beshers@cs.columbia.edu (Clifford Beshers) (12/05/90)
I've just received my HP C++ compiler and discovered that update dumps core with a floating point exception when it tries to read the tape. The contents of the tape seem okay; I can extract them with tar and everything seems to work. However, I can't get update to work from disk, nor from tape. We are running HPUX version 7.0. Any suggestions? -- ----------------------------------------------- Clifford Beshers 450 Computer Science Department Columbia University New York, NY 10027 beshers@cs.columbia.edu
paul@prcrs.UUCP (Paul Hite) (01/02/91)
In article <BESHERS.90Dec4181505@division.cs.columbia.edu>, beshers@cs.columbia.edu (Clifford Beshers) writes: > I've just received my HP C++ compiler and discovered that update > dumps core with a floating point exception when it tries to read > the tape. The contents of the tape seem okay; I can extract them with > tar and everything seems to work. However, I can't get update to > work from disk, nor from tape. We are running HPUX version 7.0. Well, I'm not sure what you mean by "get update to work from disk, nor from tape." But you're right, the 7.0 update does not understand the c++ tape. The tape seems to be an ordinary tar archive and as you report, just tar'ing the stuff off seems to work fine. Sorry for the delay in responding but I just got around to trying our C++ tape. Paul Hite PRC Realty Systems McLean,Va uunet!prcrs!paul (703) 556-2243 You can't tell which way the train went by studying its tracks.
bla@hpcupt1.cup.hp.com (Brad Ahlf) (01/04/91)
> I've just received my HP C++ compiler and discovered that update > dumps core with a floating point exception when it tries to read > the tape. The contents of the tape seem okay; I can extract them with > tar and everything seems to work. However, I can't get update to > work from disk, nor from tape. We are running HPUX version 7.0. |Well, I'm not sure what you mean by "get update to work from disk, nor |from tape." But you're right, the 7.0 update does not understand the c++ |tape. The tape seems to be an ordinary tar archive and as you report, |just tar'ing the stuff off seems to work fine. |Paul Hite PRC Realty Systems McLean,Va uunet!prcrs!paul (703) 556-2243 I sent email to Paul and am investigating this problem. We currently cannot duplicate these problem reports with our HP C++ product master tapes and standard 7.0 update. It works fine for us in all possible configurations that we have tried. Brad Ahlf bla@hpda.hp.com This response does not represent the official position of, or statement by, the Hewlett-Packard Company. The above data is provided for informational purposes only. It is supplied without warranty of any kind.
tml@tik.vtt.fi (Tor Lillqvist) (01/07/91)
In article <-286539933@hpcupt1.cup.hp.com> bla@hpcupt1.cup.hp.com (Brad Ahlf) writes: |Well, I'm not sure what you mean by "get update to work from disk, nor |from tape." But you're right, the 7.0 update does not understand the c++ |tape. I had problems with a pre-release or something version of C++ and update too. The problem was that 7.0 update doesn't like the + chars in filenames (like /usr/bin/c++filt) in the CDFinfo file. Unpacking the tar file by hand worked, after which I manually created the CDF files based on the CDFinfo file (well, actually I munged it into a shell script using Emacs, of course). We bought C++ for both s800 and s300 in a mixed cluster, so we need the CDF files. -- Tor Lillqvist, working, but not speaking, for the Technical Research Centre of Finland
bla@hpcupt1.cup.hp.com (Brad Ahlf) (01/08/91)
> I had problems with a pre-release or something version of C++ and > update too. The problem was that 7.0 update doesn't like the + chars > in filenames (like /usr/bin/c++filt) in the CDFinfo file. Unpacking > the tar file by hand worked, after which I manually created the CDF > files based on the CDFinfo file (well, actually I munged it into a > shell script using Emacs, of course). We bought C++ for both s800 and > s300 in a mixed cluster, so we need the CDF files. Correct. In a HP-internal-only prerelease of HPC++, this problem did exist. No normal customer shipment tapes should have this problem. Only old tapes obtained from internal HP persons should ever have this problem. A script to tar the files and create CDFs was also temporarily made available within HP for this problem. If you have a 'normal customer shipment' tape with this problem please send me email. Note that this caused CDFinfo parsing errors in update and never (that we have seen) caused update to dump core. Brad Ahlf bla@hpda.hp.com P.S. Paul Hite, I sent you email and have not yet received a reply. This response does not represent the official position of, or statement by, the Hewlett-Packard Company. The above data is provided for informational purposes only. It is supplied without warranty of any kind.
irf@kuling.UUCP (Bo Thide') (01/08/91)
In article <1370@prcrs.UUCP> paul@prcrs.UUCP (Paul Hite) writes: >In article <BESHERS.90Dec4181505@division.cs.columbia.edu>, beshers@cs.columbia.edu (Clifford Beshers) writes: >> I've just received my HP C++ compiler and discovered that update >> dumps core with a floating point exception when it tries to read >> the tape. The contents of the tape seem okay; I can extract them with >> tar and everything seems to work. However, I can't get update to >> work from disk, nor from tape. We are running HPUX version 7.0. > >Well, I'm not sure what you mean by "get update to work from disk, nor >from tape." But you're right, the 7.0 update does not understand the c++ >tape. The tape seems to be an ordinary tar archive and as you report, >just tar'ing the stuff off seems to work fine. > >Sorry for the delay in responding but I just got around to trying our C++ >tape. > >Paul Hite PRC Realty Systems McLean,Va uunet!prcrs!paul (703) 556-2243 > You can't tell which way the train went by studying its tracks. Hmm.. I didn't have any problems adding C++ to my HP9000/400t (HP-UX 7.03) with the 'update' command. Everything went as expected. I even went through the update process twice to make absolutely sure I didn't miss any error messages or other strange things the first time. The label on my C++ cartridge says: HP C++ LANGUAGE HP9000 SERIES 300 HP-UX CART 1 OF 1 UPDATE Format p/n B1691-19002 Rev. 0700 (plus copyright notices from HP, Univ. of Calif., Univ. of Colo., AT&T, Sun, Digital, MIT, and Stanford Univ.) Maybe you should check and see if we have the same version of C++ update cartridge ..? Adding C++ to my system gave me another "problem". I now seem to have three different HP C (cc) compilers: /bin/cc (the usual 7.0/7.03 compiler) /usr/softbench/bin/cc (the one I got with SoftBench/Encapsulator) /usr/CC/bin/cc (the one in the C++ package, with ANSI support) Am I right in assuming I can replace /bin/cc with /usr/CC/bin/cc and get one with full ANSI support and dynamic internal table sizes? Benchmarking with Dhrystone 2.1 (no registers) gives about 17600 dhrys/sec for /bin/cc and about 17500 dhrys/sec for the other two. I am not sure if this difference is significant (I didn't run enough times to narrow down the error bars). Bo bt@irfu.se ^ Bo Thide'-------------------------------------------------------------- |I| Swedish Institute of Space Physics, S-755 91 Uppsala, Sweden |R| Phone: (+46) 18-303671. Telex: 76036 (IRFUPP S). Fax: (+46) 18-403100 /|F|\ INTERNET: bt@irfu.se UUCP: ...!uunet!sunic!irfu!bt ~~U~~ -----------------------------------------------------------------sm5dfw
bla@hpcupt1.cup.hp.com (Brad Ahlf) (01/10/91)
> Hmm.. I didn't have any problems adding C++ to my HP9000/400t (HP-UX > 7.03) with the 'update' command. Everything went as expected. > I even went through the update process twice to make absolutely sure > I didn't miss any error messages or other strange things the first time. Good to hear it! Paul & Clifford: Is your system a diskless system? Did you get your tape shipped normally from HP or given to you by an HP person? What are the p/n's on the tape? Did your update process result in CDFinfo parse errors or a core dump? Please send email to 'bla@hpda.hp.com'. > The label on my C++ cartridge says: Looks correct. Manufacturing p/n ending in '2' instead of '1' is good. > Maybe you should check and see if we have the same version of > C++ update cartridge ..? Good idea. S300: Look for manufacturing part number 'p/n B1691-19002'. That is the 'good' tape, while 'p/n B1691-19001' is the 'bad' tape. S800: Look for bad=p/n 92501-13501 and good=p/n 92501-13502, etc, etc. Once again, only 'good' p/n tapes should have been sent in customer shipments. The 'bad' tapes were early HP-internal-only tapes. Brad Ahlf bla@hpda.hp.com This response does not represent the official position of, or statement by, the Hewlett-Packard Company. The above data is provided for informational purposes only. It is supplied without warranty of any kind.
paul@prcrs.UUCP (Paul Hite) (01/10/91)
In article <-286539930@hpcupt1.cup.hp.com>, bla@hpcupt1.cup.hp.com (Brad Ahlf) writes: > > P.S. Paul Hite, I sent you email and have not yet received a reply. > [ I tried mailing this, but I don't believe that I ever got through...] Hmmmm... I didn't get any mail from you. But anyway here is what I think you want. Since my posting I have found that the C++ Developer tape is also not understood by update. Both my C++ tape and my C++ Developer tapes are real customer releases. The part number on the C++ Developer tape is B1697-10001 and the Rev is A.00.00. In the script below I first mounted a 128 user license tape and got a table of contents. Then I switched to the C++ Dev tape and attempted a table of contents. Finally, I just use tar to display the files but I truncated the tar listing. I get identical results on two different systems both running 7.0... Script started on Wed Jan 9 12:49:06 1991 <21> id uid=0(root) gid=1(other) <22> update -c # PARTITION.FILESET # bytes media flags OS_KERNEL.UNL_USER # 53627 1 .C.H. ======<<<<< tape switch occurs here >>>>>>======= <23> update -c update: ERROR: The architecture of the files on the update media is unknown. Use the -S option to indicate whether the files on the media are for a Series 300 or Series 800 system. <24> tar tvf /dev/rmt/0m rw-r--r-- 0/3 315 Oct 17 16:36 1990 system/INDEX rw-r--r-- 0/3 2271 Oct 17 16:36 1990 system/INFO rw-r--r-- 0/3 89 Oct 17 16:36 1990 system/CDFinfo rwxr-xr-x 2/2 0 Oct 17 16:36 1990 CXXDEV/.././ rwxr-xr-x 2/2 0 Oct 17 16:33 1990 CXXDEV/../usr/ rwxr-xr-x 2/2 0 Oct 17 16:33 1990 CXXDEV/../etc/ rwxr-xr-x 2/2 0 Oct 17 16:36 1990 CXXDEV/../system/ rwxr-xr-x 2/2 0 Oct 17 16:32 1990 CXXDEV/../usr/bin/ rwxr-xr-x 2/2 1394 Oct 17 15:28 1990 CXXDEV/../usr/bin/cdvi rwxr-xr-x 2/21486848 Oct 17 15:28 1990 CXXDEV/../usr/bin/cxxdev I hope that this is what you wanted. I don't know what happened to your mail. If uunet!prcrs!paul doesn't work try paul@prcrs.prc.com But after my experiences trying to mail to you, I'm not sure that you can get here from there. Paul Hite PRC Realty Systems McLean,Va uunet!prcrs!paul (703) 556-2243 You can't tell which way the train went by studying its tracks.
bla@hpcupt1.cup.hp.com (Brad Ahlf) (01/24/91)
An 'update' on the HPC++ and HPC++Developer 'update' tape problem: The 'update' format for these tapes is not 100% correct. An 'update' format specification (for architecture type) was omitted. This leads to an error message from 'update -c'. Most customers will never see this problem. It is also very easily worked around by running update without the -c option or by specifying the -S option with the -c option. These 'update' tapes will still work just fine for updating your HPC++ and HPC++Developer software using the 'update' program in all of the other normal ways such as using 'update' without -c for interactive use and also 'update' with -s and -S for command-line use. Future 'update' tapes will, of course, not have this problem. Brad Ahlf bla@hpda.hp.com This response does not represent the official position of, or statement by, the Hewlett-Packard Company. The above data is provided for informational purposes only. It is supplied without warranty of any kind.