5.8.8 RC1
Nicholas Clark
See to it!
Stiletto: Si, Barone. Subito, Barone.
From "Dangermouse", by Cosgrove Hall Productions.
Currently it's at
http://www.ccl4.org/~nick/P/perl-5.8.8-RC1.tar.bz2
Please don't publicise that URL outside p5p - instead
ftp://ftp.cpan.org/pub/CPAN/authors/id/N/NW/NWCLARK/perl-5.8.8-RC1.tar.bz2
once it's had time to propagate to CPAN's mirrors.
Nicholas Clark
H.Merijn Brand
> Greenback: And the world is mine, all mine. Muhahahahaha.
> See to it!
> Stiletto: Si, Barone. Subito, Barone.
>
> From "Dangermouse", by Cosgrove Hall Productions.
>
> Currently it's at
>
> http://www.ccl4.org/~nick/P/perl-5.8.8-RC1.tar.bz2
DOR: http://home.hccnet.nl/h.m.brand/dor-588-rc1.diff
> Please don't publicise that URL outside p5p - instead
>
> ftp://ftp.cpan.org/pub/CPAN/authors/id/N/NW/NWCLARK/perl-5.8.8-RC1.tar.bz2
DOR: ftp://ftp.cpan.org/pub/CPAN/authors/id/H/HM/HMBRAND/dor-588-rc1.diff
> once it's had time to propagate to CPAN's mirrors.
--
H.Merijn Brand Amsterdam Perl Mongers (http://amsterdam.pm.org/)
using Perl 5.6.2, 5.8.0, 5.8.5, & 5.9.2 on HP-UX 10.20, 11.00 & 11.11,
AIX 4.3 & 5.2, SuSE 9.2 & 9.3, and Cygwin. http://www.cmve.net/~merijn
Smoking perl: http://www.test-smoke.org, perl QA: http://qa.perl.org
reports to: smokers...@perl.org, per...@perl.org
Jason Vas Dias
I can report that the latest perl-5.8.8-RC1 tarball builds fine
and all tests pass under Fedora Core 5 test 2 / Rawhide on all
supported architectures ( i386, x86_64, ia64, ppc, ppc64, s390, s390x )
(running: Linux kernel-2.6.15-1.1863_FC5, gcc 4.1.0 20060117, glibc 2.3.90-30) .
Binary RPMs for all architectures and the source RPM are available for
testing from:
http://people.redhat.com/~jvdias/perl/FC-5/5.8.8-0_RC1
When 5.8.8 is officially released, I'll build it into Fedora Core 5.
Thanks & Regards,
Jason Vas Dias<jvd...@redhat.com>,
perl package maintainer
Red Hat, Inc.
Nicholas Clark
> Good day -
>
> I can report that the latest perl-5.8.8-RC1 tarball builds fine
> and all tests pass under Fedora Core 5 test 2 / Rawhide on all
> supported architectures ( i386, x86_64, ia64, ppc, ppc64, s390, s390x )
> (running: Linux kernel-2.6.15-1.1863_FC5, gcc 4.1.0 20060117, glibc 2.3.90-30) .
Nice. Thanks for the report.
Nicholas Clark
Gisle Aas
options" ugly. It does not look like an option I should consider
turning off either; it's just some fact about how my libc behaves.
Does it have to be there?
It was introduced by change 25840 because of some Windows build issue.
$ perl -V
...
Characteristics of this binary (from libperl):
Compile-time options: PERL_MALLOC_WRAP SPRINTF_RETURNS_STRLEN
USE_LARGE_FILES USE_PERLIO
Locally applied patches:
RC1
Gisle Aas
> If SPRINTF_RETURNS_STRLEN were not listed in this output then
> makedef.pl would write a perldll.def file on Win32 specifying that
> Perl_my_sprintf should be exported from the DLL, which will cause a
> linker error since no such symbol is defined (given that
> SPRINTF_RETURNS_STRLEN is #define'd on Win32).
But perl-5.8.8 doesn't have Perl_my_sprintf so there shouldn't be any
reason for this output in that perl.
--Gisle
Vadim Konovalov
>
> As of change 25209, makedef.pl reads the output of "miniperl
> -V" to pick
> up various build options (including SPRINTF_RETURNS_STRLEN) which it
> otherwise wouldn't know about.
>
> If SPRINTF_RETURNS_STRLEN were not listed in this output then
> makedef.pl
> would write a perldll.def file on Win32 specifying that
> Perl_my_sprintf
> should be exported from the DLL, which will cause a linker
> error since
> no such symbol is defined (given that SPRINTF_RETURNS_STRLEN is
> #define'd on Win32).
>
> things listed
> by "-V", though, so perhaps it should be handled in some other way.
I beleive those should be in win32/config.* files; yet following lines in
makedef.pl, which serve purposes of providing SPRINTF_RETURNS_STRLEN for a
build process, and ask "miniperl -V" too early, should also be removed:
if ($PLATFORM eq 'win32' or $PLATFORM eq 'wince' or $PLATFORM eq "aix") {
# Add the compile-time options that miniperl was built with to
%define.
# On Win32 these are not the same options as perl itself will be
built
# with since miniperl is built with a canned config (one of the
win32/
# config_H.*) and none of the BUILDOPT's that are set in the
makefiles,
# but they do include some #define's that are hard-coded in various
# source files and header files and don't include any BUILDOPT's
that
# the user might have chosen to disable because the canned configs
are
# minimal configs that don't include any of those options.
my $opts = ($PLATFORM eq 'wince' ? '-MCross' : ''); # for wince need
Cross.pm to get Config.pm
my $config = `$^X $opts -Ilib -V`;
my($options) = $config =~ /^ Compile-time options: (.*?)\n^ \S/ms;
$options =~ s/\s+/ /g;
print STDERR "Options: ($options)\n";
foreach (split /\s+/, $options) {
$define{$_} = 1;
}
}
I remember I adjusted those for WinCE build namely for that
SPRINTF_RETURNS_STRLEN, which isn't a way to go.
Best regards,
Vadim.
Steve Hay
Oh. Yes, you're right. Looks like (at least some of) the various config
file changes that #define SPRINTF_RETURNS_STRLEN have been integrated,
but Perl_my_sprintf etc itself hasn't been.
So yes, SPRINTF_RETURNS_STRLEN could indeed be removed from the "-V"
output in maint.
------------------------------------------------
Radan Computational Ltd.
The information contained in this message and any files transmitted with it are confidential and intended for the addressee(s) only. If you have received this message in error or there are any problems, please notify the sender immediately. The unauthorized use, disclosure, copying or alteration of this message is strictly forbidden. Note that any views or opinions presented in this email are solely those of the author and do not necessarily represent those of Radan Computational Ltd. The recipient(s) of this message should check it and any attached files for viruses: Radan Computational will accept no liability for any damage caused by any virus transmitted by this email.
Nicholas Clark
> Gisle Aas wrote:
> >Steve Hay <stev...@uk.radan.com> writes:
> >
> >
> >>If SPRINTF_RETURNS_STRLEN were not listed in this output then
> >>makedef.pl would write a perldll.def file on Win32 specifying that
> >>Perl_my_sprintf should be exported from the DLL, which will cause a
> >>linker error since no such symbol is defined (given that
> >>SPRINTF_RETURNS_STRLEN is #define'd on Win32).
> >
> >
> >But perl-5.8.8 doesn't have Perl_my_sprintf so there shouldn't be any
> >reason for this output in that perl.
>
> Oh. Yes, you're right. Looks like (at least some of) the various config
> file changes that #define SPRINTF_RETURNS_STRLEN have been integrated,
> but Perl_my_sprintf etc itself hasn't been.
>
> So yes, SPRINTF_RETURNS_STRLEN could indeed be removed from the "-V"
> output in maint.
Until Perl_my_sprintf gets integrated?
And then we're back where we started?
Nicholas Clark
Jan Dubois
> As of change 25209, makedef.pl reads the output of "miniperl -V" to pick
> up various build options (including SPRINTF_RETURNS_STRLEN) which it
> otherwise wouldn't know about.
>
> would write a perldll.def file on Win32 specifying that Perl_my_sprintf
> should be exported from the DLL, which will cause a linker error since
> no such symbol is defined (given that SPRINTF_RETURNS_STRLEN is
> #define'd on Win32).
It seems like *all* supported compilers for Win32 define this symbol.
Therefore I don't see why you don't just add Perl_my_sprintf to the
list of symbols always skipped on Windows. There is already a big
static section of symbols skipped that starts with this:
if ($PLATFORM eq 'win32') {
skip_symbols [qw(
PL_statusvalue_vms
...
Cheers,
-Jan
Michael Schroeder
> Currently it's at
>
> http://www.ccl4.org/~nick/P/perl-5.8.8-RC1.tar.bz2
Compiles with all tests passing for the following SuSE releases:
Factory (i.e. 10.1/sles10 beta1): gcc-4.1.0_20060109, glibc-2.3.90
i586, x86_64, ia64, ppc, s390, s390x
10.0: gcc-4.0.2_20050901, glibc-2.3.5
i586, x86_64, ppc, ppc64
9.3: gcc-3.3.5, glibc-2.3.4
i386, x86_64
9.2: gcc-3.3.4, glibc-2.3.3
i386, x86_64
9.1/sles9: gcc-3.3.3, glibc-2.3.3
i386, x86_64, ia64, ppc, ppc64, s390, s390x
Some comments:
- the build runs 'miniperl -I../../lib mktables -w' seven times.
- there are two "comparison is always false due to limited range of
data type" warnings:
regcomp.c:723, the Move() macro
byterun.c:899, somewhere in the BSET_newsv() macro
- there are some "ignoring return value of <function>, declared with
attribute warn_unused_result" warnings.
perl.c:224 write in INIT_TLS_AND_INTERP
util.c:87,136,189,820 write of no_mem messages
sv.c:3005,3222,8552,9315 gcvt
sv.c:5222 Perl_io_close in Perl_sv_clear()
doio.c:914 fchown() in Perl_nextargv
doio.c:1496,1641 write of report after exec
perlio.c:479 PerlIO_debug write
perlio.c:649 fp_dup
Cwd.xs:202 fchdir in bsd_realpath()
- one "'S_emulate_eaccess' declared 'static' but never defined"
warning for pp_sys.c
- some "'<var>' may be used uninitialized" warnings
B.c: 3318 XS_B__MAGIC_precomp: RETVAL
Dumper.xs:529 DD_dump: entry
Dumper.xs:936 XS_Data__Dumper_Dumpxs: sortkeys
DProf.xs:490 boot_Devel__DProf: s,u,r
encoding.xs:258 PerlIOEncode_fill: s
Normalize.c:715 XS_Unicode__Normalize_isComp2nd: ix
Normalize.xs:519 XS_Unicode__Normalize_checkFCD: canlen
Normalize.xs:341 XS_Unicode__Normalize_compose: uv,uvS
Normalize.xs:252 XS_Unicode__Normalize_reorder: uvlast
Cheers,
Michael.
--
Michael Schroeder mlsc...@informatik.uni-erlangen.de
main(_){while(_=~getchar())putchar(~_-1/(~(_|32)/13*2-11)*13);}
Nicholas Clark
> - the build runs 'miniperl -I../../lib mktables -w' seven times.
Thanks, yes. It doesn't do this with FreeBSD make. However it does do it
with GNU make on FreeBSD. The culprit seemed to be me not integrating a
patch from Jarkko to Makefile.SH. His patch was intended to solve problems
with parallel builds, but at the time didn't seem to fully help, so I never
integrated it. I think subsequently removing mv-if-diff made his patch more
important. It's now integrated with change 26938.
Thanks for the other reported warnings. Most look like they are warnings that
are also in blead, and I'm not sure how to avoid those.
Nicholas Clark
H.Merijn Brand
> Rafael Garcia-Suarez wrote:
> > On 1/24/06, Steve Hay <stev...@uk.radan.com> wrote:
> >
> >>>It seems like *all* supported compilers for Win32 define this symbol.
> >>>
> >>>Therefore I don't see why you don't just add Perl_my_sprintf to the
> >>>list of symbols always skipped on Windows. There is already a big
> >>>static section of symbols skipped that starts with this:
> >>>
> >>> if ($PLATFORM eq 'win32') {
> >>> skip_symbols [qw(
> >>> PL_statusvalue_vms
> >>> ...
> >>
> >>Very true. The attached patch works OK on Win32, and removes the symbol
> >>from the "-V" output.
> >>
> >>Is the following chunk from further down makedef.pl:
> >>
> >>if ($define{'SPRINTF_RETURNS_STRLEN'}) {
> >> skip_symbols [qw(
> >> Perl_my_sprintf
> >> )];
> >>}
> >>
> >>still required for other OS's?
> >
> >
> > I don't know.
>
> I'll play it safe and leave it for now then. Perhaps somebody with
> another OS that uses makedef.pl could shed some light? e.g. Is it
> required on AIX? Merijn?
All tests successful (5 subtests UNEXPECTEDLY SUCCEEDED), 61 tests and 277 subtests skipped.
Files=1194, Tests=160902, 895 wallclock secs (356.91 cusr + 152.28 csys = 509.19 CPU)
i4:/pro/3gl/CPAN/perl-current 115 > ./perl -Ilib -V
Summary of my perl5 (revision 5 version 9 subversion 3 patch 26936) configuration:
Platform:
osname=aix, osvers=5.2.0.0, archname=aix-64all
uname='aix i4 2 5 0056a8fd4c00 '
config_args='-Dusedevel -Duse64bitall -des'
hint=recommended, useposix=true, d_sigaction=define
useithreads=undef, usemultiplicity=undef
useperlio=define, d_sfio=undef, uselargefiles=define, usesocks=undef
use64bitint=define, use64bitall=define, uselongdouble=undef
usemymalloc=n, bincompat5005=undef
Compiler:
cc='xlc', ccflags ='-q64 -DDEBUGGING -qlanglvl=extended -D_ALL_SOURCE -D_ANSI_C_SOURCE -D_POSIX_SOURCE -qmaxmem=-1 -qnoansialias -DUSE_NATIVE_DLOPEN -I/pro/local/include -q64 -DUSE_64_BIT_ALL -q64',
optimize='-O',
cppflags='-q64 -DDEBUGGING -qlanglvl=extended -D_ALL_SOURCE -D_ANSI_C_SOURCE -D_POSIX_SOURCE -qmaxmem=-1 -qnoansialias -DUSE_NATIVE_DLOPEN -I/pro/local/include'
ccversion='7.0.0.3', gccversion='', gccosandvers=''
intsize=4, longsize=8, ptrsize=8, doublesize=8, byteorder=87654321
d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=8
ivtype='long', ivsize=8, nvtype='double', nvsize=8, Off_t='off_t', lseeksize=8
alignbytes=8, prototype=define
Linker and Libraries:
ld='ld', ldflags ='-L/usr/local/ppc64/lib64 -b64 -q64 -L/pro/local/lib -brtl -bdynamic -bmaxdata:0x80000000 -q64 -b64'
libpth=/usr/local/ppc64/lib64 /lib /usr/lib /usr/ccs/lib /usr/local/lib
libs=-lbind -lnsl -ldbm -ldb -ldl -lld -lm -lcrypt -lc -lbsd
perllibs=-lbind -lnsl -ldl -lld -lm -lcrypt -lc -lbsd
libc=/lib/libc.a, so=a, useshrplib=false, libperl=libperl.a
gnulibc_version=''
Dynamic Linking:
dlsrc=dl_aix.xs, dlext=so, d_dlsymun=undef, ccdlflags=' -bE:/pro/lib/perl5/5.9.3/aix-64all/CORE/perl.exp'
cccdlflags=' ', lddlflags='-b64 -bhalt:4 -bexpall -G -bnoentry -lc -L/usr/local/ppc64/lib64 -L/pro/local/lib'
Characteristics of this binary (from libperl):
Compile-time options: DEBUGGING PERL_DONT_CREATE_GVSV PERL_MALLOC_WRAP
USE_64_BIT_ALL USE_64_BIT_INT USE_LARGE_FILES
USE_PERLIO
Built under aix
Compiled at Jan 24 2006 15:02:15
%ENV:
PERL5_CPANPLUS_CONFIG="/u/usr/merijn/.cpanplus/config"
@INC:
lib
/pro/lib/perl5/5.9.3/aix-64all
/pro/lib/perl5/5.9.3
/pro/lib/perl5/site_perl/5.9.3/aix-64all
/pro/lib/perl5/site_perl/5.9.3
/pro/lib/perl5/site_perl
.
i4:/pro/3gl/CPAN/perl-current 116 >
Vadim Konovalov
>
> I made it skip on WinCE too, since that also defines
> SPRINTF_RETURNS_STRLEN in its canned config. Vadim, could
> you test that
> this still works OK?
Works ok, thanks!
Vadim.
H.Merijn Brand
> H.Merijn Brand wrote:
> > On Tue, 24 Jan 2006 09:25:54 +0000, Steve Hay <stev...@uk.radan.com>
> > wrote:
> >
> >
> >>Rafael Garcia-Suarez wrote:
> >>
> >>>On 1/24/06, Steve Hay <stev...@uk.radan.com> wrote:
> >>>
> >>>
> >>>>Is the following chunk from further down makedef.pl:
> >>>>
> >>>>if ($define{'SPRINTF_RETURNS_STRLEN'}) {
> >>>> skip_symbols [qw(
> >>>> Perl_my_sprintf
> >>>> )];
> >>>>}
> >>>>
> >>>>still required for other OS's?
> >>>
> >>>
> >>>I don't know.
> >>
> >>I'll play it safe and leave it for now then. Perhaps somebody with
> >>another OS that uses makedef.pl could shed some light? e.g. Is it
> >>required on AIX? Merijn?
> >
> >
> > All tests successful (5 subtests UNEXPECTEDLY SUCCEEDED), 61 tests and
> > 277 subtests skipped.
>
> Thanks, but was that with or without the chunk of makedef.pl shown above?
>
> I haven't removed it yet, and was wondering whether I could remove it or
> whether I should leave it. It isn't required on Win32 any more, but I'm
> not sure if AIX still needs it.
It was from rsync at 26936, and as I read that you applied it, I guess that
it is in, as your patch was applied in 26935
Demerphq
> On Mon, Jan 23, 2006 at 02:41:30PM +0100, Michael Schroeder wrote:
>
> > - the build runs 'miniperl -I../../lib mktables -w' seven times.
>
> Thanks, yes. It doesn't do this with FreeBSD make. However it does do it
> with GNU make on FreeBSD. The culprit seemed to be me not integrating a
> patch from Jarkko to Makefile.SH. His patch was intended to solve problems
> with parallel builds, but at the time didn't seem to fully help, so I never
> integrated it. I think subsequently removing mv-if-diff made his patch more
> important. It's now integrated with change 26938.
Did the mktables patch i put together some time back get applied to
maint? (The one that makes the other six times (close to) a noop)
yves
--
perl -Mre=debug -e "/just|another|perl|hacker/"
Steve Hay
You mean 24056? If so, then yes. Looks like it was in 5.8.7, in fact.
H.Merijn Brand
> H.Merijn Brand wrote:
> > On Tue, 24 Jan 2006 14:54:28 +0000, Steve Hay <stev...@uk.radan.com>
> >>H.Merijn Brand wrote:
> >>>On Tue, 24 Jan 2006 09:25:54 +0000, Steve Hay <stev...@uk.radan.com>
> >>>>Rafael Garcia-Suarez wrote:
> >>>>>On 1/24/06, Steve Hay <stev...@uk.radan.com> wrote:
> >>>>>
> >>>>>>Is the following chunk from further down makedef.pl:
> >>>>>>
> >>>>>>if ($define{'SPRINTF_RETURNS_STRLEN'}) {
> >>>>>> skip_symbols [qw(
> >>>>>> Perl_my_sprintf
> >>>>>> )];
> >>>>>>}
> >>>>>>
> >>>>>>still required for other OS's?
> >>>>>
> >>>>>I don't know.
> >>>>
> >>>>I'll play it safe and leave it for now then. Perhaps somebody with
> >>>>another OS that uses makedef.pl could shed some light? e.g. Is it
> >>>>required on AIX? Merijn?
> >>>
> >>>All tests successful (5 subtests UNEXPECTEDLY SUCCEEDED), 61 tests and
> >>>277 subtests skipped.
> >>
> >>Thanks, but was that with or without the chunk of makedef.pl shown above?
> >>
> >>I haven't removed it yet, and was wondering whether I could remove it or
> >>whether I should leave it. It isn't required on Win32 any more, but I'm
> >>not sure if AIX still needs it.
> >
> >
> > It was from rsync at 26936, and as I read that you applied it, I guess
> > that it is in, as your patch was applied in 26935
>
> Change 26935 did some other stuff but didn't remove the chunk above!
> Sorry if I wasn't clear.
>
> Would you be able to edit your makedef.pl to remove the chunk above and
> then try a rebuild to see if it still works? I think it might not be
> necessary any more.
# vi makedef.pl
# rm perl.exp
# make test_harness
All tests successful.
u=2.28 s=2.65 cu=373.82 cs=75.13 scripts=1130 tests=160889
Nicholas Clark
> OK. I assume that SPRINTF_RETURNS_STRLEN is #define'd on your AIX?
>
> If so, then I'm puzzled how this worked, really. Without the bit of
> makedef.pl concerned, the symbol Perl_my_sprintf is emitted to whatever
> the AIX equivalent of a DLL export definition file is, and yet the
> function itself isn't actually defined. Is that not considered an error?
>
> On Win32 that causes a linker error, hence the need to skip the symbol.
IIRC from my experiences of pain, er, AIX, it was only a warning, not an error.
However, this might have been for the inverse case, using a symbol that wasn't
listed for export.
Nicholas Clark
H.Merijn Brand
> OK. I assume that SPRINTF_RETURNS_STRLEN is #define'd on your AIX?
i4:/pro/3gl/CPAN/perl-current 102 > grep -i sprintf config.sh
d_sprintf_returns_strlen='define'
i4:/pro/3gl/CPAN/perl-current 103 >
> If so, then I'm puzzled how this worked, really. Without the bit of
> makedef.pl concerned, the symbol Perl_my_sprintf is emitted to whatever
> the AIX equivalent of a DLL export definition file is, and yet the
> function itself isn't actually defined. Is that not considered an error?
>
> On Win32 that causes a linker error, hence the need to skip the symbol.
--