search

Perl / perl5

🐪 The Perl programming language
https://dev.perl.org/perl5/
Other
1.91k stars 542 forks source link

perl5.6-RC1 not OK - modules complain. #1344

Closed p5pRT closed 20 years ago

p5pRT commented 24 years ago

Migrated from rt.perl.org#2379 (status was 'resolved')

Searchable as RT2379$

p5pRT commented 24 years ago

From ht.000@foa.dk

Created by ht000@foa.dk

Build of perl5.6-RC1 completes without failure. The tests and installation likewise.

The problems come when I try to install modules.

Bundle​::CPAN has a small glitch in CPAN​::WAIT​:

.... mkdir blib/man3 cp lib/CPAN/WAIT.pm blib/lib/CPAN/WAIT.pm cp lib/WAIT/Client.pm blib/lib/WAIT/Client.pm Manifying blib/man3/CPAN​::WAIT.3   /sbin/make -- OK Running make test PERL_DL_NONLAZY=1 /tools/perl5.6/bin/perl -Iblib/arch -Iblib/lib -I/tools/perl5. 6/lib/5.6.0/alpha-dec_osf -I/tools/perl5.6/lib/5.6.0 -e 'use Test​::Harness qw(&r untests $verbose); $verbose=0; runtests @​ARGV;' t/*.t t/basic.............Use of uninitialized value in string eq at /tools/perl5.6/li b/5.6.0/ExtUtils/MM_Unix.pm line 110. t/basic.............ok All tests successful. Files=1\, Tests=5\, 2 wallclock secs ( 0.77 cusr + 0.25 csys = 1.02 CPU)   /sbin/make test -- OK Running make install Installing /tools/perl5.6/lib/site_perl/5.6.0/WAIT/Client.pm Installing /tools/perl5.6/lib/site_perl/5.6.0/CPAN/WAIT.pm Installing /tools/perl5.6/man/man3/CPAN​::WAIT.3 Writing /tools/perl5.6/lib/site_perl/5.6.0/alpha-dec_osf/auto/CPAN/WAIT/.packlis ....

Net​::Daemon (a part of Bundle​::DBI) fails miserably​:

.... cp lib/Net/Daemon.pm blib/lib/Net/Daemon.pm b/Net/Daemon/Test.pm blib/lib/Net/Daemon/Test.pm Manifying blib/man3/Net​::Daemon​::Log.3 Manifying blib/man3/Net​::Daemon.3 Manifying blib/man3/Net​::Daemon​::Test.3   /sbin/make -- OK Running make test PERL_DL_NONLAZY=1 /tools/perl5.6/bin/perl -Iblib/arch -Iblib/lib -I/tools/perl5. 6/lib/5.6.0/alpha-dec_osf -I/tools/perl5.6/lib/5.6.0 -e 'use Test​::Harness qw(&r untests $verbose); $verbose=0; runtests @​ARGV;' t/*.t t/base..............ok t/config............Your vendor has not defined the Sys​::Syslog macro _PATH_LOG at /tools/perl5.6/lib/5.6.0/alpha-dec_osf/Sys/Syslog.pm line 277. Server process didn't create a file 'ndtest.prt'. at blib/lib/Net/Daemon/Test.pm line 317. t/config............dubious   Test returned status 2 (wstat 512\, 0x200) DIED. FAILED tests 1-5   Failed 5/5 tests\, 0.00% okay t/fork..............Your vendor has not defined the Sys​::Syslog macro _PATH_LOG at /tools/perl5.6/lib/5.6.0/alpha-dec_osf/Sys/Syslog.pm line 277. ....

After a *LOT* of fiddling I can get as far as getting most of the "usual" modules installed\, but always there are many complaints along the way. Most use of Sys​::Syslog works - so Net​::Daemon must do something special in 5.6-RC1\, in 5.005_03 there is no error.

I have tried to reinstall several times on different machines but nowhere have I got something that is nearly as stable as 5.005_03.

Henrik Tougaard\, htoug@​cpan.org

Perl Info ``` This perlbug was built using Perl v5..5.670 - Thu Mar 9 12:33:21 MET 2000 It is being executed now by Perl v5.6.0 - Tue Mar 14 09:30:53 MET 2000. Site configuration information for perl v5.6.0: Configured by ht000 at Tue Mar 14 09:30:53 MET 2000. Summary of my perl5 (revision 5.0 version 6 subversion 0) configuration: Platform: osname=dec_osf, osvers=4.0, archname=alpha-dec_osf uname='osf1 panama2.foa.dk v4.0 1091 alpha ' config_args='' hint=recommended, useposix=true, d_sigaction=define usethreads=undef use5005threads=undef useithreads=undef usemultiplicity=undef useperlio=undef d_sfio=undef uselargefiles=define use64bitint=define use64bitall=define uselongdouble=undef usesocks=undef Compiler: cc='cc', optimize='-O4', gccversion= cppflags='-std -ieee -D_INTRINSICS -I/usr/local/include -DLANGUAGE_C' ccflags ='-std -fprm d -ieee -D_INTRINSICS -I/usr/local/include -DLANGUAGE_C' stdchar='unsigned char', d_stdstdio=define, usevfork=false intsize=4, longsize=8, ptrsize=8, doublesize=8 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, usemymalloc=y, prototype=define Linker and Libraries: ld='ld', ldflags =' -L/usr/local/lib' libpth=/usr/local/lib /usr/shlib /usr/ccs/lib /usr/lib/cmplrs/cc /usr/lib /var/shlib libs=-ldbm -ldb -lm -liconv libc=/usr/shlib/libc.so, so=so, useshrplib=true, libperl=libperl.so Dynamic Linking: dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags=' -Wl,-rpath,/tools/perl5.6/lib/5.6.0/alpha-dec_osf/CORE' cccdlflags=' ', lddlflags='-shared -expect_unresolved "*" -O4 -msym -std -s -L/usr/local/lib' Locally applied patches: @INC for perl v5.6.0: /usr/local/lib/perlmod /tools/perl5.6/lib/5.6.0/alpha-dec_osf /tools/perl5.6/lib/5.6.0 /tools/perl5.6/lib/site_perl/5.6.0/alpha-dec_osf /tools/perl5.6/lib/site_perl/5.6.0 /tools/perl5.6/lib/site_perl . Environment for perl v5.6.0: HOME=/usr/users/ht LANG=da_DK.ISO8859-1 LANGUAGE (unset) LD_LIBRARY_PATH=/ingres_oa/ingres/lib:/usr/shlib LOGDIR (unset) PATH=/tools/perl5.6/bin:/usr/local/scripts:/usr/local/bin:/usr/local:/sbin:/usr/sbin:/usr/bin:/usr/bin/X11:/usr/opt/networker/bin:/progs/CVS/cvs-1.10/bin:/usr/users/ht/bin:/ingres_oa/ingres/utility:/ingres_oa/ingres/bin PERL5LIB=/usr/local/lib/perlmod PERL_BADLANG (unset) SHELL=/usr/bin/ksh ```
p5pRT commented 24 years ago

From @gsar

On Wed\, 15 Mar 2000 10​:37​:38 +0100\, Henrik Tougaard wrote​:

The problems come when I try to install modules.

We're up to RC2\, in case you haven't noticed.

Bundle​::CPAN has a small glitch in CPAN​::WAIT​:

.... mkdir blib/man3 cp lib/CPAN/WAIT.pm blib/lib/CPAN/WAIT.pm cp lib/WAIT/Client.pm blib/lib/WAIT/Client.pm Manifying blib/man3/CPAN​::WAIT.3 /sbin/make -- OK Running make test PERL_DL_NONLAZY=1 /tools/perl5.6/bin/perl -Iblib/arch -Iblib/lib -I/tools/perl5. 6/lib/5.6.0/alpha-dec_osf -I/tools/perl5.6/lib/5.6.0 -e 'use Test​::Harness qw(&r untests $verbose); $verbose=0; runtests @​ARGV;' t/*.t t/basic.............Use of uninitialized value in string eq at /tools/perl5.6/li b/5.6.0/ExtUtils/MM_Unix.pm line 110. t/basic.............ok All tests successful. Files=1\, Tests=5\, 2 wallclock secs ( 0.77 cusr + 0.25 csys = 1.02 CPU) /sbin/make test -- OK

Other than the warning\, I don't see anything wrong here.

Net​::Daemon (a part of Bundle​::DBI) fails miserably​:

.... cp lib/Net/Daemon.pm blib/lib/Net/Daemon.pm b/Net/Daemon/Test.pm blib/lib/Net/Daemon/Test.pm Manifying blib/man3/Net​::Daemon​::Log.3 Manifying blib/man3/Net​::Daemon.3 Manifying blib/man3/Net​::Daemon​::Test.3 /sbin/make -- OK Running make test PERL_DL_NONLAZY=1 /tools/perl5.6/bin/perl -Iblib/arch -Iblib/lib -I/tools/perl5. 6/lib/5.6.0/alpha-dec_osf -I/tools/perl5.6/lib/5.6.0 -e 'use Test​::Harness qw(&r untests $verbose); $verbose=0; runtests @​ARGV;' t/*.t t/base..............ok t/config............Your vendor has not defined the Sys​::Syslog macro _PATH_LOG at /tools/perl5.6/lib/5.6.0/alpha-dec_osf/Sys/Syslog.pm line 277. Server process didn't create a file 'ndtest.prt'. at blib/lib/Net/Daemon/Test.pm line 317. t/config............dubious Test returned status 2 (wstat 512\, 0x200) DIED. FAILED tests 1-5 Failed 5/5 tests\, 0.00% okay t/fork..............Your vendor has not defined the Sys​::Syslog macro _PATH_LOG at /tools/perl5.6/lib/5.6.0/alpha-dec_osf/Sys/Syslog.pm line 277.

These may have been silent errors earlier (Sys​::Syslog is more verbose now about failed attempts to find constants).

After a *LOT* of fiddling I can get as far as getting most of the "usual" modules installed\,

If you're doing a lot of fiddling\, you're doing something wrong. (Perhaps you don't know about "perl Makefile.PL POLLUTE=1".)

              but always there are many complaints along the way\.

Most use of Sys​::Syslog works - so Net​::Daemon must do something special in 5.6-RC1\, in 5.005_03 there is no error.

I have tried to reinstall several times on different machines but nowhere have I got something that is nearly as stable as 5.005_03.

Just for the record\, ActiveState has built a large number of extensions from CPAN on Solaris\, Linux and Windows. The Windows builds have needed fixups for a couple of modules that used undocumented internals (like DBI) to pass the extra context parameter\, but all the rest of them build and test with zero changes.

Sarathy gsar@​ActiveState.com

p5pRT commented 24 years ago

From [Unknown Contact. See original ticket]

From​: Gurusamy Sarathy [mailto​:gsar@​ActiveState.com]

Just for the record\, ActiveState has built a large number of extensions from CPAN on Solaris\, Linux and Windows. The Windows builds have needed fixups for a couple of modules that used undocumented internals (like DBI) to pass the extra context parameter\, but all the rest of them build and test with zero changes.

FWIW\, here's a (partial) build report on stuff I did last night (RC2\, Win32\, MSVC\, core build -- I hope to do a fork-emulation build soon).

Some points - "patch available" here means that I managed to patch it. I've not necessarily passed the patch back to the maintainer(s) yet. Many of the test failures have been around for ages - the "new" ones (as far as I recall) are Bit-Vector\, Filter\, Math-Fraction\, ParseLex and String-Approx. Also\, there are definitely a quite few modules which still need POLLUTE=1.

The nasties are

1. DBI and DBD-Oracle - although there are patches   so hopefully a new release will fix these. 2. Tk - I only hit this last night - Tk800.018 built   with patches\, so hopefully this will get sorted soon. 3. Archive-Tar - this seems unrelated to the core\, so it's   not a release issue.

Also\, last thing last night\, mod_perl failed to build (link errors on _Perl_cv_undef and _Perl_cv_const_sv) but I haven't had any time to check this yet.

I have to agree with whoever it was who commented that releasing 5.6.0 with key modules on CPAN not building in the latest version\, would be a problem. At the very least\, the release notes for the core should explicitly note that there are build issues with things like DBI-1.13\, expected to be fixed in the next release. Preferably including an "unofficial" patch in the Perl release\, if Tim can't get a new DBI release out soon enough (Not meaning to pick on Tim\, thye same applies to other "big" modules). Perlbug messages along the lines of "perl 5.6 broke DBI" wouldn't be nice...

Just my 2p worth\, Paul.

Module Build Report for Perl 5.6.0-RC2 (core build)

Notation   Y​: succeeded   P​: needs POLLUTE=1   T​: test failures (believed to be benign)   ?​: test failures (status unknown)   U​: use of uninitialised value warning   FP​: failed - patch available   F​: failed - no patch available

Module Version Notes ----------------------- ------- ----- Compress-Zlib 1.08 Y DB_File 1.72 Y libwin32 0.151 Y   APIFile P\, ? (tests 46\,48\,50)   NetResource ? (tests 5\,6)   ODBC T (note 1)   PerfLib ? (tests 3)   Shortcut ? (note 2)   Sound P Mime-Base64 2.11 Y URI 1.05 ? (old-base.t 4-8) Tk 800.019 F (note 3) Net-Daemon 0.29 U (t\forkm.t line 18) Storable 0.6.7 P PlRPC 0.2012 Y DBI 1.13 FP DBD-Oracle 1.03 FP\, T (note 4) Bit-ShiftReg 2.0 Y Bit-Vector 5.7 P\, T (note 5) Archive-Tar 0.21 F (note 6) DBD-ODBC 0.27 Y Curses 1.02 P Data-Locations 5.2 Y libnet 1.0702 P XML-Parser 2.27 FP (note 7) Digest-MD5 2.09 Y HTML-Parser 3.05 Y libwww-perl 5.47 Y GD 1.25 Y Algorithm-Diff 0.59 Y Sample 1.3 Y Shuffle 1.4 Y TransitiveClosure 1.4 Y Heap 0.01 Y Graph 0.201 Y IO-stringy 1.207 Y IO-Zlib 0.02 Y MIME-Lite 1.135 Y MailTools 1.13 Y MIME-tools 4.124 Y FreezeThaw 0.3 Y MLDBM 2.00 T (note 8) Alias 2.32 Y Array-PrintCols 2.1 Y Array-Lookup 2.1 T (note 9) Date-Calc 4.3 Y DateManip 5.37 T Filter 1.17 ? (line termination???) Time-HiRes 01.20 Y Scalar-List-Utils 1.00 T (odd msg from taint test) Set-IntRange 5.0 Y Text-CSV_XS 0.20 Y Text-Filter 1.7 Y Text-Format 0.52 Y Win32-DDE 0.02 Y Win32API 0.011 Y Devel-Symdump 2.00 T (extra modules in test 6) Math-Fraction v.53b ? (note 10) Math-Random 0.61 P\, T (test 2\, rounding error) Parse-RecDescent 1.77 Y ParseLex 2.15 ? (test 4) SQL-Statement 0.1016 Y (note 11) DBD-CSV 0.1022 Y (note 12) Tie-DBI 0.85 Y Tie-Dir 1.02 Y Tie-Watch 1.0 Y String-Approx 3.09 ? (t\aslice tests 4-6) String-CRC 1.0 Y Text-Balanced 1.76 Y MIDI-Perl 0.74 Y Data-Flow 0.05 Y Make 1.00 Y

Notes

1. The database file required for the tests is missing 2. The tests fail to resolve a shortcut. Reason unknown. 3. It seems that the panic() function clashes with Perl's. Check... 4. There is an error in the tests​: on cleanup\, cannot call FETCH on an   undefined value in Win32​::TieRegistry 5. The tests check message texts\, which have changed. 6. Suspected to be caused by changes in Compress​::Zlib\, not perl. 7. The generated Expat\Makefile defines O_FILES with the .obj files   in subdirectories. Visual C puts them in the current dir. Fix is   to remove directories from the files in O_FILES. 8. Test 1 fails as Dat​::Dumper is now better at detecting numbers vs strings 9. Differences due to change in hash keys order. 10. Test failures - due to precision issues? 11. tests 50eval and 60hash say "main​::ArrEq() called too early to check   prototype". 12. tests 10dsnlist\, dbdadmin use defined(@​array)

p5pRT commented 24 years ago

From @gsar

On Thu\, 16 Mar 2000 10​:06​:03 GMT\, "Moore\, Paul" wrote​:

The nasties are

All three of which are problems in the extensions themselves... (just to drive home a point :)

1. DBI and DBD-Oracle - although there are patches so hopefully a new release will fix these.

2. Tk - I only hit this last night - Tk800.018 built with patches\, so hopefully this will get sorted soon.

3. Archive-Tar - this seems unrelated to the core\, so it's not a release issue.

Also\, last thing last night\, mod_perl failed to build (link errors on _Perl_cv_undef and _Perl_cv_const_sv) but I haven't had any time to check this yet.

cv_undef() and cv_const_sv() are not in the list of public API functions. I'm not sure why mod_perl needs them\, but it seems like a bad idea to use them. (SvREFCNT_dec() should be used rather than cv_undef()\, because the latter will clobber a CV regardless of how many people own it. And cv_const_sv() is purely internal.)

I have to agree with whoever it was who commented that releasing 5.6.0 with key modules on CPAN not building in the latest version\, would be a problem.

FWIW\, I (mildly) disagree. New releases must happen to effect change in something as big as CPAN--we can't wait around for everyone to "catch up"\, because in my experience\, very few of them will "catch up" unless there's a release out to "create" demand for the changes.

But I do agree that we should make the known issues with existing modules amply clear in the release notes.

Module Build Report for Perl 5.6.0-RC2 (core build)

Thanks for the table.

I'm going to ignore the need for POLLUTE=1 or any new warnings as non-issues.

Notation ?​: test failures (status unknown) FP​: failed - patch available F​: failed - no patch available [...] libwin32 0.151 Y APIFile P\, ? (tests 46\,48\,50) NetResource ? (tests 5\,6) ODBC T (note 1) PerfLib ? (tests 3) Shortcut ? (note 2) Sound P

I have fixes for these (obviously). There should be new libwin32 out when 5.6.0 ships.

URI 1.05 ? (old-base.t 4-8)

IIRC\, that test never worked on Windows\, for some reason.

Tk 800.019 F (note 3)

Problem is in Tk\, apparently.

DBI 1.13 FP

I sent a patch for this.

Bit-Vector 5.7 P\, T (note 5)

Problem in test suite.

Archive-Tar 0.21 F (note 6)

Problem in module.

XML-Parser 2.27 FP (note 7)

Makefile.PL is simply not portable enough.

Filter 1.17 ? (line termination???) Math-Fraction v.53b ? (note 10) Math-Random 0.61 P\, T (test 2\, rounding error) ParseLex 2.15 ? (test 4) String-Approx 3.09 ? (t\aslice tests 4-6)

These need investigation.

For comparison\, I've attached the list of modules that folks at ActiveState have built successfully. (The build on windows uses ithreads for fork()\, while the others are vanilla builds.)

Sarathy gsar@​ActiveState.com

# Format​: # package name|pollute|install|platform1\,testflag\,platform2\,testflag... # # - 'pollute' field should be 1 if the extension requires -DPOLLUTE # - 'make install' will be performed if 'install' is 1 # - platforms list is comma-separated platform name + testflag # 'make test' will be performed if 'testflag' is 1 # # N.B. Packages will be built in the order in which they occur in this file. # Class-Tom|0|1|MSWin32\,1\,linux\,1\,solaris\,1 GD|0|1|MSWin32\,1\,linux\,1 DBI|0|1|MSWin32\,1\,linux\,1\,solaris\,1 Text-CSV_XS|0|1|MSWin32\,1\,linux\,1\,solaris\,1 SQL-Statement|0|1|MSWin32\,1\,linux\,1\,solaris\,1 Set-IntSpan|0|1|MSWin32\,1\,linux\,1\,solaris\,1 Convert-BER|0|1|MSWin32\,1\,linux\,1\,solaris\,1 Alias|0|0|MSWin32\,1\,linux\,1\,solaris\,1 B-Graph|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Bit-ShiftReg|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Bit-Vector|1|1|MSWin32\,1\,linux\,1\,solaris\,1 Business-CreditCard|0|0|MSWin32\,1\,linux\,1\,solaris\,1 CGI-Screen|0|0|MSWin32\,1\,linux\,1\,solaris\,1 CGI-XMLForm|0|0|MSWin32\,1\,linux\,1\,solaris\,1 CGI_Imagemap|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Chart|0|0|MSWin32\,1\,linux\,1 Class-Eroot|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Class-MethodMaker|0|0|MSWin32\,1\,linux\,1\,solaris\,1 DBD-CSV|0|0|MSWin32\,1\,linux\,1\,solaris\,1 # need an ODBC DB running to test the following databases # also\, doesn't build on Unix (missing libs?) DBD-ODBC|0|0|MSWin32\,0 # we need SDKs for the following DBDs on linux DBD-Oracle|0|0|MSWin32\,0 DBD-Sybase|0|0|MSWin32\,0 # not building libdb on linux (yet) DB_File|0|1|MSWin32\,1 Date-Calc|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Devel-Coverage|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Devel-Symdump|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Digest-MD5|0|0|MSWin32\,1\,linux\,1\,solaris\,1 File-Slurp|0|0|MSWin32\,1\,linux\,1\,solaris\,1 File-Tools|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Getopt-EvaP|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Getopt-Long|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Getopt-Mixed|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Getopt-Tabular|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Graph-Kruskal|0|0|MSWin32\,1\,linux\,1\,solaris\,1 HTML-Stream|0|0|MSWin32\,1\,linux\,1\,solaris\,1 IO-stringy|0|1|MSWin32\,1\,linux\,1\,solaris\,1 Image-Size|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Locale-Codes|0|0|MSWin32\,1\,linux\,1\,solaris\,1 MD5|0|0|MSWin32\,1\,linux\,1\,solaris\,1 MIME-Lite|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Mail-POP3Client|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Math-Approx|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Math-Matrix|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Math-MatrixBool|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Math-MatrixReal|0|0|MSWin32\,1\,linux\,1\,solaris\,1 MLDBM|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Net-Bind|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Net-DNS|0|0|MSWin32\,1\,linux\,1\,solaris\,1 # test failure on windows (uses alarm()) #Net-Ident|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Net-Ping|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Net-Telnet|0|0|MSWin32\,1\,linux\,1\,solaris\,1 # Test currently fails because author's server is down? Net-Whois|0|0|MSWin32\,0\,linux\,0\,solaris\,0 # needs DB_File on Unix Netscape-History|0|0|MSWin32\,1 # needs NNTP server for test News-NNTPClient|0|0|MSWin32\,1\,linux\,0\,solaris\,0 News-Newsrc|0|0|MSWin32\,1\,linux\,1\,solaris\,1 NNML|0|0|MSWin32\,1\,linux\,1\,solaris\,1 # test failures on windows #PNGgraph|0|0|MSWin32\,1\,linux\,1\,solaris\,1 POE|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Parse-Yapp|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Penguin-Easy|0|0|MSWin32\,1\,linux\,1\,solaris\,1 # need libraries on linux PerLDAP|1|0|MSWin32\,1 Period|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Roman|0|0|MSWin32\,1\,linux\,1\,solaris\,1 # build failure on linux SGMLS|0|0|MSWin32\,1 SHA|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Set-Object|0|0|MSWin32\,1\,linux\,1\,solaris\,1 # test failure on linux Set-Scalar|0|0|MSWin32\,1\,linux\,0\,solaris\,0 Sort-PolySort|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Statistics-ChiSquare|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Statistics-Descriptive|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Storable|0|1|MSWin32\,1\,linux\,1\,solaris\,1 String-BitCount|0|0|MSWin32\,1\,linux\,1\,solaris\,1 String-CRC|0|0|MSWin32\,1\,linux\,1\,solaris\,1 String-Parity|0|0|MSWin32\,1\,linux\,1\,solaris\,1 String-Scanf|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Term-ANSIColor|0|0|MSWin32\,1\,linux\,1\,solaris\,1 TermReadKey|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Text-German|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Text-Metaphone|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Text-Striphigh|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Text-Template|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Text-Vpp|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Tie-CPHash|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Tie-Dir|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Tie-IxHash|0|1|MSWin32\,1\,linux\,1\,solaris\,1 Tie-Watch|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Time-HiRes|0|0|MSWin32\,1\,linux\,1\,solaris\,1 TimeDate|0|0|MSWin32\,1\,linux\,1\,solaris\,1 # Skip the following Tk tests (can't currently test from churn) Tk|0|1|MSWin32\,1\,linux\,0\,solaris\,0 Tk-GBARR|0|0|MSWin32\,1\,linux\,0\,solaris\,0 Tk-ObjScanner|0|0|MSWin32\,1\,linux\,0\,solaris\,0 WWW-Search|0|0|MSWin32\,1\,linux\,1\,solaris\,1 Win32-API|1|1|MSWin32\,1 # non-standard build process #Win32-SerialPort|0|0|MSWin32\,1 # test failure on linux (build env problem) X11-Protocol|0|0|MSWin32\,1\,linux\,0\,solaris\,0 XML-DOM|0|0|MSWin32\,1\,linux\,1\,solaris\,1 perl-ldap|0|0|MSWin32\,1\,linux\,1\,solaris\,1 # need libs for Unix platforms sybperl|1|0|MSWin32\,0

p5pRT commented 24 years ago

From [Unknown Contact. See original ticket]

Gurusamy Sarathy \gsar@​ActiveState\.com writes​:

2. Tk - I only hit this last night - Tk800.018 built with patches\, so hopefully this will get sorted soon.

Does Tk800.015 build as-is ? that is the production version right now.

Tk 800.019 F (note 3)

Problem is in Tk\, apparently.

There are two minor nits with Tk800.019 as far as I am aware​:

1. Pedantic compilers need a ; after the skip​: label in tkGlue.c 2. Win32's tkWinPort.h needs the declaration of EXTERN panic ...   commenting out.

Both of those are perl version independant. Fixes would have been out by now but for​: A. The backlight on this laptop (my fast machine) being flakey. B. My ISP failing for 1/2 of today.

-- Nick Ing-Simmons

p5pRT commented 24 years ago

From @gsar

On Thu\, 16 Mar 2000 18​:07​:12 GMT\, Nick Ing-Simmons wrote​:

Gurusamy Sarathy \gsar@​ActiveState\.com writes​:

2. Tk - I only hit this last night - Tk800.018 built with patches\, so hopefully this will get sorted soon.

Does Tk800.015 build as-is ?

I know Tk800.018 builds\, but haven't tried 800.015.

Sarathy gsar@​ActiveState.com

p5pRT commented 24 years ago

From [Unknown Contact. See original ticket]

On Thu\, 16 Mar 2000\, Gurusamy Sarathy wrote​:

cv_undef() and cv_const_sv() are not in the list of public API functions. I'm not sure why mod_perl needs them\, but it seems like a bad idea to use them. (SvREFCNT_dec() should be used rather than cv_undef()\, because the latter will clobber a CV regardless of how many people own it. And cv_const_sv() is purely internal.)

Apache​::Symbol​::undef wants to clobber the CV regardless. it was a special version of pp_undef to avoid the old mandatory warning of constant subroutine being re-defined. cv_const_sv() is handy to check if a subroutine has been constant folded. i didn't notice this breakage\, both are referenced in Apache/Symbol.so\, which builds fine\, but is not used by the 'core'. now i notice mod_perl skips the modules/symbol test because it fails to load. it would be nice if these could be made public again.

p5pRT commented 24 years ago

From @gsar

On Thu\, 16 Mar 2000 10​:42​:01 PST\, Doug MacEachern wrote​:

On Thu\, 16 Mar 2000\, Gurusamy Sarathy wrote​:

cv_undef() and cv_const_sv() are not in the list of public API functions. I'm not sure why mod_perl needs them\, but it seems like a bad idea to use them. (SvREFCNT_dec() should be used rather than cv_undef()\, because the latter will clobber a CV regardless of how many people own it. And cv_const_sv() is purely internal.)

Apache​::Symbol​::undef wants to clobber the CV regardless. it was a special version of pp_undef to avoid the old mandatory warning of constant subroutine being re-defined. cv_const_sv() is handy to check if a subroutine has been constant folded. i didn't notice this breakage\, both are referenced in Apache/Symbol.so\, which builds fine\, but is not used by the 'core'. now i notice mod_perl skips the modules/symbol test because it fails to load. it would be nice if these could be made public again.

OK\, consider it done.

Sarathy gsar@​ActiveState.com

p5pRT commented 24 years ago

From [Unknown Contact. See original ticket]

From​: Gurusamy Sarathy [mailto​:gsar@​ActiveState.com]

On Thu\, 16 Mar 2000 10​:06​:03 GMT\, "Moore\, Paul" wrote​:

The nasties are

All three of which are problems in the extensions themselves... (just to drive home a point :)

Yes. Sorry if I didn't make that clear.

cv_undef() and cv_const_sv() are not in the list of public API functions. I'm not sure why mod_perl needs them\, but it seems like a bad idea to use them. (SvREFCNT_dec() should be used rather than cv_undef()\, because the latter will clobber a CV regardless of how many people own it. And cv_const_sv() is purely internal.)

Urk. I had heard rumours of a newer mod_perl\, but couldn't find it on CPAN when I looked. I'll check this one closer\, but I don't know anything about mod_perl\, so I probably won't find much :-(

I have to agree with whoever it was who commented that releasing 5.6.0 with key modules on CPAN not building in the latest version\, would be a problem.

FWIW\, I (mildly) disagree. New releases must happen to effect change in something as big as CPAN--we can't wait around for everyone to "catch up"\, because in my experience\, very few of them will "catch up" unless there's a release out to "create" demand for the changes.

It's a close call. But I'd tend to think that there are "big" CPAN modules which ought to work at the point of release. The ones I'd tend to put in this category are mod_perl\, DBI\, and Tk. And maybe libwin32. But I do agree that with something as big as CPAN\, trying to get everything tested and compatible is pretty much a no-hoper.

Thanks for the table.

Glad it was of use. I'll try to update it with some more comments\, plus a run with ithreads/fork for comparison. Maybe I'll put the results on my web page for general reference.

Paul.