Logo ognion
Perl QA Hackathon 2014

March 13-16, 2014 in Lyon

Logo ognion
 

Perl QA Hackathon 2014

March 13-16, 2014
in Lyon, France

Logo PQAH  

March 13-16, 2014
in Lyon, France

This page is dedicated to reporting what has been done day by day during the event.


Thursday 13th March

Tim Bunce:

Matthew Horsfall (‎alh‎):

  • Added parallel testing support to p5's Makefile for valgrind testing (test.valgrind)
  • Add deparse testing support to p5's Makefile for smoking deparse (test.deparse)
  • Played around with a Configure -i flag (ignore/rm Policy.sh and config.sh)
    • Talked to H.Merijn Brand (‎Tux‎) about ./Configure with cached data. More is needed here, perhaps for now though separate out the functionality to another shell/Perl script so it can be improved in the future without touching Configure. Might play with this later.
  • Added patch to update doc or fix code for hv_iterinit
  • Fixed Devel::Cover to work with 5.20 and later 5.19.Xs.
  • PPI review for Christian Walde (‎Mithaldu‎)
  • Took a look at test.porting supporting parallel tests.
    • It already does!
  • Bugged Tim Bunce to give me github access to Memcached::libmemcached

Steffen Schwigon (‎renormalist‎):

  • opensource released some Tapper libs
  • TAP::DOM release 0.11
    • assume TAP version 13 by default
    • generate TAP from TAP::DOM
  • Data::DPath release 0.50
    • small optimization

Philippe Bruhat (‎BooK‎):

  • showed everyone where the toilets were
  • discussion with TIMB about Test::Database

Friday 14th March

Tim Bunce:

  • More DBI::Test code refinements and design discussions with Jens Rehsack (‎Sno‎) and ribasushi +1, especially around the test variant generator.
  • About ready to start working on extracting and releasing Data::Tumbler and Test::WriteVariants as new distributions on CPAN.
  • Discussed the problems of packagers (eg rpm creators) determining dependencies for compiled perl extension.
  • Proposed an extension to the META spec as an appropriate way to capture the required information and build tooling around it.

Matthew Horsfall (‎alh‎)

  • Memcached::libmemcached
    • Updated META_MERGE a bit (change bugtracker, include repo url)
    • Check pending tickets for work / clean up bug queue. Most seem stalled on more user input
    • Release ! (Includes fixes for floop-parallel-all)
  • bisected/reported "Attempt to free unreferenced scalar" with here doc in regexop
  • File bug about B::Deparse not handling newlines in (?{...}) regex constructs

Olivier Mengué (‎dolmen‎)

  • Arriving late...
  • Digged into Module::CPANTS::Analyse and Module::CPANTS::SiteKwalitee. I discussed with charsbar and he gave me tips to continue my quests of reporting tickets on distributions that bundle MYMETA.*: I was using grep.cpan.me so far, I will now use http://cpants.cpanauthors.org/kwalitee/no_mymeta_files
  • In the evening, followed an interesting discussion about how to improve automated packaging by adding metadata to META.json. Jen, Andreas, Tim Bunce have planned to further discuss that tomorrow.

Philippe Bruhat (‎BooK‎):

  • was interviewed with NEILB by SJN about gamification of CPAN
  • did some work on Test::Database
  • talked with NEILB about our gamification of CPAN project

Saturday 15th March

Tim Bunce:

  • Released Data::Tumbler and Test::WriteVariants to CPAN.
  • Contributed to the design discussions around extending the META spec for packagers.
  • Updated the DBI::Test prototype to use the new Test::WriteVariants module.

Matthew Horsfall (‎alh‎):

  • Update test.valgrind patch to remove previous output otherwise it would interfere
    • Noticed valgrind patches abandon work done by previous work of others. Need to reconsider/rewrite.
  • Work on getting a base testing setup for Devel::PPPort work. Lots of hacking/fighting with various

tools

Steffen Schwigon (‎renormalist‎):

  • completed reviewing and polishing of literally 2 non-mythical man years of Tapper git history
    • in a way that allows coming back to a more "early and often" release cycle
    • some approved libs already published
    • prepared changelog

Olivier Mengué (‎dolmen‎)

  • Investigated the issue of `$!` https://github.com/dolmen/p5-strerror-Unicode:
    • Checked how `setlocale()` and `strerror()` work on Windows to verify if the issue applies here: it doesn't. Note that `setlocale()` works quite differently from Unix. For example, it doesn't use environment variables (LC_*) at all.
    • Got help from Sébastien Aperghis-Tramoni (‎maddingue‎) who found that the issue is at least partially fixed (only for UTF-8 locales) in blead
  • Persued in locales world on Windows to fix `use open qw< :locale :std >;` on that platform. Patch for encoding::_get_locale_encoding() in progress
  • Followed discussions:
  • Managed to kill the screen of my laptop by dropping it in the stairs when going to sleep. :'(

Philippe Bruhat (‎BooK‎):

Sunday 16th March

Steffen Schwigon (‎renormalist‎):

Philippe Bruhat (‎BooK‎):

  • opened the door (took about 3 hours)

Tim Bunce:

Matthew Horsfall (‎alh‎):

  • Devel::PPPort - Familiarized myself more with the workings. Added these APIs:
    • SvREFCNT_dec_NN
    • mg_findext
    • sv_unmagicext

Olivier Mengué (‎dolmen‎):

  • Continued my work on encoding::_get_locale_encoding for Win32. Patch in progress is here: https://github.com/dolmen/p5-Encode/commits/fix/Win32-locale-encoding
  • Installed distributions produced by my peers, to ensure they get CPAN Testers reports:
  • Dist::Zilla:
    • Participated to discussion about what could trigger Dist::Zilla 6.0
      • I waved that the ExtraTests plugin should be deprecated.
    • Followed profiling of a Dist::Zilla build with Devel::NYTProf and the analysis of the flame graph it produced
      • PPI is an area that should be improved.
      • PPI is often used to separate POD from code. Let's see if we can avoid using PPI for that case.
  • Spent some time trying to workaround my broken laptop screen :(

version 35 saved on 18/03/14 10:44 by Olivier Mengué (‎dolmen‎)

Home | Tags | Recent changes | History

This page is dedicated to reporting what has been done day by day during the event.


Thursday 13th March

Tim Bunce:

Matthew Horsfall (‎alh‎):

  • Added parallel testing support to p5's Makefile for valgrind testing (test.valgrind)
  • Add deparse testing support to p5's Makefile for smoking deparse (test.deparse)
  • Played around with a Configure -i flag (ignore/rm Policy.sh and config.sh)
    • Talked to H.Merijn Brand (‎Tux‎) about ./Configure with cached data. More is needed here, perhaps for now though separate out the functionality to another shell/Perl script so it can be improved in the future without touching Configure. Might play with this later.
  • Added patch to update doc or fix code for hv_iterinit
  • Fixed Devel::Cover to work with 5.20 and later 5.19.Xs.
  • PPI review for Christian Walde (‎Mithaldu‎)
  • Took a look at test.porting supporting parallel tests.
    • It already does!
  • Bugged Tim Bunce to give me github access to Memcached::libmemcached

Steffen Schwigon (‎renormalist‎):

  • opensource released some Tapper libs
  • TAP::DOM release 0.11
    • assume TAP version 13 by default
    • generate TAP from TAP::DOM
  • Data::DPath release 0.50
    • small optimization

Philippe Bruhat (‎BooK‎):

  • showed everyone where the toilets were
  • discussion with TIMB about Test::Database

Friday 14th March

Tim Bunce:

  • More DBI::Test code refinements and design discussions with Jens Rehsack (‎Sno‎) and ribasushi +1, especially around the test variant generator.
  • About ready to start working on extracting and releasing Data::Tumbler and Test::WriteVariants as new distributions on CPAN.
  • Discussed the problems of packagers (eg rpm creators) determining dependencies for compiled perl extension.
  • Proposed an extension to the META spec as an appropriate way to capture the required information and build tooling around it.

Matthew Horsfall (‎alh‎)

  • Memcached::libmemcached
    • Updated META_MERGE a bit (change bugtracker, include repo url)
    • Check pending tickets for work / clean up bug queue. Most seem stalled on more user input
    • Release ! (Includes fixes for floop-parallel-all)
  • bisected/reported "Attempt to free unreferenced scalar" with here doc in regexop
  • File bug about B::Deparse not handling newlines in (?{...}) regex constructs

Olivier Mengué (‎dolmen‎)

  • Arriving late...
  • Digged into Module::CPANTS::Analyse and Module::CPANTS::SiteKwalitee. I discussed with charsbar and he gave me tips to continue my quests of reporting tickets on distributions that bundle MYMETA.*: I was using grep.cpan.me so far, I will now use http://cpants.cpanauthors.org/kwalitee/no_mymeta_files
  • In the evening, followed an interesting discussion about how to improve automated packaging by adding metadata to META.json. Jen, Andreas, Tim Bunce have planned to further discuss that tomorrow.

Philippe Bruhat (‎BooK‎):

  • was interviewed with NEILB by SJN about gamification of CPAN
  • did some work on Test::Database
  • talked with NEILB about our gamification of CPAN project

Saturday 15th March

Tim Bunce:

  • Released Data::Tumbler and Test::WriteVariants to CPAN.
  • Contributed to the design discussions around extending the META spec for packagers.
  • Updated the DBI::Test prototype to use the new Test::WriteVariants module.

Matthew Horsfall (‎alh‎):

  • Update test.valgrind patch to remove previous output otherwise it would interfere
    • Noticed valgrind patches abandon work done by previous work of others. Need to reconsider/rewrite.
  • Work on getting a base testing setup for Devel::PPPort work. Lots of hacking/fighting with various

tools

Steffen Schwigon (‎renormalist‎):

  • completed reviewing and polishing of literally 2 non-mythical man years of Tapper git history
    • in a way that allows coming back to a more "early and often" release cycle
    • some approved libs already published
    • prepared changelog

Olivier Mengué (‎dolmen‎)

  • Investigated the issue of `$!` https://github.com/dolmen/p5-strerror-Unicode:
    • Checked how `setlocale()` and `strerror()` work on Windows to verify if the issue applies here: it doesn't. Note that `setlocale()` works quite differently from Unix. For example, it doesn't use environment variables (LC_*) at all.
    • Got help from Sébastien Aperghis-Tramoni (‎maddingue‎) who found that the issue is at least partially fixed (only for UTF-8 locales) in blead
  • Persued in locales world on Windows to fix `use open qw< :locale :std >;` on that platform. Patch for encoding::_get_locale_encoding() in progress
  • Followed discussions:
  • Managed to kill the screen of my laptop by dropping it in the stairs when going to sleep. :'(

Philippe Bruhat (‎BooK‎):

Sunday 16th March

Steffen Schwigon (‎renormalist‎):

Philippe Bruhat (‎BooK‎):

  • opened the door (took about 3 hours)

Tim Bunce:

Matthew Horsfall (‎alh‎):

  • Devel::PPPort - Familiarized myself more with the workings. Added these APIs:
    • SvREFCNT_dec_NN
    • mg_findext
    • sv_unmagicext

Olivier Mengué (‎dolmen‎):

  • Continued my work on encoding::_get_locale_encoding for Win32. Patch in progress is here: https://github.com/dolmen/p5-Encode/commits/fix/Win32-locale-encoding
  • Installed distributions produced by my peers, to ensure they get CPAN Testers reports:
  • Dist::Zilla:
    • Participated to discussion about what could trigger Dist::Zilla 6.0
      • I waved that the ExtraTests plugin should be deprecated.
    • Followed profiling of a Dist::Zilla build with Devel::NYTProf and the analysis of the flame graph it produced
      • PPI is an area that should be improved.
      • PPI is often used to separate POD from code. Let's see if we can avoid using PPI for that case.
  • Spent some time trying to workaround my broken laptop screen :(

version 35 saved on 18/03/14 10:44 by Olivier Mengué (‎dolmen‎)

Home | Tags | Recent changes | History