mf@ircam.fr (Michel Fingerhut) (11/11/90)
Setup: DECsystem 5810
Ultrix 4.0
Using cc to compile perl
no optimization
undef'ed volatile
When testing,
lib.big.........FAILED on test 51
Help, anyone?cander@unisoft.UUCP (Charles Anderson) (11/14/90)
From article <1990Nov10.230258.21946@ircam.ircam.fr>, by mf@ircam.fr (Michel Fingerhut): > Setup: DECsystem 5810 > Ultrix 4.0 > When testing, > lib.big.........FAILED on test 51 > Help, anyone? I can't offer any help, but I too am seeing problems with lib.big. System: Pyramid-9820 OSx128MNQ 5.0b-891013 BSD universe. Compiled with perl's malloc: mallocsrc='malloc.c' mallocobj='malloc.o' usemymalloc='y' When running lib.big alone,I get the following: ok 55 Bad free() ignored at ./lib.big line 14, <DATA> line 60. ok 56 ... ok 217 not ok 218 # '&bmod('+5000000000','+9');' expected: '+5' got: '+4' ok 219 not ok 220 # '&bmod('+7000000000','+9');' expected: '+7' got: '+6' ok 221 -- Unix is the Winchester Mystery | Charles Anderson, UniSoft Corp. House of software design. | {sun, ucbvax, uunet}!unisoft!cander
gorpong@ping.uucp (Gordon C. Galligher) (11/14/90)
In article <1990Nov10.230258.21946@ircam.ircam.fr> mf@ircam.fr (Michel Fingerhut) writes: >Setup: DECsystem 5810 > Ultrix 4.0 > > Using cc to compile perl > no optimization > undef'ed volatile > >When testing, > lib.big.........FAILED on test 51 > >Help, anyone? I have an 80386 and used the GCC 1.37 compiler (with -g -O -fpcc-struct-return) and lib.big fails on test 6, when run from "make test." If I cd t and then execute ./perl lib.big, every test comes out OK. I am confused, but it does work. I also believe that the first line of lib.big should be ./perl -x not just perl, in order to take advantage of the __END__ stuff in the file. I executed ./perl lib.big and ./perl -x lib.big and it works regardless of the -x switch. Is it a side effect that if an __END__ exists then the rest is automatically used as <DATA> and the -x is not needed, or is the -x merely there to tell perl to skip to a #!perl and start executing to an __END__? -- Gordon. -- Gordon C. Galligher 9127 Potter Rd. #2E Des Plaines, IL 60016-4881 ...!uunet!telxon!teleng!ping!gorpong
lwall@jpl-devvax.JPL.NASA.GOV (Larry Wall) (11/16/90)
In article <1990Nov14.140021.5953@ping.uucp> gorpong@ping.uucp (Gordon C. Galligher) writes:
: I also believe that the first line of lib.big should be ./perl -x
: not just perl, in order to take advantage of the __END__ stuff in the file.
: I executed ./perl lib.big and ./perl -x lib.big and it works regardless of
: the -x switch. Is it a side effect that if an __END__ exists then the rest
: is automatically used as <DATA> and the -x is not needed, or is the -x merely
: there to tell perl to skip to a #!perl and start executing to an __END__?
The -x has nothing to do with __END__, other than the similarity that
both are used to trim excess garbage from a script. The -x lets you trim
garbage before #!. The __END__ lets you trim garbage after the __END__.
The -x is really only useful when the #! line is not the first line of the
file. Ordinary scripts never need it--the intent is to allow you to pipe
articles to perl that contain scripts.
<DATA> will work any time there is an __END__.
Your make problem sounds like it stems from a make that supplies a different
context than the shell. Perhaps there is some subtle interaction between
DATA and STD{IN,OUT,ERR}. Some makes attach STDIN to /dev/null, for
instance. I don't know why that would louse up DATA though.
Larry