search

Perl / perl5

🐪 The Perl programming language
https://dev.perl.org/perl5/
Other
1.93k stars 551 forks source link

Not OK: perl v5.6.1 +v5.6.1-TRIAL1 on VMS_AXP V7.2-1 #3011

Closed p5pRT closed 20 years ago

p5pRT commented 23 years ago

Migrated from rt.perl.org#4966 (status was 'resolved')

Searchable as RT4966$

p5pRT commented 23 years ago

From dsugalski@northernlight.com

lib/peek.t test 12 fails with a full 64-bit build. All the other tests go OK.

Perl Info ``` Flags: category=install severity=none Site configuration information for perl v5.6.1: Configured by dsugalski at Mon Dec 18 14:52:18 UTC-5:00 2000. Summary of my perl5 (revision 5.0 version 6 subversion undef) configuration: Platform: osname=VMS, osvers=V7.2-1, archname=VMS_AXP uname='VMS monsoon V7.2-1 Compaq AlphaServer GS140 6/700' config_args='' hint=none, useposix=false, d_sigaction=define usethreads=undef use5005threads=undef useithreads=undef usemultiplicity=undef useperlio=undef d_sfio=undef uselargefiles=define usesocks=undef use64bitint=define use64bitall=define uselongdouble=define Compiler: cc='CC/DECC', ccflags ='/Include=[]/Standard=Relaxed_ANSI/Prefix=All/Obj=.obj/NoList/NOANSI_ALIAS', optimize='', cppflags='undef' ccversion='60290007', gccversion='', gccosandvers='undef' intsize=4, longsize=4, ptrsize=4, doublesize=8, byteorder=12345678 d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=16 ivtype='long long', ivsize=8, nvtype='long double', nvsize=16, Off_t='int', lseeksize=4 alignbytes=8, usemymalloc=N, prototype=define Linker and Libraries: ld='Link', ldflags ='/NoTrace/NoMap' libpth=/sys$share /sys$library libs= perllibs= libc=(DECCRTL), so=exe, useshrplib=undef, libperl=undef Dynamic Linking: dlsrc=dl_vms.c, dlext=exe, d_dlsymun=undef, ccdlflags='' cccdlflags='', lddlflags='/Share' Locally applied patches: v5.6.1-TRIAL1 @INC for perl v5.6.1: perl_root:[lib.VMS_AXP.5_6_1] perl_root:[lib] perl_root:[lib.site_perl.VMS_AXP] perl_root:[lib.site_perl] /perl_root/lib/site_perl . Environment for perl v5.6.1: HOME=user01:[dsugalski] LANG (unset) LANGUAGE (unset) LD_LIBRARY_PATH (unset) LOGDIR (unset) PATH=user01:[dsugalski.perl-5_6_1-trial1] PERLSHR=PERL_ROOT:[000000]PERLSHR.EXE PERL_BADLANG (unset) PERL_ROOT=USER01:[DSUGALSKI.PERL-5_6_1-TRIAL1.] SHELL (unset) ```
p5pRT commented 23 years ago

From @jhi

On Mon\, Dec 18\, 2000 at 04​:17​:25PM -0500\, dsugalski@​northernlight.com wrote​:

This is a build failure report for perl from scomer@​MONSOON.STRATUS.NORTHERNLIGHT.COM\, generated with the help of perlbug 1.33 running under perl v5.6.1.

----------------------------------------------------------------- [Please enter your report here]

lib/peek.t test 12 fails with a full 64-bit build. All the other tests go OK.

Could you please run the test separately so that we can see what's the difference it is complaining about.

Summary of my perl5 (revision 5.0 version 6 subversion undef) configuration​:

subversion undef?

p5pRT commented 23 years ago

From [Unknown Contact. See original ticket]

At 03​:52 PM 12/18/00 -0600\, Jarkko Hietaniemi wrote​:

On Mon\, Dec 18\, 2000 at 04​:17​:25PM -0500\, dsugalski@​northernlight.com wrote​:

This is a build failure report for perl from scomer@​MONSOON.STRATUS.NORTHERNLIGHT.COM\, generated with the help of perlbug 1.33 running under perl v5.6.1.

----------------------------------------------------------------- [Please enter your report here]

lib/peek.t test 12 fails with a full 64-bit build. All the other tests go OK.

Could you please run the test separately so that we can see what's the difference it is complaining about.

Yup​:

[SV = RV(0x263168) at 0x1ec0d0   REFCNT = 1   FLAGS = (ROK)   RV = 0x1ebf2c   SV = PVHV(0x207118) at 0x1ebf2c   REFCNT = 2   FLAGS = (SHAREKEYS)   IV = 1   NV = 0   ARRAY = 0x2c8fd8 (0​:7\, 1​:1)   hash quality = 0.0%   KEYS = 1   FILL = 1   MAX = 7   RITER = -1   EITER = 0x0   Elt "123" HASH = 0xddce   SV = PVNV(0x26f948) at 0x1ec10c   REFCNT = 1   FLAGS = (IOK\,NOK\,pIOK\,pNOK)   IV = 456   NV = 456   PV = 0 ] vs [SV = RV\(0x[[​:xdigit​:]]+\) at 0x[[​:xdigit​:]]+   REFCNT = 1   FLAGS = \(ROK\)   RV = 0x[[​:xdigit​:]]+   SV = PVHV\(0x[[​:xdigit​:]]+\) at 0x[[​:xdigit​:]]+   REFCNT = 2   FLAGS = \(SHAREKEYS\)   IV = 1   NV = 0   ARRAY = 0x[[​:xdigit​:]]+ \(0​:7\, 1​:1\)   hash quality = 150.0%   KEYS = 1   FILL = 1   MAX = 7   RITER = -1   EITER = 0x0   Elt "123" HASH = 0x[[​:xdigit​:]]+   SV = PVNV\(0x[[​:xdigit​:]]+\) at 0x[[​:xdigit​:]]+   REFCNT = 1   FLAGS = \(IOK\,NOK\,pIOK\,pNOK\)   IV = 456   NV = 456   PV = 0] not ok 12

Summary of my perl5 (revision 5.0 version 6 subversion undef) configuration​:

subversion undef?

Looks like configure.com needs another whacking about...

  Dan

--------------------------------------"it's like this"------------------- Dan Sugalski even samurai dan@​sidhe.org have teddy bears and even   teddy bears get drunk

p5pRT commented 23 years ago

From @jhi

 hash quality = 0\.0%

 hash quality = 150\.0%

Looks like NVff ain't right.

p5pRT commented 23 years ago

From @craigberry

At 4​:55 PM -0500 12/18/00\, Dan Sugalski wrote​:

At 03​:52 PM 12/18/00 -0600\, Jarkko Hietaniemi wrote​:

Summary of my perl5 (revision 5.0 version 6 subversion undef) configuration​:

subversion undef?

Looks like configure.com needs another whacking about...

The problem appears to be the following commented-out line in configure.com​:

$! WC "PERL_SUBVERSION='" + subversion + "'" ! VMS specific to descrip_mms.template

Anyone know why this is commented out or what descrip_mms.template has to do with it?

p5pRT commented 23 years ago

From @craigberry

At 10​:55 PM -0600 12/18/00\, Craig A. Berry wrote​:

At 4​:55 PM -0500 12/18/00\, Dan Sugalski wrote​:

At 03​:52 PM 12/18/00 -0600\, Jarkko Hietaniemi wrote​:

Summary of my perl5 (revision 5.0 version 6 subversion undef) configuration​:

subversion undef?

Looks like configure.com needs another whacking about...

The problem appears to be the following commented-out line in configure.com​:

$! WC "PERL_SUBVERSION='" + subversion + "'" ! VMS specific to descrip_mms.template

Anyone know why this is commented out or what descrip_mms.template has to do with it?

It's been almost a month\, I haven't heard a reply to this question\, and our subversion is still missing. I say we put it back in and place the burden of explanation on whoever commented it out in the first place.

Inline Patch ```diff --- configure.com;-0 Sun Jan 14 23:11:41 2001 +++ configure.com Tue Jan 16 23:12:15 2001 @@ -4835,7 +4835,7 @@ $ WC "CONFIG='true'" $ WC "Makefile_SH='" + Makefile_SH + "'" $ WC "Mcc='" + Mcc + "'" -$! WC "PERL_SUBVERSION='" + subversion + "'" ! VMS specific to ```

descrip_mms.template +$ WC "PERL_SUBVERSION='" + subversion + "'" ! VMS specific to descrip_mms.template   $ WC "PERL_VERSION='" + patchlevel + "'" ! VMS specific to descrip_mms.template   $ WC "alignbytes='" + alignbytes + "'"   $ WC "aphostname='" + "'"

p5pRT commented 23 years ago

From @jhi

On Tue\, Jan 16\, 2001 at 11​:38​:46PM -0600\, Craig A. Berry wrote​:

At 10​:55 PM -0600 12/18/00\, Craig A. Berry wrote​:

At 4​:55 PM -0500 12/18/00\, Dan Sugalski wrote​:

At 03​:52 PM 12/18/00 -0600\, Jarkko Hietaniemi wrote​:

Summary of my perl5 (revision 5.0 version 6 subversion undef) configuration​:

subversion undef?

Looks like configure.com needs another whacking about...

The problem appears to be the following commented-out line in configure.com​:

$! WC "PERL_SUBVERSION='" + subversion + "'" ! VMS specific to descrip_mms.template

Anyone know why this is commented out or what descrip_mms.template has to do with it?

It's been almost a month\, I haven't heard a reply to this question\, and our subversion is still missing. I say we put it back in and place the burden of explanation on whoever commented it out in the first place.

It seems to have been the "Change 6613 by jhi@​alpha on 2000/08/14 13​:45​:33"\, "VMS configure.com update continues." I *think* I got it from Peter Prymmer\, unfortunately I didn't seem to document it back then...

p5pRT commented 23 years ago

From [Unknown Contact. See original ticket]

On Wed\, 17 Jan 2001\, Jarkko Hietaniemi wrote​:

On Tue\, Jan 16\, 2001 at 11​:38​:46PM -0600\, Craig A. Berry wrote​:

At 10​:55 PM -0600 12/18/00\, Craig A. Berry wrote​:

At 4​:55 PM -0500 12/18/00\, Dan Sugalski wrote​:

At 03​:52 PM 12/18/00 -0600\, Jarkko Hietaniemi wrote​:

Summary of my perl5 (revision 5.0 version 6 subversion undef) configuration​:

subversion undef?

Looks like configure.com needs another whacking about...

The problem appears to be the following commented-out line in configure.com​:

$! WC "PERL_SUBVERSION='" + subversion + "'" ! VMS specific to descrip_mms.template

Anyone know why this is commented out or what descrip_mms.template has to do with it?

It's been almost a month\, I haven't heard a reply to this question\, and our subversion is still missing. I say we put it back in and place the burden of explanation on whoever commented it out in the first place.

It seems to have been the "Change 6613 by jhi@​alpha on 2000/08/14 13​:45​:33"\, "VMS configure.com update continues." I *think* I got it from Peter Prymmer\, unfortunately I didn't seem to document it back then...

It might have come from me although I don't recall it very well.

It might have been fallout from a change in patchlevel.h in the 5.6.0 release of some sort (confusion over the intent of subversion vs api_subversion?). FWIW here are the versions that a unix box has​:

% grep -i version config.sh api_subversion='0' api_version='5' api_versionstring='5.005' ccversion='V5.6-082' gccversion='' ignore_versioned_solibs='' inc_version_list='5.005/alpha-dec_osf 5.005' inc_version_list_init='"5.005/alpha-dec_osf"\,"5.005"\,0' pm_apiversion='5.005' subversion='0' version='5.7.0' versiononly='undef' xs_apiversion='5.005' PERL_VERSION=7 PERL_SUBVERSION=0 PERL_API_VERSION=5 PERL_API_SUBVERSION=0

The other possiblilty is that it was to head off an unitialized variable warning in DCL. For example​:

% type subver.com $ WC := write sys$output $! without initialization $ WC "subversion='" + subversion + "'" $! with initialization to empty string $ subversion = "" $ WC "subversion='" + subversion + "'" $! with initialization to nonempty string $ subversion = "0" $ WC "subversion='" + subversion + "'" % @​subver %DCL-W-UNDSYM\, undefined symbol - check validity and spelling \SUBVERSION\ subversion='' subversion='0'

And it could very well be that a bug in configure.com prevents the $ subversion="0" assignment from kicking in.

Perhaps another thing worth testing is what munchconfig.exe and descrip.mms would do with​:

PERL_SUBVERSION=0

rather than with the 'single quotation marks'​:

PERL_SUBVERSION='0'

I'll try to look into it.

Peter Prymmer

p5pRT commented 23 years ago

From [Unknown Contact. See original ticket]

On Wed\, 17 Jan 2001\, Peter Prymmer wrote​:

And it could very well be that a bug in configure.com prevents the $ subversion="0" assignment from kicking in.

There was a bug but my speculation was not on the mark. Our read patchlevel.h conditional would quit reading before it necessarily found the various API macros was the problem.

Perhaps another thing worth testing is what munchconfig.exe and descrip.mms would do with​:

PERL_SUBVERSION=0

Sans 'quotes' munchconfig and mms seem pretty happy\, will mmk be happy too?

I'll try to look into it.

And here is the patch that supercedes Craig's (MMK users and external module builders of either the mms or mmk persuasion may want to test this out). POSIX compliant platform folks may want to laugh at the verbosity of DCL since all of this just replaces the wee little awk one liner from the Bourne-ish Configure​:

$test -f $src/patchlevel.h && \ awk '/^#define[ ]+PERL_/ {printf "%s=%s\n"\,$2\,$3}' $src/patchlevel.h >>config.sh

OK here it is w.r.t. 8453​:

Inline Patch ```diff --- configure.com.orig Wed Jan 17 12:08:15 2001 +++ configure.com Wed Jan 17 12:44:15 2001 @@ -1224,12 +1224,20 @@ $ api_subversion = F$ELEMENT(2," ",line) $ got_api_subversion = "true" $ ENDIF -$ IF (.NOT.got_patch).OR.(.NOT.got_sub) THEN GOTO Patchlevel_h_loop +$ IF (.NOT. got_patch) .OR. - + (.NOT. got_sub) .OR. - + (.NOT. got_api_revision) .OR. - + (.NOT. got_api_version) .OR. - + (.NOT. got_api_subversion) - + THEN GOTO Patchlevel_h_loop $Close_patch: $ CLOSE CONFIG $ ELSE $ patchlevel="0" $ subversion="0" +$ api_revision="0" +$ api_version="0" +$ api_subversion="0" $ ENDIF $ IF (F$STRING(subversion) .NES. "0") $ THEN @@ -4835,8 +4843,11 @@ $ WC "CONFIG='true'" $ WC "Makefile_SH='" + Makefile_SH + "'" $ WC "Mcc='" + Mcc + "'" -$! WC "PERL_SUBVERSION='" + subversion + "'" ! VMS specific to descrip_mms.template -$ WC "PERL_VERSION='" + patchlevel + "'" ! VMS specific to descrip_mms.template +$ WC "PERL_REVISION=" + revision +$ WC "PERL_VERSION=" + patchlevel +$ WC "PERL_SUBVERSION=" + subversion +$ WC "PERL_API_VERSION=" + api_version +$ WC "PERL_API_SUBVERSION=" + api_subversion $ WC "alignbytes='" + alignbytes + "'" $ WC "aphostname='" + "'" $ WC "ar='" + "'" ```

End of Patch.

Peter Prymmer

p5pRT commented 23 years ago

From @craigberry

At 01​:07 PM 1/17/2001 -0800\, Peter Prymmer wrote​:

And here is the patch that supercedes Craig's (MMK users and external module builders of either the mms or mmk persuasion may want to test this out).

Thanks Peter. Looks good to me when built with MMK​:

$ perl -"V" Summary of my perl5 (revision 5.0 version 7 subversion 0) configuration​:   Platform​:   osname=VMS\, osvers=V7.1\, archname=VMS_AXP

$ perl -"MConfig" -e "print $Config{PERL_SUBVERSION};" 0

p5pRT commented 23 years ago

From [Unknown Contact. See original ticket]

On Wed\, 17 Jan 2001\, Craig A. Berry wrote​:

$ perl -"V" Summary of my perl5 (revision 5.0 version 7 subversion 0) configuration​: Platform​: osname=VMS\, osvers=V7.1\, archname=VMS_AXP

$ perl -"MConfig" -e "print $Config{PERL_SUBVERSION};" 0

Thank you\, here is another test I've done with my MMS build​:

$ perl "-V​:PERL_.*" PERL_API_SUBVERSION='0' PERL_API_VERSION='5' PERL_REVISION='5' PERL_SUBVERSION='0' PERL_VERSION='7'

where the quoting is done by Config.pm no doubt.

Peter Prymmer