defert@apostroph.cernvax.cern.ch (Philippe Defert) (10/02/90)
I generated perl kit 28 on an apollo 3500 using gcc:
- everything compiled correctly
- but test failed in 3 places, 2 of which do not seem important
. op.fs on test 5 ( $mode from stat)
. op.dbm on test 2 ( $mode from stat)
and the last failure seems to be really problematic, it is:
. op.regexp that produces a segmentation fault at the 46 th line
of the test regexp file, and on some other following lines that I
tested separately.
Has anybody already faced these problems ?
Any hints ?
Can I anyhow rely on this perl when it does not crash ?
Any help would be welcome.
Thanks
Philippe.
--
--------------------------------+---------------------------------------------
Philippe Defert: Computing and Networks Division / Application Software Group/
European Organisation for Nuclear Research (CERN)
--------------------------------+---------------------------------------------
Earth mail: | E-mail:
CERN / CN / AS | defert@cernvax.bitnet
CH-1211 Geneve 23. Switzerland. | defert@cernvax.UUCP
Voice: | defert@cernvax.cern.ch
+ 41 22 7674933 | psi%02284681140551::defertkgallagh@digi.lonestar.org (Kevin Gallagher) (10/03/90)
In article <DEFERT.90Oct2114010@apostroph.cernvax.cern.ch> defert@apostroph.cernvax.cern.ch (Philippe Defert) writes: > >I generated perl kit 28 on an apollo 3500 using gcc: >- everything compiled correctly >- but test failed in 3 places, 2 of which do not seem important > . op.fs on test 5 ( $mode from stat) > . op.dbm on test 2 ( $mode from stat) > and the last failure seems to be really problematic, it is: > . op.regexp that produces a segmentation fault at the 46 th line > of the test regexp file, and on some other following lines that I > tested separately. > >[rest deleted] Well, I built Perl, patch level 28, on an Apollo 3500 and it passes all the tests. But I did not use gcc. I did it under bsd4.3, SR10.2, using /bin/cc. Here's how I answered the Configure script questions: Question Answer Models none Compiler cc Optimizer -O Additional cc flags -DDEBUGGING Additional ld flags -ldbm -lclib Byte Order 4321 Library /lib/libc Setuid/setgid emulation n Use index(), not strchr() y Void support flag bits total 7 Group ids for getgroups() int Size of integer 4 Bits in rand 31 User ids in getuid() uid_t Use malloc that comes with Perl y Compiler compiler yacc NOTE: /bin/cc invokes Apollo's Unix pre-processor which then invokes Apollo's C compiler. Apollo's C library routines are contained in TWO libraries: /lib/libc and /lib/clib. The compiler knows about both libraries and searches both automatically. To ensure that the Configure script finds everything available on the Apollo, you must inform the script about BOTH libraries. However, since the Compiler finds /lib/clib on its own, you must REMOVE reference to it in the Makefiles generated by Configure. -- ---------------------------------------------------------------------------- Kevin Gallagher kgallagh@digi.lonestar.org OR ...!uunet!digi!kgallagh DSC Communications OR apcihq!apcidfw!digi!kgallagh ----------------------------------------------------------------------------
ianh@bhpmrl.oz.au (Ian Hoyle) (10/05/90)
defert@apostroph.cernvax.cern.ch (Philippe Defert) writes: >I generated perl kit 28 on an apollo 3500 using gcc: >- everything compiled correctly >- but test failed in 3 places, 2 of which do not seem important > . op.fs on test 5 ( $mode from stat) > . op.dbm on test 2 ( $mode from stat) > and the last failure seems to be really problematic, it is: > . op.regexp that produces a segmentation fault at the 46 th line > of the test regexp file, and on some other following lines that I > tested separately. >Has anybody already faced these problems ? >Any hints ? Don't compile with the optimization flag. All should then be well ian -- Ian Hoyle /\/\ Image Processing & Data Analysis Group / / /\ BHP Melbourne Research Laboratories / / / \ 245 Wellington Rd, Mulgrave, 3170 / / / /\ \ AUSTRALIA \ \/ / / / \ / / / Phone : +61-3-560-7066 \/\/\/ FAX : +61-3-561-6709 E-mail : ianh@bhpmrl.oz.au