re: DCPS V2.2 field test available" Re: DCPS V2.2 field test available" Re: DCPS V2.2 field test available: Re: DECW$XFS (X font service) - anyone successfully using?: Re: DECW$XFS (X font service) - anyone successfully using?3 Re: DELL Itanium Servers - May be we can run OVMS ? 3 Re: DELL Itanium Servers - May be we can run OVMS ? 3 Re: DELL Itanium Servers - May be we can run OVMS ? 3 Re: DELL Itanium Servers - May be we can run OVMS ? 3 Re: DELL Itanium Servers - May be we can run OVMS ? 3 Re: DELL Itanium Servers - May be we can run OVMS ? 3 Re: DELL Itanium Servers - May be we can run OVMS ?  Re: Get owner info from UAF  Re: Get owner info from UAF  Re: Get owner info from UAF  Re: Gnat Ada for VMS Re: Gnat Ada for VMS Re: Gnat Ada for VMS1 Re: How to delete a command from the DCL-Tables ? - Re: I give up, need Solaris for VMS jobs now. P Re: Is there a hobbyist program for other HP products? Was Re: What's going on w Re: Jaw dropping EV7 systems Re: Jaw dropping EV7 systems Re: Jaw dropping EV7 systems Re: Misuse of SYSPRV in VMS  Re: Misuse of SYSPRV in VMS  Re: Misuse of SYSPRV in VMS  Re: OT: Hamburger Flipping Re: OT: Hamburger Flipping Re: Pathworks Mac vs VMS 7.3-1% Re: poor Gigabit Ethernet performance % Re: poor Gigabit Ethernet performance % Re: poor Gigabit Ethernet performance $ Re: Read/Write Locks in an SMP setup$ Re: Read/Write Locks in an SMP setup? Re: Sue Skonetski recognized at OpenVMS Symposium for hard work ' Re: System uptime - The Uptimes Project  TCPIP$XDM : Security issue Re: Termination mailbox problem  Re: Termination mailbox problem  Re: Termination mailbox problem  Re: Termination mailbox problem  Re: Termination mailbox problem  Re: Termination mailbox problem  Re: TPU scripting for MAIL ?+ Trouble with NFS since upgrade 7.3 -> 7.3-1 ! Re: What is DAP status code 5067? @ Re: What's going on with the 7.3.1 VAX hobbyist kit at Montagar?* Re: Where can I find patches to my system?  F ----------------------------------------------------------------------  # Date: Fri, 22 Nov 2002 11:20:48 GMT - From: "Peter Jan" <PeterJan.Rusch@Compaq.com> 4 Subject: Re: Advanced server 7.3 on OVMS 7.3 problem0 Message-ID: <kqoD9.1$2U4.91226@news.cpqcorp.net>   Try to find the domain with    $ nbshow knbstatus MERKABE   You get a list like.   Name            Soc Num Status+ NodeName          x20   1 Unique Registered + NodeName          x00   2 Unique Registered / SPSD                  x00   3 Group  Registered * DomainName       x03   4 Group  Registered* DomainName       x20   5 Group  Registered, INet~Services      x1c   6 Group  Registered& IS~DomName^^ x00   7 Unique Registered* DomainName       x01   8 Unique Registered   Peter Jan Rusch   : "rob merritt" <merritt.robert@spsd.sk.ca> wrote in message6 news:b6bf97d5.0211211521.2b8d6e1@posting.google.com... > Hi >  > & > I have a Advanced server 7.3 install >  > $ ucx sho ver  > 9 >   Compaq TCP/IP Services for OpenVMS Alpha Version V5.1 4 >   on a AlphaStation 250 4/266 running OpenVMS V7.3 > 6 > when I go and pwrk$config it I cant join the domain: >  > F > Enter the name of the primary domain controller for domain ACADEMUS:	 > MERKABA > > Enter the name of the administrator account: [Administrator]2 > Enter the account password in the required case: > Re-enter to verify password: > % > Confirming domain name with MERKABA  > G > PWRK-F-COMERR, error communicating with the primary domain controller  > Error getting domain name  >  >  > the logs give me:  >  > : > PWRK$COMMONROOT:[000000.LOGS]PWRK$KNBDAEMON_TSCTST.LOG;8 > 9 >  Sun Nov 17 17:55:32 2002 PH Address: AA-00-04-00-62-08 4 >  Sun Nov 17 17:55:33 2002 IP Address: 172.16.0.1997 >  Sun Nov 17 17:55:33 2002 NBNS IP Address: 172.16.1.2 2 >  Sun Nov 17 17:55:33 2002 KNB Driver linked with  > <netbios/streams/pwips> succes > sfully= >  Sun Nov 17 17:55:33 2002 KNB Driver : Starting knblink ... ? >  Sun Nov 17 17:55:33 2002 KNB Driver : Starting knblmhost ... ? >  Sun Nov 17 17:55:33 2002 preloaded nbname MERKABA         at 
 > 172.16.1.10 E >  Sun Nov 17 17:55:33 2002 KNB Driver : Starting knbbindresolver ... E >  Sun Nov 17 23:43:21 2002 resolve_name: LMHOSTS Lookup for LANGROUP 
 >     [1b] > failed knbcb 38 8 >  Sun Nov 17 17:55:31 2002  dllink t_open successful on > netbios/streams/dl8 >  Sun Nov 17 17:55:31 2002  dllink t_open successful on > netbios/streams/dlpi1 G >  Sun Nov 17 17:55:31 2002  Exiting dllink with success we have one or  > more dlpi 	 > devices D >  Sun Nov 17 17:55:31 2002 netbios/streams/nbes opened successfullyD >  Sun Nov 17 17:55:31 2002 Microsoft NetBEUI : Starting nbelink ...  >  Sun Nov 17 17:55:31 2002 done > $  >  >  >  > $ > 1: I can ping and nslookup the PDC >  > $ ping  merkaba + > PING MERKABA (172.16.1.10): 56 data bytes 9 > 64 bytes from 172.16.1.10: icmp_seq=0 ttl=128 time=2 ms 9 > 64 bytes from 172.16.1.10: icmp_seq=1 ttl=128 time=1 ms 9 > 64 bytes from 172.16.1.10: icmp_seq=2 ttl=128 time=0 ms 9 > 64 bytes from 172.16.1.10: icmp_seq=3 ttl=128 time=0 ms  >  > ! > ----MERKABA PING Statistics---- ; > 4 packets transmitted, 4 packets received, 0% packet loss ) > round-trip (ms)  min/avg/max = 0/1/2 ms  >  > and nslookup merkaba:  >  > $ nslookup merkaba > Server:  sps2s.spsd.sk.ca  > Address:  172.16.1.2 >  > Name:    merkaba.SPSD.SK.CA  > Address:  172.16.1.10  >  > + > 2: how would I check Netbios connectivity  > G > the PDC is Win2000 in a natvie 2000 domain the evernt log is clean no ! > reference to the try to join...  > - > the password was tried upper and lower case  >  >  > ! > what do you think I should try?  > F > I have been throught the as Iguide doing trouble shooting for a week > and I am at a loss.  >  >  >  >  >  >  > ( > any Ideas what causing me this problem   ------------------------------  % Date: Fri, 22 Nov 2002 07:12:52 -0500 ) From: "Scott Greig" <jsgreig@geminaq.com> 4 Subject: Re: Advanced server 7.3 on OVMS 7.3 problem6 Message-ID: <TapD9.37$ed3.49118@news20.bellglobal.com>  : "rob merritt" <merritt.robert@spsd.sk.ca> wrote in message6 news:b6bf97d5.0211211521.2b8d6e1@posting.google.com... <snip>  + > 2: how would I check Netbios connectivity  > G > the PDC is Win2000 in a natvie 2000 domain the evernt log is clean no ! > reference to the try to join...  > = This might be your problem. My experience is that there is NO : primary domain controller in a Windows 2000 native domain.> So, you might try making your VMS system a member server only.   Just a thought,  Scott   - > the password was tried upper and lower case  >  >  > ! > what do you think I should try?  > F > I have been throught the as Iguide doing trouble shooting for a week > and I am at a loss.  >  >  >  >  >  >  > ( > any Ideas what causing me this problem   ------------------------------  % Date: Fri, 22 Nov 2002 16:07:53 -0000 - From: wspencer@ap.nospam.org (Warren Spencer) 5 Subject: Re: ANN: libxml2 2.4.27 XML parser available 5 Message-ID: <92CE79293warrenspencer1977@216.168.3.30>   E craigberry@nospam.telocity.com (Craig A. Berry) wrote in <craigberry- 0 F356F5.17292517112002@news.directvinternet.com>:  I >Libxml2 is an XML library written in C as part of the Gnome project. It  I >includes a fully validating parser and supports a number of XML-related  F >standards.  Version 2.4.27 has updated VMS support and builds out of I >the box on OpenVMS Alpha using DEC/Compaq C 5.7 or later.  More details  ' >and download links are available here:  >  >http://www.xmlsoft.org/ >    Craig,  J This is great news - thank you!  I've downloaded the kit and attempted to I build it on OpenVMS 7.2-1 with the DEC C compiler version 6.0.  It fails  H when compiling TRIO.C, with the errors shown below.  I poked around the L source, but didn't see why this is happening.  The TRIO_VA_START() macro is ? defined in TRIODEF.H, and TRIO.C includes it.  Any suggestions?    ws     $   write sys$output "CC/NAMES= 1 (SHORTENED)/FLOAT=IEEE/IEEE_MODE=DENORM_RESULTS/o A bject=AP1:[WSPENCER.LIBXML.LIBXML2-2_4_27.DEBUG]TRIO.OBJ; TRIO.C" D CC/NAMES=(SHORTENED)/FLOAT=IEEE/IEEE_MODE=DENORM_RESULTS/object=AP1: [WSPENCER.LI* BXML.LIBXML2-2_4_27.DEBUG]TRIO.OBJ; TRIO.CK $   cc_command /object=AP1:[WSPENCER.LIBXML.LIBXML2-2_4_27.DEBUG]TRIO.OBJ;   TRIO.  C      TRIO_VA_START(args, format); ..^ 4 %CC-E-IDEXPECTED, Identifier expected but not found.? at line number 3556 in file XXX:[LIBXML.LIBXML2-2_4_27]TRIO.C;1      TRIO_VA_START(args, format); ..^ 4 %CC-E-IDEXPECTED, Identifier expected but not found.? at line number 3623 in file XXX:[LIBXML.LIBXML2-2_4_27]TRIO.C;1      TRIO_VA_START(args, format); ..^ 4 %CC-E-IDEXPECTED, Identifier expected but not found.? at line number 3695 in file XXX:[LIBXML.LIBXML2-2_4_27]TRIO.C;1      TRIO_VA_START(args, format);   ... more of the same --     Warren Spencer' Senior Software Engineer (not a writer)  The Associated Press  + ** What's brown and sticky?    A stick.  **    ------------------------------  % Date: Fri, 22 Nov 2002 08:16:10 -0800 ' From: David Mathog <mathog@caltech.edu> 5 Subject: Re: ANN: libxml2 2.4.27 XML parser available 8 Message-ID: <20021122081610.27d39f42.mathog@caltech.edu>  " On Fri, 22 Nov 2002 16:07:53 -0000. wspencer@ap.nospam.org (Warren Spencer) wrote:   > The TRIO_VA_START() macro is  A > defined in TRIODEF.H, and TRIO.C includes it.  Any suggestions?  >  > ws >  > ! > $   write sys$output "CC/NAMES= 3 > (SHORTENED)/FLOAT=IEEE/IEEE_MODE=DENORM_RESULTS/o C > bject=AP1:[WSPENCER.LIBXML.LIBXML2-2_4_27.DEBUG]TRIO.OBJ; TRIO.C" F > CC/NAMES=(SHORTENED)/FLOAT=IEEE/IEEE_MODE=DENORM_RESULTS/object=AP1: > [WSPENCER.LI, > BXML.LIBXML2-2_4_27.DEBUG]TRIO.OBJ; TRIO.CM > $   cc_command /object=AP1:[WSPENCER.LIBXML.LIBXML2-2_4_27.DEBUG]TRIO.OBJ;   > TRIO.  > C  >   >   TRIO_VA_START(args, format); > ..^ 6 > %CC-E-IDEXPECTED, Identifier expected but not found.A > at line number 3556 in file XXX:[LIBXML.LIBXML2-2_4_27]TRIO.C;1   A Are you sure it's being included?  That's the sort of error you'd I expect if it didn't have TRIO_VA_START defined.   Look for ifdef's around H it in the .h file that might be causing it to be skipped.  If that failsL run it through the preprocessor, and then compile the output of that.  Could> be that it's a missing definition one or two layers deep under TRIO_VA_START.   Regards,   David Mathog mathog@caltech.edu> Manager, Sequence Analysis Facility, Biology Division, Caltech   ------------------------------  % Date: Fri, 22 Nov 2002 10:24:32 -0600 7 From: "Craig A. Berry" <craigberry@nospam.telocity.com> 5 Subject: Re: ANN: libxml2 2.4.27 XML parser available G Message-ID: <craigberry-65BAEA.10243222112002@news.directvinternet.com>   5 In article <92CE79293warrenspencer1977@216.168.3.30>, /  wspencer@ap.nospam.org (Warren Spencer) wrote:   , >  I've downloaded the kit and attempted to K > build it on OpenVMS 7.2-1 with the DEC C compiler version 6.0.  It fails  J > when compiling TRIO.C, with the errors shown below.  I poked around the N > source, but didn't see why this is happening.  The TRIO_VA_START() macro is A > defined in TRIODEF.H, and TRIO.C includes it.  Any suggestions?   F Hmm.  My testing was with CC 6.4 and 6.5.  They may have some feature G 6.0 doesn't.  Take one of the CC commands it spits at you verbatim and  F add /LIST/SHOW=INCLUDE at the end.  Poke around in the resulting .LIS F file to see where your macro should be getting defined but isn't.  If E you want you can send me the listing privately and I'll see if I can   spot what's happening.   ------------------------------    Date: 22 Nov 2002 10:48:41 -0800" From: Rostig1@hotmail.com (Rostig): Subject: Anybody with a HDS 7700 need some DK308-90 disks?= Message-ID: <74089cec.0211221048.61c64584@posting.google.com>   E I have 161 Hitachi DK308-90 hard disks on a pallet that I have no use F for.  I can also not find anybody else that needs them.  You can email/ me if you want them.  I am in the Southern USA.    ------------------------------    Date: 22 Nov 2002 09:40:44 -0800& From: jwparker@sfasu.edu (John Parker)' Subject: Changing vms passwords via web = Message-ID: <229a787c.0211220940.53b218f9@posting.google.com>   A We are trying to come up with a way to allow vms users to change  A their password via a web page.  (It seems to be getting just too  ? hard for people to understand how to use a telnet program). We  ? would like to be able to use the normal checks provided by vms  @ such that we can use the history list, dictionary, etc.  I have B heard about people allowing users to change passwords via the web A but using these checks.  I would like to hear from anyone who has > successfully done this.  Until then, here is what I have been A thinking and I would like to hear the pros and cons against this   idea.   B The (CSWS) web page would allow the user to input their username, B old password, new password and confirmation of the new password.  F Javascript confirms that the new pass and confirmation are the same.    > The web page then submits to a .com file using the OSU script  handler.  > The .com file gets the user device and directory using getuai.  9 The .com file creates a file that contains the following: *     $ define sys$output change_pass.txt        $ set password         oldpass          newpass          newpass      $ deassign sys$output   < The .com file creates a new login.com for the user with the 
 following:
    $ set noon     $ @change_pass.com     $ lo   ( The .com then has as the next statement:    $glogin 'username  : The .com file then looks at change_pass.txt for the result9 of the password change.  If that file is empty, then the  : password change was a success, otherwise return the error 8 message to the web page and allow the user to try again.  + The .com file deletes all files it created.   > I realize that I have to deal with a potential login.com file > that was created with logout as the last command that doesn't ? get deleted and I have somewhat of solution for this.  I write  < to a log file as soon as the .com starts with a record that 8 includes fields for start and finished.  A job will run ; periodically to check the finished field and if it's blank  ; will look for the files that may have been created at that  < time and look for a marker that I will put in files the web   process created and delete them.  A I don't think this is a great solution but it does what I want.    Any input will be appreciated.   Thanks for reading,    John   ------------------------------   Date: 22 Nov 02 08:21:11 +0100) From: p_sture@elias.decus.ch (Paul Sture) : Subject: Clustering, CSWS/Apache, and TCP/IP cluster alias) Message-ID: <z6CwJvl9jymu@elias.decus.ch>   > A colleague is looking at setting up a cluster to run CSWS and> wondering what he can do in terms of failover capability using a TCP/IP cluster alias.   " Any tips, hints, pointers welcome.   TIA.   --  
 Paul Sture Switzerland    ------------------------------  % Date: Fri, 22 Nov 2002 05:53:17 -0800 , From: "James Gessling" <jgessling@yahoo.com>> Subject: Re: Clustering, CSWS/Apache, and TCP/IP cluster alias4 Message-ID: <arlcse$jtf09$1@ID-46415.news.dfncis.de>  L In the simplest form, just have the clients use the alias in the URL.  Like:   http://mycluster.domain.com   I This mostly works fine.  However, if a node goes down that the client was  attachedL to, the client may not be smart enough to do a new dns lookup to find a host that3 is up.  Perhaps an example, from recent experience:   E Client is a cURL program.  When starts up, it connects to alias, this  amounts to client I doing a dns lookup of alias name, getting an IP address and connecting to  that.  Node I in cluster at that IP address goes down.  Stupid client continues to send 
 requests (and I fail) to that IP address.  After operations realizes what's going on, the  client is restarted.K This results in dns lookup of alias name which now returns an IP address of  a node that " is up.  Client runs happily again.  G So depending on your client this can work well or not.  I wonder what a  regular browser would ! do, probably have to restart too.    Jim   ' p.s. find cURL at: http://curl.haxx.se/   6 "Paul Sture" <p_sture@elias.decus.ch> wrote in message# news:z6CwJvl9jymu@elias.decus.ch... @ > A colleague is looking at setting up a cluster to run CSWS and@ > wondering what he can do in terms of failover capability using > a TCP/IP cluster alias.  > $ > Any tips, hints, pointers welcome. >  > TIA. >  > -- > Paul Sture
 > Switzerland    ------------------------------  + Date: Fri, 22 Nov 2002 14:58:18 +0100 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> > Subject: Re: Clustering, CSWS/Apache, and TCP/IP cluster alias; Message-ID: <01KP67XB0XWKA24IEW@sysdev.deutsche-boerse.com>   H > In the simplest form, just have the clients use the alias in the URL.  > Like:  >  > http://mycluster.domain.com  > J > This results in dns lookup of alias name which now returns an IP address4 > of a node that is up.  Client runs happily again.   G This is true of a "dynamic DNS style cluster alias".  The "VMS C_ style H cluster alias" presents ONE IP address to the outside world, regardless G of what node is handling it.  Note that this is not load balancing but  	 failover.    ------------------------------  % Date: Fri, 22 Nov 2002 06:44:40 -0800 , From: "James Gessling" <jgessling@yahoo.com>> Subject: Re: Clustering, CSWS/Apache, and TCP/IP cluster alias4 Message-ID: <arlfsp$jdvig$1@ID-46415.news.dfncis.de>  F "Phillip Helbig" <HELBPHI@sysdev.deutsche-boerse.com> wrote in message5 news:01KP67XB0XWKA24IEW@sysdev.deutsche-boerse.com... I > > In the simplest form, just have the clients use the alias in the URL. 	 > > Like:  > >  > > http://mycluster.domain.com  > > L > > This results in dns lookup of alias name which now returns an IP address5 > > of a node that is up.  Client runs happily again.  > I > This is true of a "dynamic DNS style cluster alias".  The "VMS C_ style I > cluster alias" presents ONE IP address to the outside world, regardless H > of what node is handling it.  Note that this is not load balancing but > failover.   C That is an important distinction.  In my case the load balancing is 
 important and J we've modified the client to restart when it gets a failure.  This type of alias K is what Multinet offers (at least as far as I know).  What stack offers the  other  kind of alias?   Jim    ------------------------------  % Date: Fri, 22 Nov 2002 14:49:48 -0000 * From: "Richard Brodie" <R.Brodie@rl.ac.uk>> Subject: Re: Clustering, CSWS/Apache, and TCP/IP cluster alias, Message-ID: <arlg6d$1loo@newton.cc.rl.ac.uk>  7 "James Gessling" <jgessling@yahoo.com> wrote in message . news:arlfsp$jdvig$1@ID-46415.news.dfncis.de...  Y > That is an important distinction.  In my case the load balancing is important and we've d > modified the client to restart when it gets a failure.  This type of alias is what Multinet offersJ > (at least as far as I know).  What stack offers the other kind of alias?  O UCX used to only support failover before it had BIND support. Now it soes both.    ------------------------------  % Date: Fri, 22 Nov 2002 09:04:47 +0100 6 From: =?iso-8859-1?Q?Jean=2DFran=E7ois=20PI=C9RONNE?=  Subject: Re: Database for VMS + Message-ID: <3DDDE51F.B4B6E6E4@laposte.net>    >  > Lyndon Bartels wrote:  > > 
 > > Hello, > > J > > I know this has been brought up before, but I didn't follow the thread > > at the time. > > ? > > What are the options/opinions of a "free" database for VMS?  > > J > > I went out and looked as MySQL... But there's isn't a VMS port for it.1 > > I downloaded it anyway... Maybe I'll port it.  > J > I noticed several messages in the group mailing.database.mysql-internalsF > from a french guy who is in the process of porting MySql to OpenVMS.F > The last one was about a week ago (november 14). If you want to knowF > how he's doing, or want to give him a hand, you can search in GoogleE > for last weeks posts on "OpenVMS", "MySql" and "port", and you will  > find his E-mail address. >  > --  : You are right, we are currently porting MySQL (V4) to VMS.P We have succesfully compile, link the server and establish an connection between an client and the server.   = Anyone who want to work on this project with us is welcome...   
 Jean-Franois    ------------------------------   Date: 22 Nov 2002 09:55:22 GMT( From: per@nospam.mimer.se (Per Schrder) Subject: Re: Database for VMS 8 Message-ID: <Xns92CE705178AD9pescatmimer@195.58.103.121>  G "Craig A. Berry" <craigberry@nospam.telocity.com> wrote in <craigberry- 0 4B2E31.20010421112002@news.directvinternet.com>:  . >In article <3DDD330B.7268AD2@pressenter.com>,1 > Lyndon Bartels <lbartels@pressenter.com> wrote:  > ? >> What are the options/opinions of a "free" database for VMS?   > 3 >I think Mimer gives you the first copy free.  See:  >  ><http://www.mimer.com>  >    (I develop Mimer on openVMS)  I The download package for the Mimer SQL engine contains a 10-user license   for development and testing.  ( You can read about the VMS version here:3 http://developer.mimer.com/platforms/platform_9.htm   & Version 9.1.3 is due for release soon.  L Please mail me if you have any specific questions or need help setting up a  system.    --  
 /Per Schrder  http://developer.mimer.com   ------------------------------  % Date: Fri, 22 Nov 2002 11:59:35 -0500 ; From: "Brian Tillman" <tillman_brian@notnoone.notnohow.com> + Subject: Re: DCPS V2.2 field test available $ Message-ID: <3dde62a7$1@news.si.com>  C >The major features of this release are new printer support and the 1 >addition of autostart capability to DCPS queues.   K We've been using autostart on our DCPS queues for a while now.  It seems to ! work.  Why is this a new feature?  --  A Brian Tillman                   Internet: tillman_brian at si.com A Smiths Aerospace                          tillman at swdev.si.com = 3290 Patterson Ave. SE, MS      Addresses modified to prevent < Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@"8        This opinion doesn't represent that of my company   ------------------------------  # Date: Fri, 22 Nov 2002 17:15:05 GMT . From: peter@langstoeger.at (Peter LANGSTOEGER)+ Subject: Re: DCPS V2.2 field test available 2 Message-ID: <tCtD9.29140$up.325233@news.chello.at>  b In article <3dde62a7$1@news.si.com>, "Brian Tillman" <tillman_brian@notnoone.notnohow.com> writes:D >>The major features of this release are new printer support and the2 >>addition of autostart capability to DCPS queues. > L >We've been using autostart on our DCPS queues for a while now.  It seems to" >work.  Why is this a new feature?  6 I did use it also for some years now without problems.4 I understand it as it was not supported until now...   --   Peter "EPLAN" LANGSTOEGER % Network and OpenVMS system specialist  E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------  # Date: Fri, 22 Nov 2002 17:28:17 GMT * From: Paul Anderson <paul.anderson@hp.com>+ Subject: Re: DCPS V2.2 field test available 5 Message-ID: <221120021226498080%paul.anderson@hp.com>   2 In article <3dde62a7$1@news.si.com>, Brian Tillman, <tillman_brian@notnoone.notnohow.com> wrote:  M > We've been using autostart on our DCPS queues for a while now.  It seems to # > work.  Why is this a new feature?   @ Autostart was never officially supported, nor was there a way to> implement them with DCPS queues without a bit of muddling withG DCPS$STARTUP.  Now you can specify autostart nodes in the P2 parameter.   F We also added a faster way of running DCPS$STARTUP that eliminates allC INITIALIZE /QUEUE commands.  It's useful for those using autostart.    Paul   --    Paul Anderson   OpenVMS Engineering    Hewlett-Packard Company    ------------------------------  % Date: Fri, 22 Nov 2002 10:52:29 -0000 * From: "Richard Brodie" <R.Brodie@rl.ac.uk>C Subject: Re: DECW$XFS (X font service) - anyone successfully using? + Message-ID: <arl29e$k18@newton.cc.rl.ac.uk>   < "Mark Daniel" <Mark.Daniel@wasd.vsm.com.au> wrote in message- news:arh27q$k7t$1@fang.dsto.defence.gov.au...   6 > Is the DECW font server daemon congenitally useless?  \ I gave up in the end, although partly because I needed to do some magic with scalable fonts.Y That said, later versions were markedly improved over earlier ones. XFS counts as part of T base VMS, so it will probably be updated in GRAPHICS or UPDATE; definitely check you0 are on the latest versions before anything else.   ------------------------------  % Date: Fri, 22 Nov 2002 12:55:52 -0500 5 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com> C Subject: Re: DECW$XFS (X font service) - anyone successfully using? . Message-ID: <3dde70a9$1_3@hpb10302.boi.hp.com>   I'd file a bug report.    Mark Daniel wrote in message ...L >I've been asked to look into why font services from our VMS systems seem to@ >provide intermittent service at best.  Primary use is to supplyK >DECW-specific fonts to some X-terms and PC X display servers.  We start it H >during system DECW startup and the process DECW$XFS is present where itJ >should be.  There is nothing in the SYS$MANAGER:DECW$FS-ERRORS.LOG.  If ID >telnet to port 7100 sometimes I can type a few characters before itL >disconnects (no output) sometimes it disconnects immediately.  I have triedJ >various configuration file items based in the XFS man page (some of which# >cause the DECW$XFS.EXE to ACCVIO).  > 5 >Is the DECW font server daemon congenitally useless?  > J >If not, is there some technique or tool I can use to diagnose this issue? >  >AATIA.  >    ------------------------------    Date: 22 Nov 2002 06:35:22 -0600- From: koehler@encompasserve.org (Bob Koehler) < Subject: Re: DELL Itanium Servers - May be we can run OVMS ?3 Message-ID: <gF6PU5bjGgpl@eisner.encompasserve.org>   c In article <vFWIIg5XjxrA@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes:   @ > Intel sets those rules.  All vendors should be following them.  D    I have never seen that happen in the PC market.  It's been more aC    "my copy works, clone it on the assembly line" type of business. C    This keeps both prices and reliability down, but the buyers seem &    mostly to be impressed with prices.   ------------------------------  % Date: Fri, 22 Nov 2002 07:31:52 -0700 6 From: "Michael D. Ober" <obermd.@.alum.mit.edu.nospam>< Subject: Re: DELL Itanium Servers - May be we can run OVMS ?/ Message-ID: <sdrD9.2$b14.13951@news.uswest.net>   = "JF Mezei" <jfmezei.spamnot@vl.videotron.ca> wrote in message ) news:3DDD40C2.D5E8F44D@vl.videotron.ca...  > Rob Young wrote:G > >         It is indeed getting harder to FUD up IA64 with Dell lining F > >         up and all that.  Because... we know it means the hardware6 > >         will be very reasonable in cost, don't we? > E > Perhaps it is just me, but I don't see Dell's IA64 program as truly 
 genuine. II > have a feeling that Intel may have grabbed Mikey's balls and told it to  start H > producing IA64 stuff otherwise the price of the 8086 chips might rise. > L > There really is no business or technocal reason for Windows to go to IA64,I > especially since the amount of software on windows-IA64 will be a small  subset# > of what is available on the 8086.   A So?  This was the same argument against the 80386 on the desktop. I Personally, I'd love to see VMS ported to a generic Itanium 2 system, and E baring that, WindowsNT being brought up to the stability and security J standards of VMS.  I'm not holding my breadth for either, though because IE don't trust VMS marketing and I don't think MS will be able to follow + through on their secure computing platform.   
 Mike Ober.   ------------------------------    Date: 22 Nov 2002 08:59:15 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) < Subject: Re: DELL Itanium Servers - May be we can run OVMS ?3 Message-ID: <K72b9alFI7z2@eisner.encompasserve.org>   c In article <gF6PU5bjGgpl@eisner.encompasserve.org>, koehler@encompasserve.org (Bob Koehler) writes: e > In article <vFWIIg5XjxrA@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes:  > A >> Intel sets those rules.  All vendors should be following them.  > F >    I have never seen that happen in the PC market.  It's been more aE >    "my copy works, clone it on the assembly line" type of business.   F I presume that is why Intel has been a bit more firm this time around.> _They_ know about the reliability problems that have resulted.   ------------------------------    Date: 22 Nov 2002 09:07:13 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) < Subject: Re: DELL Itanium Servers - May be we can run OVMS ?3 Message-ID: <yCEAN50gW4tI@eisner.encompasserve.org>   h In article <sdrD9.2$b14.13951@news.uswest.net>, "Michael D. Ober" <obermd.@.alum.mit.edu.nospam> writes:  K > Personally, I'd love to see VMS ported to a generic Itanium 2 system, and G > baring that, WindowsNT being brought up to the stability and security L > standards of VMS.  I'm not holding my breadth for either, though because IG > don't trust VMS marketing and I don't think MS will be able to follow - > through on their secure computing platform.   B My faith was reinforced by the (non-disclosure) discussions at theF just-completed VMS Symposium, but of course that event did not attractC those who are already discouraged from spending their budget on VMS  "stuff".  I As an example (without violating non-disclosure) there was a presentation C for which I had as many as 10 "what-if" angles not addressed by the A manager-type who gave the presentation.  Talking afterward to the @ technical fellow in the back of the room who implements what the@ manager was talking about, I found out they were on their way to0 good (my opinion) solutions to all of my issues.  D No, the technical fellow should not have given that particular talk,E because it had to come from management, and no the details should not D have been in the presentation because I was the only one in the roomD who cared at that level of detail.  Certainly there were other talksE which led with the technical side rather than the business side.  (As E there were other topics for which I was incompetent to understand the 2 technical details -- we all have our specialties.)   ------------------------------  % Date: Fri, 22 Nov 2002 12:39:25 -0500 5 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com> < Subject: Re: DELL Itanium Servers - May be we can run OVMS ?. Message-ID: <3dde6cce$1_3@hpb10302.boi.hp.com>  D Shane Smith wrote in message <01C29143.AD668A50@sulfer.icius.com>...F >Did you just say "firmware and hardware rules"? Are you implying thatI >the VMS port to Itanium will run on any Itanic based system that follows I >Intel's guidelines regardless of manufacturer? If that's what you meant,  >I'm interested and encouraged.  >   E Nothing we are doing is tied to something HP-specific.  We are moving L forward using the interfaces (like ACPI) which in theory will allow us to beH platform specific.  What that does is shift "correctness" to making sureL that the interfaces on the platform conform to specs - like ACPI, and DIG64.E We will probably recognize some HP-specific attributes to allow us to H support HW capabilities that aren't available on other platforms, but weI "should" be able to genericly support any Itanium-2 (and forward) system.   J [Note: There *may* be some issues short term.  For instance, Alpha driversH often assume that that map registers are available.  The Intel referenceH platform, for example, doesn't have map registers.  So 32-bit devices onI systems with more than 4Gb are problematic - as are drivers that *always* J use map registers.  At minimum, those drivers need to change, and that mayJ take a while to happen -- since the HP platforms *do* have map registers -1 it isn't a high priority for the early releases.]n  G >Call me cynical (because I am, and I know it) but I'd sort of expectedlB >there to be some widget or other that tied VMS to HP's own Itanic2 >systems, probably in the form of custom firmware. >b  K Nope.  BUT I said *nothing* about licensing.  Our focus is on supporting HPvI hardware.  I don't know that anyone has given any thought to how or if we ; would *support/license* VMS on non-HP (like Dell) hardware.A   >Shane >P >-----Original Message-----e; >From: Fred Kleinsorge [mailto:kleinsorge@star.zko.dec.com] * >Sent: Thursday, November 21, 2002 9:44 AM >To: Info-VAX@Mvb.Saic.Com= >Subject: Re: DELL Itanium Servers - May be we can run OVMS ?i >  >t >a% >Larry Kilgallen wrote in message ...b= >>In article <3DDC50D6.3254BEC9@fsi.net>, "David J. Dachtera"   ><djesys.nospam@fsi.net> writes: >>> Fabio Cardoso wrote: >>>>
 >>>> Click >>>>. >>>> http://news.com.com/2100-1040-966499.html >>>sI >>> Unless the folks at Dell have been working with the folks at OpenVMS,O >>> don't count on it. >>9 >>I would not count it out as being technically possible.- >>F >>Certainly HP would not provide technical support, but it is possible; >>you might be able to buy a license (and not get support).e >eI >As he said, I think the only question is one of licensing, I don't thinkoB >there is a technical issue, as long as they are following all the >firmware &  >hardware rules. >r >t >D >T   ------------------------------  % Date: Fri, 22 Nov 2002 12:48:00 -0500b5 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com>n< Subject: Re: DELL Itanium Servers - May be we can run OVMS ?, Message-ID: <3dde6ed1_3@hpb10302.boi.hp.com>   Dan Allen wrote in message ...  B >Perhaps the question is whether HP is in the hardware or softwareJ >business.  Since they manufacture HP-UX and VMS but not Itanium they lookK >like a software company to me. If software is the only part of the packageGK >you actually manufacture why should you care greatly where the rest of theWK >package comes from (aside from the support issue) as long as you sell your  >product  H The CPU is the part that we will buy and not build.  We are building theJ value added portion of the system.  Adding performance, reliability, cost,K and scaleability.  The core logic chipset is an example.  System managementaF HW.  Cellular technology for large systems.  Etc.  Software is also anD example of the added value.  The combination of our software, on ourL hardware, with our field support - is how we build enterprise class systems.; This isn't the Dell model.  This isn't the Microsoft model.-   ------------------------------  % Date: Fri, 22 Nov 2002 12:53:56 -0500,5 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com>.< Subject: Re: DELL Itanium Servers - May be we can run OVMS ?. Message-ID: <3dde7034$1_2@hpb10302.boi.hp.com>  A David J. Dachtera wrote in message <3DDD8CE5.1040810A@fsi.net>...R >Fred Kleinsorge wrote:k >>' >> Larry Kilgallen wrote in message ...e? >> >In article <3DDC50D6.3254BEC9@fsi.net>, "David J. Dachtera"n" >> <djesys.nospam@fsi.net> writes: >> >> Fabio Cardoso wrote: >> >>> >> >>> Click >> >>>0 >> >>> http://news.com.com/2100-1040-966499.html >> >>rK >> >> Unless the folks at Dell have been working with the folks at OpenVMS,  >> >> don't count on it. >> >; >> >I would not count it out as being technically possible.m >> >H >> >Certainly HP would not provide technical support, but it is possible= >> >you might be able to buy a license (and not get support).3 >>K >> As he said, I think the only question is one of licensing, I don't thinkVD >> there is a technical issue, as long as they are following all the
 firmware & >> hardware rules. >cF >...and who set those rules? If it wasn't OpenVMS, forget it - I'm not( >betting on anything until I see it run. >n  H Intel/Microsoft for the most part.  Some (like DIG64) are consortiums ofH large companies.  You'll note however, that with the exception of Alpha,I there were very few HAL's needed for NT on Intel platforms.  So they *do*i know how to do it.  F >Other o.s.-es have a supported hardware list. Solaris/Intel is a goodE >parallel here. I would expect OpenVMS-IPF to likewise have a list of8B >supported chipsets, etc. outside of which I would expect marginal >success, if any.3 >0  E I have no idea.  Nor do I believe does anyone else right now.  We are1G focused on our short term deliverables, and frankly this is just way tos early for us to worry about.   ------------------------------  + Date: Fri, 22 Nov 2002 11:54:56 +0000 (UTC) + From: david20@alpha1.mdx.ac.uk (David Webb) $ Subject: Re: Get owner info from UAF+ Message-ID: <arl5ug$lk8$1@aquila.mdx.ac.uk>1  a In article <gPAk7RdoRvfq@eisner.encompasserve.org>, young_r@encompasserve.org (Rob Young) writes:0 >ed >> In article <819f1cc4.0211210955.c6a6de3@posting.google.com>, daveparboo@hotmail.com (DXP) writes:7 >>> $ owner = f$edit(f$extract(85, 31, record), "trim")n >> ...5 >>> Well, if p1 .eqs. plebby_user then no problem....tG >>> However, if p1 .eqs. super_user_with_loads_of_privs_and_identifiers E >>> then I get the message, "%DCL-W-BUFOVF, command buffer overflow -b' >>> shorten expression or command line"S >>> G >>> Now, I am aware of the 1024k symbol buffer size limit, but how do I . >>> get the info from this record that I need. >n >nI >	One of the best UAF extracting programs is getting harder to find.  Joe)B >	Meadows , (in a much earlier DECUS submission I suppose) has two? >	programs (in addition to Verb) that are in the same bundle.  m9 >	GUESS_PASSWORD and UAF.  I just went and found it here:m > $ >http://www.uniud.it/ftp/vms/uaf.zip >t= >	Maybe there is another place... real tricky to find though.  >a5 >	To do what you are after looks like this using UAF:  >o= >	$ uaf /sele=(user=username_goes_here)/display=owner/nolabela > ! >	Or numerous variations on that.r >s: >	Then play with PIPE to force it into a symbol to make it5 >	usuable.  Many examples of that about.  That or you:% >	can do it the old fashioned way by:  > K >  $ uaf /select=(user=username_goes_here)/display=owner/nolabel/output=a.a  >e >	And then reading from a.a	 >  >				Rob >dI >"An old Oregon rancher once told me, there are three types of men in thetM >world.  One type learns from books.  One type learns from observations.  Andk= >one type just has to urinate on the electric fence himself."a >h= >                                        -- Carl Barney, 1996    Better yet get getuai     O This allows you to access individual fields in the uaf records and return them r as symbols 7   eg  # Alpha2:getuai "DUMMY"/exists=existse Alpha2:sh sym exists   EXISTS = "YES" Alpha2:sh sym owner.= %DCL-W-UNDSYM, undefined symbol - check validity and spellingg! Alpha2:getuai "DUMMY"/owner=owneru Alpha2:sh sym ownern   OWNER = "dummy test account"     getuai is available from  0 http://vms.process.com/ftp/vms-freeware/FILSERV/    
 David Webb VMS and Unix team leader CCSS Middlesex University   ------------------------------  % Date: Fri, 22 Nov 2002 10:02:47 -0500o& From: David M Smith <dsmit115@csc.com>$ Subject: Re: Get owner info from UAF8 Message-ID: <llhstusahfob6lks3dsd06e4u5c32p9585@4ax.com>  K On 21 Nov 2002 21:03:13 -0600, young_r@encompasserve.org (Rob Young) wrote:a  > >	While GETUAI I suppose has its feature, the two really don't- >	compare feature-for-feature, in my opinion.c >i >				Rob  M Rob, I agree -- the two are aimed at different uses. I have both on my system0I (or equivalents). GETUAI is used within command procedures to to what thelO non-existent lexical function F$GETUAI would do, while UAF is used for "ad hoc". queries and reporting.  D So, I recommend system managers have both tools available for use...I ------------------------------------------------------------------------- I David M. Smith 302.391.8533                       dsmit115 at csc dot com I Computer Sciences Corporation     (Opinions are those of the writer only) I -------------------------------------------------------------------------C   ------------------------------  % Date: Fri, 22 Nov 2002 10:12:08 -0500 & From: David M Smith <dsmit115@csc.com>$ Subject: Re: Get owner info from UAF8 Message-ID: <q4istucdhbeonkd0m2thm34235sat8hpvl@4ax.com>  M On 21 Nov 2002 15:14:13 -0600, koehler@encompasserve.org (Bob Koehler) wrote:,  G >   According to FDL the key at index 0 is length 32.  That makes sensewE >   because you probably should be looking by username.  So why wouldFI >   p1 be "super_user_with_loads_of_privs_and_identifiers"?  Is it reallyp >   record that's too long?L  P I've run into SYSUAF.DAT files with "long records" which could not be handled in# DCL. If you look at the FDL you seeE   RECORD#         BLOCK_SPAN              yes=$         CARRIAGE_CONTROL        none(         FORMAT                  variable$         SIZE                    1412  M and if you look at $UAFDEF in SYS$LIBRARY:LIB.REQ (if you have it loaded) you-K see that the UAF records have a fixed area which is 644 bytes long, and theXJ remainder is "available" for "user data" (IIRC, I don't know where this is documented).  L The number of privileges and identifiers does not change the record size, asI every record contains the same privilege bits (either 1 or 0) and grantedX4 identifiers are "stored" in the RIGHTSLIST.DAT file.I -------------------------------------------------------------------------rI David M. Smith 302.391.8533                       dsmit115 at csc dot comrI Computer Sciences Corporation     (Opinions are those of the writer only)cI -------------------------------------------------------------------------    ------------------------------    Date: 22 Nov 2002 08:34:23 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen)  Subject: Re: Gnat Ada for VMS73 Message-ID: <A02PFbweYY5h@eisner.encompasserve.org>0  M In article <3ddcf20f$1@news.post.ch>, "Jakob Erber" <erberj@yahoo.de> writes:(M >> I presume you have informed ACT.  Please be sure to inform VMS Development  >> at HP as well.  >  > Which way should I go?  I I had occasion last week to recommend that someone considering conversionNH from VAX to Alpha use DEC Ada.  There were HP folk in the room, and theyG did not object to my suggestion, but of course what I said was tailoredL: to the particular folks in the discussion and their needs.  A For Itanium HP has said there will be no DEC Ada, but by the time D folks who are just now switching to Alpha get around to switching toD Itanium, there should be lots of improvements in GNAT (or maybe some1 giant customer will convince HP to port DEC Ada).'   ------------------------------  % Date: Fri, 22 Nov 2002 10:49:21 -0500e: From: Charlie McCutcheon <charlie.mccutcheonNO@SPAMhp.com> Subject: Re: Gnat Ada for VMS * Message-ID: <3DDE5201.C0127947@SPAMhp.com>  & --------------BC8F8108F8EBC8942FC3EECC* Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit    Jakob Erber wrote:   > Hello, >,H > when we complained, about Compaq would stopping the support for CompaqL > Ada83, Compaq replied with full right, why we could not use Gnat Ada as an > alternative.L > Now, I'm able to start a list of weak points of this compiler, gathered in > our own porting process: >eO > 1) The Gnat Ada compiler is incredible slow in comaprison to the Compaq Ada83c > compiler.tK >      When it took about 2 and a half hours to build our whole system withr > Ada83, we can be3 >      glad if we manage the same in 24h with Gnat!uN > 2) We make use of very complex variant rekord defintions. Where Compaq Ada83
 > compilesM >      those in a view seconds, the Gnat compiler needs incredible amounts ofs > memory and timeeK >      to perform the same task. In our last attempt, we configured 1 GB ofu > virtual memory forO >      the compiler to use. But it still run out of memory. We will not we able. > to avoid codeo/ >      modifications to make it pass this code.d > H > These are the points till now. I let you know, when more are occuring. >t > best regards >c
 > Jakob Erbert  8 Compaq Ada is still supported on  OpenVMS VAX and Alpha.  & We do respond to user problem reports.  P ACT's GNAT Pro is available on OpenVMS Alpha (and Tru64), and is being ported to Itanium.  K Please note that Compaq/HP funded some performance improvements to GNAT forDJ OpenVMS Alpha.  I believe the changes are mostly runtime improvements, not0 compile time.  Contact ACT directly for details.  P ACT is typically interested in user problems.  Make sure you report the problems you see to GNAT.   Charlie McCutcheon Compaq Ada project leader     & --------------BC8F8108F8EBC8942FC3EECC) Content-Type: text/html; charset=us-asciiu Content-Transfer-Encoding: 7bitg  > <!doctype html public "-//w3c//dtd html 4.0 transitional//en"> <html> Jakob Erber wrote: <blockquote TYPE=CITE>Hello,I <p>when we complained, about Compaq would stopping the support for CompaqeH <br>Ada83, Compaq replied with full right, why we could not use Gnat Ada as anr <br>alternative.K <br>Now, I'm able to start a list of weak points of this compiler, gatheredn in <br>our own porting process:J <p>1) The Gnat Ada compiler is incredible slow in comaprison to the Compaq Ada83 compiler. K <br>&nbsp;&nbsp;&nbsp;&nbsp; When it took about 2 and a half hours to buildh& our whole system with Ada83, we can beI <br>&nbsp;&nbsp;&nbsp;&nbsp; glad if we manage the same in 24h with Gnat!aJ <br>2) We make use of very complex variant rekord defintions. Where Compaq Ada83 compilesG <br>&nbsp;&nbsp;&nbsp;&nbsp; those in a view seconds, the Gnat compilert+ needs incredible amounts of memory and time K <br>&nbsp;&nbsp;&nbsp;&nbsp; to perform the same task. In our last attempt, ( we configured 1 GB of virtual memory forF <br>&nbsp;&nbsp;&nbsp;&nbsp; the compiler to use. But it still run out, of memory. We will not we able to avoid codeE <br>&nbsp;&nbsp;&nbsp;&nbsp; modifications to make it pass this code. I <p>These are the points till now. I let you know, when more are occuring.i <p>best regardsr <p>Jakob Erber</blockquote>sD <b>Compaq Ada is still supported on&nbsp; OpenVMS VAX and Alpha.</b>) <p>We do respond to user problem reports.hI <p>ACT's GNAT Pro is available on OpenVMS Alpha (and Tru64), and is beingh ported to Itanium.J <p>Please note that Compaq/HP funded some performance improvements to GNATO for OpenVMS Alpha.&nbsp; I believe the changes are mostly runtime improvements,e9 not compile time.&nbsp; Contact ACT directly for details.aK <p>ACT is typically interested in user problems.&nbsp; Make sure you reportc the problems you see to GNAT.e <p>Charlie McCutcheonf <br>Compaq Ada project leader  <br>&nbsp;</html>o  ( --------------BC8F8108F8EBC8942FC3EECC--   ------------------------------  % Date: Fri, 22 Nov 2002 10:41:57 -0500c: From: Charlie McCutcheon <charlie.mccutcheonNO@SPAMhp.com> Subject: Re: Gnat Ada for VMSd* Message-ID: <3DDE5045.6614CACD@SPAMhp.com>   Jakob Erber wrote:  N > > I presume you have informed ACT.  Please be sure to inform VMS Development > > at HP as well. >t > Which way should I go? >c > Jakob   M Compaq Ada product manager is Pat Mulvey; pat.mulveyAThp.com (change AT -> @,  trying to avoid spam for her).   ------------------------------  % Date: Fri, 22 Nov 2002 10:22:31 +0100u; From: "Thomas Egenberger" <thomas.egenberger@technidata.de>o: Subject: Re: How to delete a command from the DCL-Tables ?& Message-ID: <arl088$en$1@news.dtag.de>  A "Peter LANGSTOEGER" <peter@langstoeger.at> schrieb im Newsbeitragp, news:WZdD9.16821$up.193857@news.chello.at... > >e= > >How can I delete the old java command from the DCL Table ?0 >0 > $ SET0E COM/TA=SYS$COMMON:[SYSLIB]DCLTABLES/OU=SYS$COMMON:[SYSLIB]DCLTABLES -r > /DEL=JAVAc+ > $ INSTALL RE SYS$COMMON:[SYSLIB]DCLTABLESl+ > $ SET COM/TA=SYS$COMMON:[SYSLIB]DCLTABLESe  % Thanks. That's the way it has to work    > >And maybe? > >How can I make the new command to be part of the DCL-Table ?  >eI > There is no .CLD file in newer JAVA versions. Do you want to build youre own ?n   Thats right. Yes. 3 There is only a .CLD file for the old Java Version.h. Has anyone a .CLD File for VM (build 1.3.1-1)?% (I know, it's not the newest version)i   >nE > >Work around is in the moment to define a symbol to the new versiona > >to override the old command.n >i8 > That's no workaround. That's the supported method now. >.3 > $ JAVA == "$ sys$common:[java$131.bin]java$java "a > ? > Note the space at the end. This (and umpteen other) symbol(s)hH > (and logicals) is (are) defined in JAVA$131_SETUP.COM which you shouldI > invoke in you LOGIN.COM (and JAVA$STARTUP.COM in you SYSTARTUP_VMS.COM)i   That is the way I did it.i  7 What's about the space at the end? Any special effects?r   >  > -- > Peter "EPLAN" LANGSTOEGERt' > Network and OpenVMS system specialist: > E-mail  peter@langstoeger.atH > A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist >.   Thanks   ------------------------------  % Date: Fri, 22 Nov 2002 15:50:29 -0000t! From: Z  <zarlenga@conan.ids.net>a6 Subject: Re: I give up, need Solaris for VMS jobs now./ Message-ID: <utski55scnb594@corp.supernews.com>   / Fabio Cardoso <fabiopenvms@yahoo.com.br> wrote:-	 : Well !  P : A database is a database .... why not migrate from Sybase to Oracle or RDB ???   A database is NOT a database!:  > They were tied to Sybase's Open Server product and when you're> processing 100k financial transctions per minute, you need to ( exploit the database's special fetaures.  ? With more than 500 Sybase-specific stored procedures to rewrite > for Oracle, and then to certify, it was easier to move to AIX.   ------------------------------  # Date: Fri, 22 Nov 2002 09:43:36 GMTh. From: peter@langstoeger.at (Peter LANGSTOEGER)Y Subject: Re: Is there a hobbyist program for other HP products? Was Re: What's going on wl2 Message-ID: <c%mD9.22088$up.245673@news.chello.at>  V In article <3DDC2BD0.7A50CE14@swbell.net>, Patrick Jankowiak <eccm@swbell.net> writes:4 >The hobbyist program for VMS is still going strong.  1 Yes, the license request process does still work.e- I downloaded my last licenses on 12-Sep-2002.$  F But nevertheless it could be better. VMS kits are not downloadable andD even not orderable as a CD. Maybe there are still some old CDs (withJ VMS V7.2 and so on) remaining, but that would I not call "going strong"...  L >I do have an "OpenVMS" live free or die license plate, but not for sale 8-P  6 Me too. That's what DECUS Symposia are for, isn't it ?   -- P Peter "EPLAN" LANGSTOEGERt% Network and OpenVMS system specialist  E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------  % Date: Thu, 21 Nov 2002 23:58:50 -0500i( From: David Froble <davef@tsoft-inc.com>% Subject: Re: Jaw dropping EV7 systemsr, Message-ID: <3DDDB98A.8060000@tsoft-inc.com>   Fred Kleinsorge wrote:  ! > John Smith wrote in message ...e > B >>"Fred Kleinsorge" <kleinsorge@star.zko.dec.com> wrote in message/ >>news:zU8C9.26$w34.1038486@news.cpqcorp.net...  >>J >>>Remember that there is more than just a new CPU chip.  It's an entirely >>>  >>new: >>D >>>system platform from the ground up.  As in *any* new major system >>>n >>platform,e >>I >>>things always appear to stretch out at the end while the last few nitsm >>>s >>(and >>J >>>not just HW, but SW) are cleaned up as real customer workloads start to >>>. >>hith >>L >>>them in field test.  The Marvel platform is unreal in how fast it is, andJ >>>how well it scales - even the 800Mhz proto's were unbelievable.  As far >>>s > as >  >>Ie >>J >>>can detect, there is no feet dragging to get it officially shipped, but >>>  >>care >>J >>>is being taken to make sure that it is 100% rock solid when it hits the
 >>>street. >>>- >>I >>So Fred, how much advertising and marketing of Marvel do you think will= >> > take > @ >>place at Sun and IBM and existing HP Superdome customer sites? >> >>Just curious.t >> >> > N > I can't imagine us wanting a happy HP-UX customer to port to VMS.  As to theK > rest, it's hard to change the path of a customer committed to another OS.aE > But if they are OS agnostic, I expect that Marvel will sell itself.t    & Not if they are not exposed to Marvel.    p You aren't suppost to make it hard for a potential customer to know about your product.  It will need marketing.    , What about the not-so-happy HP-UX customers?   Dave   ------------------------------  # Date: Fri, 22 Nov 2002 16:41:48 GMTt. From: peter@langstoeger.at (Peter LANGSTOEGER)% Subject: Re: Jaw dropping EV7 systems 2 Message-ID: <g7tD9.28389$up.304661@news.chello.at>  f In article <3ddd1bbf$1_3@hpb10302.boi.hp.com>, "Fred Kleinsorge" <kleinsorge@star.zko.dec.com> writes:B >I can't imagine us wanting a happy HP-UX customer to port to VMS.  D Wanting a customer to port and throwing stones at customers who wantJ to port are different things. We only want HP to 'allow' customers to portE to VMS (by removing the dead-row-image of it) or even to know of VMS.v  @ And, why do you think HPsUX customers are only happy customers ?> Because the unhappy already migrated to SLOWLARIS/AIX/Linux ??  J HP should even consider helping customers migrate from HP-UX to OpenVMS !!G Migrating (is fresh blood to the customer) to a HP platform (is bad for H the competition) is something HP should always support. And what I heardH is that Alpha and mainly OpenVMS brings much more margin for HP than the* the rest of the stuff which HP also sells.  L >                                                                  As to theJ >rest, it's hard to change the path of a customer committed to another OS.  E Yes, it is hard (not only because of such decisions are based on manyeH years of experiences and market observations). And DEC/Q/HP made it evenJ harder by forcing a bad image of Alpha and/or VMS. Nobody wants to migrate  to a "soon to be dead" platform.C It is hopefully wrong, but this image persists unfortunately. Sigh.c  D >But if they are OS agnostic, I expect that Marvel will sell itself.  G Only to those who know of the machine. And only to those how have stillcI an interest using a opsys which supports this hardware (and neither Tru64>0 nor OpenVMS is any longer of _common_ interest).  G Fred, don't be silly. don't support the moving to some remaining niches F HP does with OpenVMS. Porting to IA64 (paid by INTEL) is _not_ enough.I What will _you_ do in two years when the port is finished and 'no further / development' pops up as a writing on the wall ?A  J And "will sell itself" was already a nail for DEC's coffin. Do you remeberJ the story where DEC sales persons got their telephones limited to internalK calls only [to save money] because "customers will call us, no need to call ; them" ?? It is such a ridicolous idea, but it did happen...    -- 2 Peter "EPLAN" LANGSTOEGER@% Network and OpenVMS system specialistj E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------  % Date: Fri, 22 Nov 2002 12:24:47 -0500o5 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com>a% Subject: Re: Jaw dropping EV7 systemsi, Message-ID: <3dde6961_1@hpb10302.boi.hp.com>  ? JF Mezei wrote in message <3DDD3F3D.619AA8F@vl.videotron.ca>...I >Fred Kleinsorge wrote:sD >> I can't imagine us wanting a happy HP-UX customer to port to VMS. >lJ >Prior to HP buying Digital, would you have said the same thing ? How many of* >your competitor OS do you say this with ? >i  G HP bought CPQ.  The answer is pretty obvious.  We want all users of our'G competetors systems to switch to HP systems.  On the other hand, unlesseJ there is a specific reason - we aren't going to try and churn our own base
 to switch.  A >Is Solaris the only OS you can imagine stealing customers from ?a >aL >VMS has technical features that make it unique and which are not present inE >HP-UX. In the past, HP would work hard to prevent its customers fromo	 defectingqJ >to Digital. But now that VMS is part of HP, there should be ABSOLUTELY NOA >REASON for HP to prevent a HP-UX customer from switching to VMS.  >lE >Would GM prevent some customers from switching from a chevrolet to aP
 cadillac ?  H What's good for the goose, is good for the gander.  If you don't want HPK coaxing VMS users to become HP-UX users - then you shouldn't complain if weI also don't do the opposite.t  J If you are a happy VMS customer - stay with VMS.  If you are a happy HP-UXJ customer - stay with HP-UX.  If in either case you are looking to move offE of your current OS - then HP-UX customers should look at VMS, and VMS:K customers should look at HP-UX.  And a Solaris customer should look at both$ *as quickly as possible* ;-)   ------------------------------  % Date: Fri, 22 Nov 2002 00:10:17 -0500$( From: David Froble <davef@tsoft-inc.com>$ Subject: Re: Misuse of SYSPRV in VMS, Message-ID: <3DDDBC39.6070004@tsoft-inc.com>   JF Mezei wrote:d   > David Webb wrote:s > L >>The fact that you need privileges to bind to ports less than 1024 is good.P >>It stops user's setting up a web server on the official port - port 80 on your# >>system without your knowledge etct >> > N > But having to grant a privilege that allows the image to do a whole lot moreK > than just bind to that port is just as dangerous, especially when you arecU > talking about open source/freeware programs that come from other operating systems.D > N > Since they don't know about VMS, they won't have code to turn off SYSPRV andP > turn it on only when absolutely necessary, and with SYSPRV alwasy on, you riskO > that freware application corrupting your system, just like IIS did to WindowsC0 > because it runs with the equivalent of SYSPRV. > N > This is one reason I trust the OSU web server since I know that it turns off+ > its allocated privileges when not needed.n > P > However, if there were a privilege (such as LOG IO or other) which allowed theI > binding of the port without granting the application free access to thet! > system, it would be far better.o  L I have to agree with JF, at least with respect to 'doing it right'. The realN issue is that VMS privs were set up before some of today's activities existed.G When the new activities began, a decision to use existing privs for newe/ capabilities occured. This is the real problem.-    M If the capability of binding to a port at or below 1024 is to be restricted,  O than some type of capability to grant such activity, specifically, should have sO been devised.  Giving a web server what amounts to access to every file on the eM system is a serious security problem.  I'm not sure that it should even be a eM priviledge, but possibly something that could be assigned to specific users, eK programs, whatever.  Not trying to design such here, just say that it's an i% ommision that would be better filled.s   Dave   ------------------------------  % Date: Fri, 22 Nov 2002 08:52:06 +0000 % From: Alan Greig <a.greig@virgin.net>-$ Subject: Re: Misuse of SYSPRV in VMS8 Message-ID: <usrrtus9f1fabg2uif5egvckd62dqhl9fe@4ax.com>  , On Thu, 21 Nov 2002 14:58:52 -0500, JF Mezei( <jfmezei.spamnot@vl.videotron.ca> wrote:   >David Webb wrote:M >> The fact that you need privileges to bind to ports less than 1024 is good.iQ >> It stops user's setting up a web server on the official port - port 80 on your $ >> system without your knowledge etc > M >But having to grant a privilege that allows the image to do a whole lot morenJ >than just bind to that port is just as dangerous, especially when you areT >talking about open source/freeware programs that come from other operating systems.  9 DEC's own TOPS-20 had/has specific privileges relating totF Arpanet/Internet. "Absolute Arpanet Sockets" is one that comes to mind= here. "Net Wizard" was another more exotically named example.n  E While I'm at it I always thought "Wheel" sounded better than "SETPRV"s    M >Since they don't know about VMS, they won't have code to turn off SYSPRV and6O >turn it on only when absolutely necessary, and with SYSPRV alwasy on, you risklN >that freware application corrupting your system, just like IIS did to Windows/ >because it runs with the equivalent of SYSPRV.s >rM >This is one reason I trust the OSU web server since I know that it turns offI* >its allocated privileges when not needed. >SO >However, if there were a privilege (such as LOG IO or other) which allowed theeH >binding of the port without granting the application free access to the  >system, it would be far better.   -- Alan   ------------------------------  % Date: Fri, 22 Nov 2002 08:55:51 +0000 % From: Alan Greig <a.greig@virgin.net> $ Subject: Re: Misuse of SYSPRV in VMS8 Message-ID: <m2srtucs2r2j01p6vthg541kiaglm7ghte@4ax.com>  F On Fri, 22 Nov 2002 00:10:17 -0500, David Froble <davef@tsoft-inc.com> wrote:    M >I have to agree with JF, at least with respect to 'doing it right'. The realbO >issue is that VMS privs were set up before some of today's activities existed.sH >When the new activities began, a decision to use existing privs for new0 >capabilities occured. This is the real problem.  E But in the case of Arpanet/Internet, TOPS-20 Arpanet Monitor and it'sbB KA-10 earlier incarnation Tenex, had privs specifically related to& Arpranet long before VMS was released.  A DEC just tried to ignore non-DECNET (or ANF-10) protocols for too  long.e   >aN >If the capability of binding to a port at or below 1024 is to be restricted, P >than some type of capability to grant such activity, specifically, should have P >been devised.  Giving a web server what amounts to access to every file on the N >system is a serious security problem.  I'm not sure that it should even be a N >priviledge, but possibly something that could be assigned to specific users, L >programs, whatever.  Not trying to design such here, just say that it's an & >ommision that would be better filled. >m >Davey >i   -- Alan   ------------------------------  % Date: Fri, 22 Nov 2002 16:28:27 +0000i' From: Andrew Harrison SUNUK Consultancya# Subject: Re: OT: Hamburger Flipping . Message-ID: <3DDE5B2B.9010209@nospamn.sun.com>   VAXman- wrote: > In article <3DD3991A.90402@nospamn.sun.com>, Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: >  >>{...snip...} >>> >>>...but will I find Andrew Harrison there?  And if so, when? >>>k >> >>Friday lunchtimes. >  >  >  > :( > I > IIRC, I believe I leave the states on Friday eve arriving on a SaturdayoJ > morning.  I did manage to look up a few of the names here (@*.uk) when IJ > was in the UK last Spring -- over Easter and there when Queen Mum passed' > away making things quite interesting.e >    Shame, give it a try anyway.  A When you have finished head to the Tate Modern 15 minutes walk ore: a short taxi drive on the South bank of the Thames west of Borough Market..   Regardsr Andrew Harrisonc   > --Q > VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMi
 >            -7 >   "Well my son, life is like a beanstalk, isn't it?" 9 >    ------------------------------  # Date: Fri, 22 Nov 2002 16:59:52 GMTR" From:   VAXman-  @SendSpamHere.ORG# Subject: Re: OT: Hamburger Flipping$0 Message-ID: <00A175A7.2288928D@SendSpamHere.ORG>   In article <3DDE5B2B.9010209@nospamn.sun.com>, Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: >o >a >VAXman- wrote:a >> In article <3DD3991A.90402@nospamn.sun.com>, Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes:n >> o >>>{...snip...}o >>> ? >>>>...but will I find Andrew Harrison there?  And if so, when?o >>>> >>>  >>>Friday lunchtimes.P >> i >>   >> y >> :(l >> sJ >> IIRC, I believe I leave the states on Friday eve arriving on a SaturdayK >> morning.  I did manage to look up a few of the names here (@*.uk) when IIK >> was in the UK last Spring -- over Easter and there when Queen Mum passed ( >> away making things quite interesting. >> n >h >Shame, give it a try anyway.b >bB >When you have finished head to the Tate Modern 15 minutes walk or; >a short taxi drive on the South bank of the Thames west ofa >Borough Market.   ... and you'll be there?  E I'd like to take in the Tate Modern being an afficionado of the Dada, D Surrealist, and Avant/Modern genres (I own two Dali and one Miro or-C iginals) but the kids will be with me and their NJ public schoolingn7 educated minds are not ready for that much stimulation.w   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMo            e5   "Well my son, life is like a beanstalk, isn't it?"     ------------------------------    Date: 22 Nov 2002 08:41:31 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen)r' Subject: Re: Pathworks Mac vs VMS 7.3-1b3 Message-ID: <PalVh2xqWd7v@eisner.encompasserve.org>   b In article <20021121084735.0279e88f.mathog@caltech.edu>, David Mathog <mathog@caltech.edu> writes:  K > It was inevitable that some version of VMS would break PW/Mac.  Do not be-K > surprised if some release of VMS breaks connectivity to printers as well.o  J If printer connectivity broke, I am confident I could write a replacement.  B But since DOCUMENT/GRAPHICS requires something less that VMS V7.3,? those older versions will always be around.  If clustering is a<= problem, perhaps DQS would handle an AppleTalk printer queue.a   ------------------------------  % Date: Fri, 22 Nov 2002 15:28:03 +0000h' From: Andrew Harrison SUNUK Consultancy . Subject: Re: poor Gigabit Ethernet performance. Message-ID: <3DDE4D03.8010304@nospamn.sun.com>   Rudolf Wingert wrote:o > Hello, >  > Andrew Harrison wrotes:b >  > ? > If HP are saying the 26% is what you should expect regardlesso< > of what tesdt you use then that is low compared with other? > platforms, we tested a Sun 450 a long time ago at 700-800 Mbsn
 > using GigE.h > <<<t > D > Andrew, is there any testtool to test this? We do have four Sun450F > with Gigabit Ethernet and we would like to know how good the networkC > performance is. A tool to scan the network connection, did reportkB > 1000Base line, but a throughpu less then a 10Base line. We wouldB > like to verify this. May be the tool is buggy. On the other handA > user do report sometimes very slow performance, may be there ishB > some mismatch between switch and interface (may be the same with* > Alphas) like duplex, flow control or so. >   7 The test was done using netperf, Rick Jones is the besto6 source of info on this tool, he may have had a hand in writing it.    Regardse Andrew Harrisona   ------------------------------  % Date: Fri, 22 Nov 2002 15:46:35 -0000o* From: "Richard Brodie" <R.Brodie@rl.ac.uk>. Subject: Re: poor Gigabit Ethernet performance+ Message-ID: <arljgs$o34@newton.cc.rl.ac.uk>   ; "Matt Muggeridge" <Matt.Muggeridge@hp.com> wrote in messages+ news:9mWC9.3$lv1.151672@news.cpqcorp.net...aL > I don't have the time to look this up, but if memory serves me, the XP1000 > is a 33MHz 32-bit PCI bus.  W I believe it has 32 and 64. Putting it in one of the 64-bit slots would be a good plan.y   ------------------------------  % Date: Fri, 22 Nov 2002 16:31:04 +0000-' From: Andrew Harrison SUNUK Consultancy2. Subject: Re: poor Gigabit Ethernet performance. Message-ID: <3DDE5BC8.6080302@nospamn.sun.com>   Matt Muggeridge wrote:L > I don't have the time to look this up, but if memory serves me, the XP1000J > is a 33MHz 32-bit PCI bus.  Your performance is greatly reduced comparedJ > with the faster 64MHz 64-bit bus.  On higher performing systems you will9 > readily see throughput in excess of 700Mb/s using TTCP.s >   < The bandwidth available through a 33MHz 32bit PCI bus should be more than the GigE network.  * 132 MB/s if my memory serves me correctly.  , Of course if all your other I/O is using the% same bus then this could be an issue.o     Regards> Andrew Harrisono  K > Apart from file I/O getting in the way, FTP is not designed to be a speedtL > demon.  Avoid using it for performance testing.  I will add that by savingM > your file to the NL: device, you will remove the write performance from theiN > equation.  Placing the source file on a RAM DISK will minimise the impact ofD > file read IO.  Having said that, you may as well use TTCP instead. > * >     $ @sys$manager:tcpip$define_commands >     $ ttcp -hb >  >     SERVER> ttcp -r -s6 >     CLIENT> ttcp -t -s -l 65535 -n 10000 -f M SERVER > " > Enable jumbo frames with either: >  >     Edit systartup_vms.com:aB >     $ mcr lancp set dev ewa/jumbo ! enable jumbo frames for GbE. >  >     or edit modparams.dat:E >     LAN_FLAGS=64 ! Enable jumbo frames on GbE, a dynamic parameter.s= >     ! Same as MC LANCP SET DEV EWA/JUMBO and restart TCP/IPV >     Now restart TCP/IP.i= >     $ ifconfig we0  ! check the ipmtu is set to 7534 bytes.c > J > You need to be sure your switch also supports jumbo-frames and that bothN > ends negotiate the correct size.  The most reliable way to do this is with a > trace. >  > Try this experiment: >  >   Setup 3 windows:$ >     1 - Server side ($ ttcp -r -s)+ >     2 - Client side ($ ttcp -t -s server)2I >     3 - Client side ($ pipe tcptrace/prot=tcp/pack=2/port=remo=5001 | -s. >                        search sys$pipe 0402) >  >   Here is a sample output: > M >   NOTE that the results show two lines.  The first line is from the TCP SYN9. >   packet, and the second is the TCP SYN ACK. >  >   Ethernet
 >   ---------:I >   $ pipe tcptrace/prot=tcp/pack=2/port=remo=5001 | search sys$pipe 0402lL >      05030301   B4050402   00006F9C   FFFF02B0    0020    .....o..........L >      05030301   B4050402   000094AA   FFFF12B0    0020    ................ > # >   NOTE: B405 => 05B4 = 1460 bytes  >  >  >   GbEn >   ---n > I >   $ pipe tcptrace/prot=tcp/pack=2/port=remo=5001 | search sys$pipe 0402hL >      05030301   461D0402   000090E5   FFFF02B0    0020    ...........F....L >      05030301   461D0402   000041F0   FFFF12B0    0020    .....A.....F.... > $ >   NOTE: 461D => 1D46 = 7494 bytes. >  > Matt.	 >  > --? > -------------------------------------------------------------S > OpenVMS TCP/IP Engineering > Enterprise Computing Group > Hewlett-Packard Companyp > Gold Coast, AUSTRALIAo? > -------------------------------------------------------------a >  > 5 > "Rudolf Wingert" <win@fom.fgan.de> wrote in messages2 > news:200211200746.IAA15385@sinet1.fom.fgan.de... >  >>Hello, >>F >>I have measured the Gigabit performance and was very unhappy, to seeE >>only about 26% of the possible network performance. Is this normal? E >>(HP/Compaq meens yes) Or must I change any device parameter in con-rB >>junction with the switch parameter? In case of a mixed speed LANF >>i have the jumbo frames disabled and the PIPELINE QUOTA set to a lowD >>value. Could this be the reason of slow performance? Are there anyJ >>SYSGEN parameter or user QUOTA, which influence the network performance.7 >>Aaaaah I did forget to write the configuration infos:  >>: >>XP1000, DS20 and/or ES40 with DEGPA-SA and OpenVMS 7.1-2 >>  >>TIA and regards Rudolf Wingert >> >  >  >    ------------------------------    Date: 22 Nov 2002 03:08:46 -08007 From: stephen_bainbridge@yahoo.co.uk (Steve Bainbridge) - Subject: Re: Read/Write Locks in an SMP setups= Message-ID: <a48f6f51.0211220308.4f76390f@posting.google.com>e   Hi,u  C Thanks for the reply's so far, but it seems I was a little light onv detail. Here it is./  D I need to provide protection to a shared memory data structure. ThisE is a large Index structure which essentially maps a text name onto an-E address in a global section using hashing. There will be a maximum oftA 1 million entries in this index. The data structure for the indexh looks like:3  0 struct ts_idx           /* idx global section */   {04   UNSIGNED sic_hash_table[C_IDX_NUM_HASH_ENTRIES+1];4   UNSIGNED ric_hash_table[C_IDX_NUM_HASH_ENTRIES+1];  8   struct ts_idx_entry  idx_table[C_IDX_NUM_IDX_ENTRIES];     UNSIGNED creates;e   UNSIGNED deletes;    UNSIGNED finds;I   UNSIGNED undeletes;0   UNSIGNED change_key;     UNSIGNED numentries;   UNSIGNED numdelentries;    struct ts_head freehd;     UNSIGNED intact;   UNSIGNED chkpt;f    struct ts_system_time chktime;   };  C The protection method must allow multiple readers to simultaneously-E access the data structure and perform lookups and allow only a single A writer to update the structure at any point in time. There may be 7 multiple writers, but only one can update it at a time.   D The software will run on a single node ES40 or ES45 with 4x1GHz CPUs@ under OpenVMS7.3-1.  But as this is only a small part of a largeB application it needs to be efficent as most of the machine will be- consumed with other parts of the application.U  B I know that the $ENQ functions will work (using Protected Read andE Protected Write access) but my concern is that they will use too much,E resources and leave little left for the remainder of the application.e  6 I was considering using the atomic C functions such asE __INTERLOCKED_TESTBITSS_QUAD. Are there any examples of code that useiC similar functions that will allow multiple readers/single writers ?    Cheers,  Stevet   ------------------------------  # Date: Fri, 22 Nov 2002 13:26:20 GMTD? From: Jim.Johnson@software-exploration.nospam.com (Jim Johnson) - Subject: Re: Read/Write Locks in an SMP setup 0 Message-ID: <3dde2c56.13790559@news.demon.co.uk>   Steve,  C There are a number of assumptions that you're making.  I don't know > that they're wrong, but it's worth pointing a few of them out:  E - the cost of even a local $ENQ/$DEQ is too high to allow you to meetC< your real time requirements.  This one if probably the least contentious in my book.d  C - it is ok for a blocked process to spin, consuming CPU time.  ThisiF may be the best bet if the length of time it is going to spin is very,; very short.  Do realize this will get in the way of process-C scheduling, which might then get in the way of getting the spinning  process unblocked.  A - serializing reads will preclude you from meeting your real time D requirements, even though the amount of time you need to hold a lock: must be small (or the previous assumption could not hold).  . So, with those assumptions, a couple of ideas:    B Can the hashed entries be chained, or is this really a flat table?  C I ask, because if you don't have to worry about walking chains, thedA reader problem comes down to "is there a chance that the data has>@ changed since I started to read the entry?".  If you can get theA problem down to that, then you can do something that should scalea quite well:s  B - add two modification numbers to each entry.  The first is bumpedB before any data is updated in the entry, and the second is updated# once all the data has been updated.r  D - used testbitss for single stream the updaters.  If you have a flatD structure, you can probably put the lock bit in each entry.  If thatB doesn't work, you can certainly put it against the global section.1 This will ensure that two updaters don't collide.f  @ - a reader first copies the data entry out.  It does so by firstD copying both modification numbes, and then the data.  It then checks@ to see if both modification numbers match (so we weren't lookingC during an update), and they match the current setting in the globalDF section (to ensure that a modification couldn't have occurred while we were copying the data).h    A If you have to deal with chains, or some other global structuring F problem, then my suggestion would be to look at implementing somethingB like the mutex structure that the VMS kernel uses, and probably toB implement it in such a way that you can segment the global sectionF across multiple mutexes to limit collisions.  The structure and designD of the kernel mutexes are described in the black VMS internals book.   Fwiw,e Jim.  D On 22 Nov 2002 03:08:46 -0800, stephen_bainbridge@yahoo.co.uk (Steve Bainbridge) wrote:   >Hi, > D >Thanks for the reply's so far, but it seems I was a little light on >detail. Here it is. >yE >I need to provide protection to a shared memory data structure. This F >is a large Index structure which essentially maps a text name onto anF >address in a global section using hashing. There will be a maximum ofB >1 million entries in this index. The data structure for the index >looks like: >t1 >struct ts_idx           /* idx global section */o >  {5 >  UNSIGNED sic_hash_table[C_IDX_NUM_HASH_ENTRIES+1];d5 >  UNSIGNED ric_hash_table[C_IDX_NUM_HASH_ENTRIES+1];w > 9 >  struct ts_idx_entry  idx_table[C_IDX_NUM_IDX_ENTRIES];  >o >  UNSIGNED creates; >  UNSIGNED deletes; >  UNSIGNED finds; >  UNSIGNED undeletes; >  UNSIGNED change_key;o >  >  UNSIGNED numentries;a >  UNSIGNED numdelentries; >  struct ts_head freehd;e >t >  UNSIGNED intact;e >  UNSIGNED chkpt;! >  struct ts_system_time chktime;l >  };  >iD >The protection method must allow multiple readers to simultaneouslyF >access the data structure and perform lookups and allow only a singleB >writer to update the structure at any point in time. There may be8 >multiple writers, but only one can update it at a time. >-E >The software will run on a single node ES40 or ES45 with 4x1GHz CPUs A >under OpenVMS7.3-1.  But as this is only a small part of a largedC >application it needs to be efficent as most of the machine will bet. >consumed with other parts of the application. >mC >I know that the $ENQ functions will work (using Protected Read and F >Protected Write access) but my concern is that they will use too muchF >resources and leave little left for the remainder of the application. >e7 >I was considering using the atomic C functions such ascF >__INTERLOCKED_TESTBITSS_QUAD. Are there any examples of code that useD >similar functions that will allow multiple readers/single writers ? >u	 >Cheers, h >Steve   Jim Johnsonw Software Exploration, Ltd.) (remove '.nospam' from the reply address)h   ------------------------------    Date: 22 Nov 2002 08:14:33 -0800* From: jdimille@partners.org (John DimIlle)H Subject: Re: Sue Skonetski recognized at OpenVMS Symposium for hard work= Message-ID: <af3672e7.0211220814.2ccb9082@posting.google.com>s   Congratulations Sue.  C  You surely deserve this award because you are so dedicated to VMS.b# Also thanks for a great conference.e   John        t VAXVMS <bounce@notmail.com> wrote in message news:<BA52530E3149734A9BAABDBBFA808E4903027BA2@rlghncst964.usps.gov>... > It's about damn time.  >  > You rule, Sue. > 7 > I'd have chipped in a few bucks if I'd known about it  > aforehand... > ; > http://www.openvms.org/stories.php?story=02/11/21/4003931A >  > WWWebb >  > -- >  > Kenneth Farmer > http://www.Tru64.org > http://www.OpenVMS.org > http://www.LinuxHPC.orga >  > ========================" > William W. Webb / DSSC/RLM, USPS > OpenVMS Support Services( > 4924 Green Road Raleigh, NC 27616-2800< > 919.874.3043 <FirstInitialDotLastNameAtEmailDotUSPSDotGov>   ------------------------------  % Date: Fri, 22 Nov 2002 12:34:30 -0500e; From: "Brian Tillman" <tillman_brian@notnoone.notnohow.com>b0 Subject: Re: System uptime - The Uptimes Project" Message-ID: <3dde6ad6@news.si.com>  H >Since there was no stats for OpenVMS, I decided to write a client in my& >spare time, and register a few hosts.  J Perhaps a programmer who knows how to call LIB$SUB_TIMES from C can modify& this to work on both Alphas and VAXes. -- 3A Brian Tillman                   Internet: tillman_brian at si.com A Smiths Aerospace                          tillman at swdev.si.com = 3290 Patterson Ave. SE, MS      Addresses modified to prevent < Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@"8        This opinion doesn't represent that of my company   ------------------------------  % Date: Fri, 22 Nov 2002 07:04:23 -0500T0 From: JF Mezei <jfmezei.spamnot@vl.videotron.ca># Subject: TCPIP$XDM : Security issuea. Message-ID: <3DDE1D45.C540D61@vl.videotron.ca>   X-NoArchive: Yes   VAX VMS 7.2, TCPIP 5.3-181  Q Tried the "new" XDM server. My MAC has MI/X X-server which I think only does XDM.   H Anyhow, I am able to get the XDM server to pop up a login dialogue once.D During that dialogue, one can enter as many wrong  username/password combinations  without :  	-intrusion detectionS+ 	-intrusion evasion (as bet LGI parameters)S 	-no opcom message generated 	-no audit log entry5 	-SYSUAF not updated to reflect number of login failsn  L So if you enable XDM server, it opens a door for potential password cracking= software that could go on forever without ringing any alarms.   L Me thinks the boys on the gold coast beach should be grounded and prohibited+ from going surfing until they fix that one.e  N And while I am at it, not sure about how the XDM protocol works, but if, afterI succesfully login in, I end the session, the X-terminal is then unable tohL restart a session from that node until the XDM server on the vax is shutdown and turned back on.e   ------------------------------  % Date: Fri, 22 Nov 2002 08:46:01 -0500)* From: "Syltrem" <syltremzulu@videotron.ca>( Subject: Re: Termination mailbox problem5 Message-ID: <%wqD9.18984$H67.85507@tor-nn1.netcom.ca>i  J "Bob Koehler" <koehler@encompasserve.org> a crit dans le message de news:( UlEHicXL89G4@eisner.encompasserve.org...  / >    Your program has a classic well known bug.a >    Can you define it, please?' I like to be bug-free whenever possiblee   Thanks --   Syltrem.I http://pages.infinit.net/syltrem (OpenVMS related web site - en franais)08 To reply to myself directly, remove zulu from my address  J "Bob Koehler" <koehler@encompasserve.org> a crit dans le message de news:( UlEHicXL89G4@eisner.encompasserve.org...A > In article <%E8D9.18925$H67.85056@tor-nn1.netcom.ca>, "Syltrem"." <syltremzulu@videotron.ca> writes:	 > > HelloB > > = > > I am having a strange problem with a termination mailbox.O > >eK > > After modifying a program, which previously returned correct (or rathereJ > > "expected") status codes from the termination mailbox, the program nowH > > returns the correct code but... the most significative bit is ALWAYS set.G > > That is, if the "watched" program exits with status 2C, the programe reading G > > the termination mailbox would previously see the status code as HEXu 0000002C > > now it sees it as 1000002C.o > > Why the difference?  > >sG > > My problem is that the watchdog program compares the status code tot > > SS$_ABORT.F > > SS$_ABORT is 0000002C, not 1000002C so the comparison is not equal anymore. > >  > C >    The high order bits of a condition code do not define a uniquetE >    condition.  Those bits are a control field (STS$M_CONTROL) who'st >    only defined bit is:V  >       STS$M_INHIB_MSG 10000000 >nI >    I've seen applications take liberties with other bits in the controlaJ >    field.  You really should be looking at only the STS$M_COND_ID field,G >    which is unique to the condition.  LIB$MATCH_COND is a conveniencee routinet  >    which will do this for you. > / >    Your program has a classic well known bug.  >    ------------------------------  # Date: Fri, 22 Nov 2002 14:49:11 GMTA/ From: brooks@cuebid.zko.dec.nospam (Rob Brooks)a( Subject: Re: Termination mailbox problem- Message-ID: <QP1cq1LeqBNA@cuebid.zko.dec.com>a  / koehler@encompasserve.org (Bob Koehler) writes:d+ Syltrem" <syltremzulu@videotron.ca> writes:0  F >> My problem is that the watchdog program compares the status code to
 >> SS$_ABORT. N >> SS$_ABORT is 0000002C, not 1000002C so the comparison is not equal anymore. > C >    The high order bits of a condition code do not define a uniquetF >    condition.  Those bits are a control field (STS$M_CONTROL) who's  >    only defined bit is:.  >       STS$M_INHIB_MSG 10000000 > I >    I've seen applications take liberties with other bits in the controlhK >    field.  You really should be looking at only the STS$M_COND_ID field,  O >    which is unique to the condition.  LIB$MATCH_COND is a convenience routinea  >    which will do this for you. > / >    Your program has a classic well known bug.   L status values are only 16 bits long.  You should only compare ^X002C against #ss$_abort.    -- g  M Rob Brooks    VMS Engineering -- I/O Exec Group     brooks!cuebid.zko.dec.comu   ------------------------------    Date: 22 Nov 2002 08:39:52 -0600 From: briggs@encompasserve.org( Subject: Re: Termination mailbox problem3 Message-ID: <t2GVgiGB+6+U@eisner.encompasserve.org>   _ In article <QP1cq1LeqBNA@cuebid.zko.dec.com>, brooks@cuebid.zko.dec.nospam (Rob Brooks) writes:y1 > koehler@encompasserve.org (Bob Koehler) writes: - > Syltrem" <syltremzulu@videotron.ca> writes:i > G >>> My problem is that the watchdog program compares the status code tol >>> SS$_ABORT.O >>> SS$_ABORT is 0000002C, not 1000002C so the comparison is not equal anymore.d >> aD >>    The high order bits of a condition code do not define a uniqueG >>    condition.  Those bits are a control field (STS$M_CONTROL) who's - >>    only defined bit is:! >>       STS$M_INHIB_MSG 10000000r >> eJ >>    I've seen applications take liberties with other bits in the controlL >>    field.  You really should be looking at only the STS$M_COND_ID field, P >>    which is unique to the condition.  LIB$MATCH_COND is a convenience routine! >>    which will do this for you.p >> U0 >>    Your program has a classic well known bug. > N > status values are only 16 bits long.  You should only compare ^X002C against
 > #ss$_abort.'  # VMS status values are 32 bits long.m  C However, in this case, the advice to mask out the low order 16 bits : and compare against SS$_ABORT is at entirely unreasonable.  A If bit 15 in the status code is zero (as it is in SS$_ABORT) that A means that the status is not facility-specific.  Which means thati> the facility number in bits 16-27 is somewhat irrelevant.  And< since the control field in bits 28-31 is not significant for< matching conditions, that leaves you with only the low order@ 16 bits being important.  And since you'll probably never see anA ABORT condition with a severity other than 4, it doesn't hurt you 9 to leave the bottom 3 bits turned on in your 16 bit mask.f   	John Briggs   ------------------------------    Date: 22 Nov 2002 08:52:30 -0600 From: briggs@encompasserve.org( Subject: Re: Termination mailbox problem3 Message-ID: <VAYQhZP$9$Es@eisner.encompasserve.org>g  T In article <t2GVgiGB+6+U@eisner.encompasserve.org>, briggs@encompasserve.org writes:E > However, in this case, the advice to mask out the low order 16 bitsi< > and compare against SS$_ABORT is at entirely unreasonable.  8 That should have read: "is _not_ entirely unreasonable".   	John Briggs   ------------------------------    Date: 22 Nov 2002 08:54:07 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen)n( Subject: Re: Termination mailbox problem3 Message-ID: <qLaHinNbYu3x@eisner.encompasserve.org>i  _ In article <QP1cq1LeqBNA@cuebid.zko.dec.com>, brooks@cuebid.zko.dec.nospam (Rob Brooks) writes:t1 > koehler@encompasserve.org (Bob Koehler) writes:p- > Syltrem" <syltremzulu@videotron.ca> writes:  > G >>> My problem is that the watchdog program compares the status code toh >>> SS$_ABORT.O >>> SS$_ABORT is 0000002C, not 1000002C so the comparison is not equal anymore.e >> tD >>    The high order bits of a condition code do not define a uniqueG >>    condition.  Those bits are a control field (STS$M_CONTROL) who's   >>    only defined bit is:! >>       STS$M_INHIB_MSG 10000000- >> VJ >>    I've seen applications take liberties with other bits in the controlL >>    field.  You really should be looking at only the STS$M_COND_ID field, P >>    which is unique to the condition.  LIB$MATCH_COND is a convenience routine! >>    which will do this for you.i >> o0 >>    Your program has a classic well known bug. > N > status values are only 16 bits long.  You should only compare ^X002C against
 > #ss$_abort.r   I disagree..  @ While the facility code for SS$_ is zero, that does not mean theD bits are "don't care".  A program extracting them from a 16 bit cellE (like an IOSB) is responsible for providing a clear upper part (aside B from the INHIB_MSG bit) when conveying that code in a 32 bit cell.  D Also, intermediate handlers may alter severity, so comparison shouldH use LIB$MATCH_COND (as indicated by a previous poster) or some mechanismD specific to a particular programming language like STS$V_COND_ID for, Bliss or CONDITION_HANDLING.COND_ID for Ada.   ------------------------------  % Date: Fri, 22 Nov 2002 11:17:52 -0500i* From: "Syltrem" <syltremzulu@videotron.ca>( Subject: Re: Termination mailbox problem5 Message-ID: <oLsD9.19013$H67.85470@tor-nn1.netcom.ca>W  G In light of this information, I still believe it's better if I leave mylK program as it is, since the "watched" program may return not only SS$_ABORT / but also any exit code that caused it to crash.nK SS$_ABORT only means to the watchdog "don't try to restart this one". Otherh: codes may mean "restart it, and see if it can live again".   --   Syltrem I http://pages.infinit.net/syltrem (OpenVMS related web site - en franais) 8 To reply to myself directly, remove zulu from my address  ; <briggs@encompasserve.org> a crit dans le message de news:o( t2GVgiGB+6+U@eisner.encompasserve.org...L > In article <QP1cq1LeqBNA@cuebid.zko.dec.com>, brooks@cuebid.zko.dec.nospam (Rob Brooks) writes:3 > > koehler@encompasserve.org (Bob Koehler) writes:S/ > > Syltrem" <syltremzulu@videotron.ca> writes:  > >zI > >>> My problem is that the watchdog program compares the status code to  > >>> SS$_ABORT.H > >>> SS$_ABORT is 0000002C, not 1000002C so the comparison is not equal anymore. > >>F > >>    The high order bits of a condition code do not define a uniqueH > >>    condition.  Those bits are a control field (STS$M_CONTROL) who's > >>    only defined bit is:# > >>       STS$M_INHIB_MSG 10000000  > >>L > >>    I've seen applications take liberties with other bits in the controlF > >>    field.  You really should be looking at only the STS$M_COND_ID field,J > >>    which is unique to the condition.  LIB$MATCH_COND is a convenience routine # > >>    which will do this for you.c > >>2 > >>    Your program has a classic well known bug. > > H > > status values are only 16 bits long.  You should only compare ^X002C againstb > > #ss$_abort.n >t% > VMS status values are 32 bits long.l >sE > However, in this case, the advice to mask out the low order 16 bitsO< > and compare against SS$_ABORT is at entirely unreasonable. >rC > If bit 15 in the status code is zero (as it is in SS$_ABORT) that C > means that the status is not facility-specific.  Which means thati@ > the facility number in bits 16-27 is somewhat irrelevant.  And> > since the control field in bits 28-31 is not significant for> > matching conditions, that leaves you with only the low orderB > 16 bits being important.  And since you'll probably never see anC > ABORT condition with a severity other than 4, it doesn't hurt youi; > to leave the bottom 3 bits turned on in your 16 bit mask.n >v
 > John Briggs    ------------------------------  % Date: Fri, 22 Nov 2002 05:18:52 +0100u2 From: martin@radiogaga.harz.de (Martin Vorlaender)% Subject: Re: TPU scripting for MAIL ?t; Message-ID: <3dddb02c.524144494f47414741@radiogaga.harz.de>h  , Peter Weaver (peter.weaver@stelco.ca) wrote: > JF Mezei wrote:m< > > TPU is a scripting language available on all VMS systemsC > > VMSmail is available on all VMS system but has basic functions.n > >... > > Comments ????- >uG > When I saw some of your questions about Base64 attachments in mail, Id > started thinking;  >TG >    Why not set up mail so SEND/EDIT will call a TPU editor that has aaL > routine called 'ADD ATTACHMENT' that will put a file in a different bufferJ > and encode it, when you exit the message your TPU macro will combine the0 > messages and add in the correct headers. [...]  I That ignores the blank line problem. You really can't do it from VMSmail;s, sending the message has to be done from TPU.   cu,o   Martin -- uG                            | Martin Vorlaender  |  VMS & WNT programmerw4  UNIX is user friendly.    | work: mv@pdv-systeme.deG  It's just selective about |   http://www.pdv-systeme.de/users/martinv/e;  who its friends are.      | home: martin@radiogaga.harz.dee   ------------------------------    Date: 22 Nov 2002 05:51:39 -0800- From: roli@barmettler.net (Roland Barmettler)e4 Subject: Trouble with NFS since upgrade 7.3 -> 7.3-1= Message-ID: <235cf5c3.0211220551.32b06601@posting.google.com>y   Helloo  C Since I've upgraded my NFS server from 7.3 to 7.3-1 (the huge TCPIPpC ECO 1 is applied), I'm unable to mount the NFS shares with my Linuxu workstation.C The server reports a successful mount through OPCOM, but the client  (Linux) logs the following messages:fA xxx kernel: call_verify: server requires stronger authentication.e, xxx kernel: nfs_get_root: getattr error = 13A xxx kernel: call_verify: server requires stronger authentication.c, xxx kernel: nfs_get_root: getattr error = 131 xxx kernel: nfs_read_super: get root inode failed-  @ And "TCPIP SHOW NFS" on the server reports 2 weak authentication calls.  F Has anybody an idea what might be the problem or what has changed from 7.3 ?v Thanks for your help!o   Greetings, Roland2   ------------------------------  # Date: Fri, 22 Nov 2002 13:03:45 GMTv, From: "Bob Knowles" <bob.knowles@compaq.com>* Subject: Re: What is DAP status code 5067?1 Message-ID: <RWpD9.1$O%4.151677@news.cpqcorp.net>i   Lawrence Bleau wrote:e@ > In article <fgLz9.4$H%1.23576@news.cpqcorp.net>, "Bob Knowles"" > <bob.knowles@compaq.com> writes: >> Lawrence Bleau wrote: >>H >>> Where would one find documentation on DAP error codes, anyway?  Just >>> curious. >>E >> They were listed in "DECnet SNA Data Transfer Facility for OpenVMS ' >> Use" (just a list - no explanation).N >0F > Do you have a URL for this document?  I searched the online docs and! > couldn't find anything similar.i .d .  .aI http://www.sysworks.com.au/disk$axpdocmar961/decw$book/d33haa14.decw$book   H I don't think it'd be among the online docs (on the ODL CDC), as it's noK longer a product of ours. So I don't know how reliable or long-lasting thisv# URL will be. It works today though.o -- bob.know1es@hp.com% (non-spammers should make the 1 an L)w   ------------------------------  # Date: Fri, 22 Nov 2002 09:07:22 GMTo. From: peter@langstoeger.at (Peter LANGSTOEGER)I Subject: Re: What's going on with the 7.3.1 VAX hobbyist kit at Montagar?82 Message-ID: <etmD9.21781$up.242260@news.chello.at>   In article <chqptuo6cqumdnndo8i06860muv4frgf0j@4ax.com>, Jamie Stallwood <this.no.work.try.something.else@project76.net> writes:N >On 13 Nov 2002 09:26:42 -0600, Kilgallen@SpamCop.net (Larry Kilgallen) wrote: >>In article <r8t4tug3a900d271ibr2470ntt3lpbkt2k@4ax.com>, Jamie Stallwood <this.no.work.try.something.else@project76.net> writes:0 >>> The site hasn't been updated for months now? >>( >>There is no VAX V7.3-1 release of VMS. >SG >So how do I actually GET a new oV hobbyist distribution for my VAX? Is. >VMS now defunct?n   Drop me a line...-   -- a Peter "EPLAN" LANGSTOEGERu% Network and OpenVMS system specialistt E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------  % Date: Fri, 22 Nov 2002 12:05:52 +0100 8 From: "Tomasz Dryjanski" <tdryjanski.nospam@hotmail.com>3 Subject: Re: Where can I find patches to my system? . Message-ID: <arl32g$186v$1@news2.ipartners.pl>  9 > >Also, to check for patches you already have installed,- > >the command is: > >-! > >PRODUCT SHOW PRODUCT VMS /FULL5 >ML > Not on a VAX, since patches are more often than not VMSINSTAL kits and not > PCSI kits.   Yes, I have already noted both.  Thanks for your help.9   T. D.s   ------------------------------   End of INFO-VAX 2002.646 ************************rites: >o >a >VAXman- wrote:a >> In article <3DD3991A.90402@nospamn.sun.com>, Andrew HarriǪ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    Ǫ    