CPAN
view release on metacpan or search on metacpan
preceded BBC saying something like "23456 breaks Term::ReadLine::Gnu".
This (at least the still open parts) should find its way into
perl5100delta.pod. Math::Pari and T:R:G being the most prominent
victims.
* Tcl maybe broke around 31941? We do not have the config of our perls
in the database, so SQL will not do it. We need to find all Tcl=OK for
perl=debugging. So some grep through all perl-V and all VKON! or so.
for d in logs/megainstall.2007{09,10}*.d; do
grep -H " optimize" $d/perl-V.txt | perl -pe 's/\n//'
grep ok= $d/VKON!Tcl-* |& perl -nle 'print /( ok=".+?")/'
done
[...]
logs/megainstall.20070922T1716.d/perl-V.txt: optimize='-g', ok="OK"
logs/megainstall.20070923T1153.d/perl-V.txt: optimize='-g', ok="NOT OK"
This means 31939-31941 and we need a config with optimize=-g. Nonono,
because 39 and 41 have long since been purged we need to order 40
directly, then the neighbor and then write the BBC (unless the neighbor
behaves differently from what we see in the database; which means 41
must fail).
31941 is easy peasy. The subject line says exactly what it is doing. But
is it correct to limit an assertion to debugging mode? Are the two
extensions doing stupid things? Let's examine the diagnostics. Tcl:
t/call.........Assertion ((svtype)((_svi)->sv_flags & 0xff)) >= SVt_RV failed: file "Tcl.xs", line 647 at t/call.t line 38.
dubious
^ITest returned status 88 (wstat 22528, 0x5800)
DIED. FAILED tests 9-10
^IFailed 2/10 tests, 80.00% okay
And Coro? 4.13 does not work with 31940.
31847/31854 MLEHMANN/Coro-3.63.tar.gz from NA to OK
31682/31684 MLEHMANN/Coro-3.63.tar.gz from OK to NA
SEGV during test: 31613, 31021, 30019, 29743, 29619
PL_parser undeclared in State.xs:1111: 29023, 29503, 29537
As this PL_parser is dependent on 5.9.0 I do not expect that any older perl will now compile
Try older Coro: 4.11 no, 4.1 no, 4.03 SEGV
SEGV during test with 4.03: 29537, 28022, 27002, 25229, 24044
State.xs:263: error: too few arguments to function Perl_pad_push: 23023, 23494, 23768, 23973
This State.xs:263 thing is even in the oldest Coro, 3.63 and I'm not up
to searching backpan.
So I will now have a look at the gdb analysis and then check in maint perl.
Failed to read a valid object file image from memory.
Core was generated by `/home/src/perl/repoperls/installed-perls/perl/pm5CFa8/perl-5.8.0@32198/bin/perl'.
Program terminated with signal 11, Segmentation fault.
#0 coro_transfer (prev=Cannot access memory at address 0xfffffffc
) at libcoro/coro.c:153
153 }
(gdb) bt
#0 coro_transfer (prev=Cannot access memory at address 0xfffffffc
) at libcoro/coro.c:153
Cannot access memory at address 0x4
Well. And maintperl? Ahhh, same thing! Stock 5.8.8 has it. 5.8.1 is
simply too old to understand current Coro code. 22251(5.8.3) has the
SEGV too. So no BBC in here.
* Bugreport to self: File::Temp needs a minimum of 0.14 if we use the
object interface. FIXED
* Devel-Caller poking around.
30955:
t/Devel-Caller....ok 1/72
# Failed (TODO) test 'with lexical $foo'
# at t/Devel-Caller.t line 36.
# got: 'GLOB(0x8319b70)'
# expected: 'SCALAR(0x840c250)'
# Failed (TODO) test 'with lexical @foo'
# at t/Devel-Caller.t line 37.
# got: 'SCALAR(0x843b3a8)'
# expected: 'ARRAY(0x84417fc)'
# Failed (TODO) test 'with lexical %foo'
# at t/Devel-Caller.t line 38.
# got: 'GLOB(0x843b268)'
# expected: 'HASH(0x8441824)'
Assertion ((svtype)((shplep)->sv_flags & 0xff)) == SVt_PVGV || ((svtype)((shplep)->sv_flags & 0xff)) == SVt_PVLV failed: file "lib/Devel/Caller.xs", line 49 at /home/sand/.cpan/build/Devel-Caller-0.11-JPp6H1/blib/lib/Devel/Caller.pm line 32.
# Looks like you planned 72 tests but only ran 43.
# Looks like your test died just after 43.
t/Devel-Caller....dubious
Test returned status 255 (wstat 65280, 0xff00)
DIED. FAILED tests 44-72
Failed 29/72 tests, 59.72% okay
32195:
t/Devel-Caller....
# Failed (TODO) test 'with lexical $foo'
# at t/Devel-Caller.t line 36.
# got: 'GLOB(0x838b8d0)'
# expected: 'SCALAR(0x83a47b0)'
# Failed (TODO) test 'with lexical @foo'
# at t/Devel-Caller.t line 37.
# got: 'SCALAR(0x84af790)'
# expected: 'ARRAY(0x84b6690)'
# Failed (TODO) test 'with lexical %foo'
# at t/Devel-Caller.t line 38.
# got: 'GLOB(0x84af678)'
# expected: 'HASH(0x84b66b8)'
dubious
Test returned status 0 (wstat 11, 0xb)
DIED. FAILED tests 44-72
Failed 29/72 tests, 59.72% okay
and Djabberd because it seemed to hang in the test even when it could
write to /tmp/djabberd/
* 5.6.2 fails one kwalify test in 12cpan.t for non obvious reason.
2007-04-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
* Crypt::OpenPGP needs a fix in the distroprefs file. What was the
problem? During XML-FOAF the usual recursion into installing OpenPGP
seemed to happen. Yes, I had PERL_AUTOINSTALL set, now added
PERL_EXTUTILS_AUTOINSTALL.
As long as this was missing, we had to kill the recursively acting
installer three times: during make, make test, and make install.
DONE.
* 1.91 is out.
2007-04-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
* bug: megainstall contains Bundle::Pause which is not published. Should
be.
2007-04-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
* see thread "echo die > Makefile.PL" on module-authors. Fortunately,
Joshua argues well.
* Slaven sent me dependency statistics on freebsd ports of perl modules.
I have checked the 208 top lines (20+ references) of 1200. I found
IO::Socket::INET6 hanging and added a few others to megainstall.
* reports command should display open issues on rt.cpan.org. Current
value to look up is
http://rt.cpan.org/Public/Dist/Display.html?Name=Template-Timer
Currently one would search for the element table and count the number of
tr minus one to get the number of new and open issues. The page is
declared as XHTML strict but isn't compliant.
2007-04-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
* Most urgent tasks?
Report Text::Hatena (JKONDO/Text-Hatena-0.20.tar.gz)
Report Locale::TextDomain (GUIDO/libintl-perl-1.16.tar.gz)
Write a bug report about Coro::Event and its interaction with CPAN.pm or
find out how to fix it and write a patch. Turn off distroprefs or work
without CPAN.pm to make things more obvious. Look:
cpan[2]> m Coro::Event
Module id = Coro::Event
CPAN_USERID MLEHMANN (Marc Lehmann <pcg@goof.com>)
CPAN_VERSION 2.1
CPAN_FILE M/ML/MLEHMANN/Coro-3.6.tar.gz
UPLOAD_DATE 2007-04-14
MANPAGE Coro::Event - do events the coro-way
INST_FILE (not installed)¯
Analyse the endless loop in "fforce test Task::Smoke". Maybe the fforce
is to blame?
FirstTime manpage?
Implement configure_requires? (testcase RPC::XML)
Implement a depends/requires keyword for distroprefs? (Testcase
Spreadsheet-ParseExcel (http://rt.cpan.org/Ticket/Display.html?id=24693)
or XML-Filter-XInclude
(http://rt.cpan.org/Ticket/Display.html?id=27401))
Speedup the cleanup? DONE [...time passes...] But I do not like the new
version. I want to see how long it takes/will take. svk rev. 2216 seems
to provide that.
Implement 30shell.t in a way that always emits useful output after a
failure? DONE in SVK rev. 2255
Find an opinion on MIME-tools. It seems that nobody cares. It's about
undef in %INC. I think it can stay as it is and we must educate
ourselves how %INC works now. But if it is not even in perldelta.pod?
Discussion on P5P was 11/12.4.
* Bug? "fforce test Task::Smoke" goes into an endless loop.
Test-TAP-Model and Task-Smoke play ping pong. All other prreqs are
satisfied quickly and prepended to PERL5LIB but these two stay active
drwx------ 6 k k 4096 Apr 16 06:58 Task-Smoke-0.15-5ksBtB
-rw-rw-r-- 1 k k 1424 Apr 16 06:58 Task-Smoke-0.15-5ksBtB.yml
-rw-rw-r-- 1 k k 1532 Apr 16 06:58 Test-TAP-Model-0.09-wUFmP9.yml
drwx------ 5 k k 4096 Apr 16 06:58 Test-TAP-Model-0.09-wUFmP9
drwx------ 6 k k 4096 Apr 16 06:58 Task-Smoke-0.15-QOiZ85
-rw-rw-r-- 1 k k 1424 Apr 16 06:58 Task-Smoke-0.15-QOiZ85.yml
-rw-rw-r-- 1 k k 1532 Apr 16 06:58 Test-TAP-Model-0.09-H3Lqy0.yml
drwx------ 5 k k 4096 Apr 16 06:58 Test-TAP-Model-0.09-H3Lqy0
drwx------ 6 k k 4096 Apr 16 06:58 Task-Smoke-0.15-6lDDuP
-rw-rw-r-- 1 k k 1424 Apr 16 06:58 Task-Smoke-0.15-6lDDuP.yml
-rw-rw-r-- 1 k k 1532 Apr 16 06:58 Test-TAP-Model-0.09-48ClDJ.yml
drwx------ 5 k k 4096 Apr 16 06:58 Test-TAP-Model-0.09-48ClDJ
drwx------ 6 k k 4096 Apr 16 06:57 Task-Smoke-0.15-V3h8gK
-rw-rw-r-- 1 k k 1424 Apr 16 06:57 Task-Smoke-0.15-V3h8gK.yml
-rw-rw-r-- 1 k k 1532 Apr 16 06:57 Test-TAP-Model-0.09-MI4CP5.yml
drwx------ 5 k k 4096 Apr 16 06:57 Test-TAP-Model-0.09-MI4CP5
drwx------ 6 k k 4096 Apr 16 06:57 Task-Smoke-0.15-0rmHRt
-rw-rw-r-- 1 k k 1424 Apr 16 06:57 Task-Smoke-0.15-0rmHRt.yml
-rw-rw-r-- 1 k k 1532 Apr 16 06:57 Test-TAP-Model-0.09-A0HeR8.yml
drwx------ 5 k k 4096 Apr 16 06:57 Test-TAP-Model-0.09-A0HeR8
drwx------ 6 k k 4096 Apr 16 06:57 Task-Smoke-0.15-BpPXIa
After I interrupt the endless loop and say "install Task::Smoke" all is
well quickly. Let's page back what was prepended to PERL5LIB?
---- Unsatisfied dependencies detected during ----
---- GAAL/Task-Smoke-0.15.tar.gz ----
Test::TAP::Model [requires]
Prepending /home/k/.cpan/build/Test-TAP-HTMLMatrix-0.08-t1Tza8/blib/arch /home/k/.cpan/build/Test-TAP-HTMLMatrix-0.08-t1Tza8/blib/lib /home/k/.cpan/build/Method-Alias-1.03-Xgby0k/blib/arch /home/k/.cpan/build/Method-Alias-1.03-Xgby0k/blib/lib /home/k...
( run in 1.836 second using v1.01-cache-2.11-cpan-2398b32b56e )