Re: Proposed removal of arch-perl (libarch-perl)
On 09 Mar 2010 12:33:44 +0100, Alex Muntada wrote: > > + Mikhael Goikhman : > > > I think README gives a handful of hints about the package. Anyway, in > > the devel branch (managed under tla, that is mentioned in README too) > > all tests should now pass even without tla or baz installed. > > I.e.: TLA=/bin/false make test > > IIRC, having Makefile.PL exit a non-zero value should make all CPAN > smokers ignore the module (no FAIL nor SKIP). Since it should be > quite straight-forward to check if tla is available on the system > and there's no point on installing Arch module unless tla is > available, failing early in Makefile.PL is a Good Thing, IMHO. This is a reasonable approach. On the other hand it is a pure perl code that may be installed before installing tla (or its equivalent), and you may even point to different backend installations at run-time. In addition, half of the functionality does not depend on presense of tla. Also, the code has explicit support for perl 5.005, 5.6, 5.8+ and it would be interesting to see whether it is actually true. So the 0.5.2 release from yesterday tries a different approach, to skip the relevant tests with "No functional arch backend" reason. Regards, Mikhael. -- perl -e 'print+chr(64+hex)for+split//,d9b815c07f9b8d1e' -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100309142118.gb21...@otaku
Re: Proposed removal of arch-perl (libarch-perl)
+ Mikhael Goikhman : > I think README gives a handful of hints about the package. Anyway, in > the devel branch (managed under tla, that is mentioned in README too) > all tests should now pass even without tla or baz installed. > I.e.: TLA=/bin/false make test IIRC, having Makefile.PL exit a non-zero value should make all CPAN smokers ignore the module (no FAIL nor SKIP). Since it should be quite straight-forward to check if tla is available on the system and there's no point on installing Arch module unless tla is available, failing early in Makefile.PL is a Good Thing, IMHO. HTH -- Alex Muntada http://alexm.org/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/35064d941003090333o5bebf853w67e50c82042a4...@mail.gmail.com
Re: Proposed removal of arch-perl (libarch-perl)
On 07 Mar 2010 18:19:48 -0500, Jonathan Yu wrote: > > I'm really embarrassed now, for not having asked you about this issue > first. I'll take a look at fixing the libarch-perl package > immediately, and look into adopting the other related packages based > on it (assuming they are Perl code). > > Perhaps something you can consider doing is having tests skipped > (rather than passed) with a message like "Please install tla to run > tests"... Otherwise, as a Perl developer, I know it's difficult to > specify "make sure is installed first", and > unfortunately, the current package doesn't leave hints as to what it > requires (which I guess is partly due to us not running the tests at > build time). I think README gives a handful of hints about the package. Anyway, in the devel branch (managed under tla, that is mentioned in README too) all tests should now pass even without tla or baz installed. I.e.: TLA=/bin/false make test > On Sun, Mar 7, 2010 at 6:04 PM, Mikhael Goikhman wrote: > > I am aware of the problem. Once you (testers) install tla (GNU Arch) > > package, it will likely be all PASSes. :) Of course I may make tests > > silently pass without tla installed, but then this perl library is > > useless on such systems, so it is hard to decide what is better. > > > > The reason to remove this package may be that nearly 100% or so users > > don't use GNU Arch. But other than this it should be a fully > > functional library. (At least I use it almost daily.) > I don't consider this a reason to remove a package, really. As long as > a package is useful to at least a handful of users and isn't totally > ridiculous to maintain, I think it's worth keeping in Debian. Have a nice day. Regards, Mikhael. -- perl -e 'print+chr(64+hex)for+split//,d9b815c07f9b8d1e' -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100308014252.gb7...@ukato
Re: Proposed removal of arch-perl (libarch-perl)
On 07 Mar 2010 16:17:56 -0500, Jonathan Yu wrote: > > On Sun, Mar 7, 2010 at 3:14 PM, Clint Adams wrote: > > Have you asked upstream? > > Actually, no I haven't, since given the number of test failures on > CPAN Testers (pretty much 100% are FAILs), I assumed the the author > was already aware of the problem. On second thought, though, you're > right -- many authors don't get the test reports for some reason or > another (maybe their CPAN e-mail account doesn't forward anywhere, > etc). I am aware of the problem. Once you (testers) install tla (GNU Arch) package, it will likely be all PASSes. :) Of course I may make tests silently pass without tla installed, but then this perl library is useless on such systems, so it is hard to decide what is better. The reason to remove this package may be that nearly 100% or so users don't use GNU Arch. But other than this it should be a fully functional library. (At least I use it almost daily.) Regards, Mikhael. -- perl -e 'print+chr(64+hex)for+split//,d9b815c07f9b8d1e' -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100307230428.ga7...@ukato
Re: Proposed removal of arch-perl (libarch-perl)
Mikhael, I'm really embarrassed now, for not having asked you about this issue first. I'll take a look at fixing the libarch-perl package immediately, and look into adopting the other related packages based on it (assuming they are Perl code). Perhaps something you can consider doing is having tests skipped (rather than passed) with a message like "Please install tla to run tests"... Otherwise, as a Perl developer, I know it's difficult to specify "make sure is installed first", and unfortunately, the current package doesn't leave hints as to what it requires (which I guess is partly due to us not running the tests at build time). On Sun, Mar 7, 2010 at 6:04 PM, Mikhael Goikhman wrote: > I am aware of the problem. Once you (testers) install tla (GNU Arch) > package, it will likely be all PASSes. :) Of course I may make tests > silently pass without tla installed, but then this perl library is > useless on such systems, so it is hard to decide what is better. > > The reason to remove this package may be that nearly 100% or so users > don't use GNU Arch. But other than this it should be a fully > functional library. (At least I use it almost daily.) I don't consider this a reason to remove a package, really. As long as a package is useful to at least a handful of users and isn't totally ridiculous to maintain, I think it's worth keeping in Debian. Thanks for your advice! Cheers, Jonathan -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/d1b732a71003071519u7156e845uda74d5f5b81d...@mail.gmail.com
Re: Proposed removal of arch-perl (libarch-perl)
Hi, Jonathan Yu wrote: On Sun, Mar 7, 2010 at 3:14 PM, Clint Adams wrote: Have you asked upstream? Actually, no I haven't, since given the number of test failures on CPAN Testers (pretty much 100% are FAILs), I assumed the the author was already aware of the problem. On second thought, though, you're right -- many authors don't get the test reports for some reason or another (maybe their CPAN e-mail account doesn't forward anywhere, etc). Did not follow the hole tread ... but failed tests on CPAN can be caused by missing non CPAN dependencies. FYI: after aptitude install tla All tests successful. Files=33, Tests=686, 5 wallclock secs ( 0.19 usr 0.02 sys + 2.00 cusr 0.54 csys = 2.75 CPU) Result: PASS (Arch-0.5.1 via CPAN on Lenny) Of course I did not enable tests which where disabled by upstream. Yours Christian -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4b942d84.7030...@cipworx.org
Re: Proposed removal of arch-perl (libarch-perl)
Clint, On Sun, Mar 7, 2010 at 3:14 PM, Clint Adams wrote: > Have you asked upstream? Actually, no I haven't, since given the number of test failures on CPAN Testers (pretty much 100% are FAILs), I assumed the the author was already aware of the problem. On second thought, though, you're right -- many authors don't get the test reports for some reason or another (maybe their CPAN e-mail account doesn't forward anywhere, etc). Hopefully we'll hear from MIGO soon. Cheers, Jonathan -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/d1b732a71003071317w22972e46k2630985de5f9e...@mail.gmail.com
Re: Proposed removal of arch-perl (libarch-perl)
On Sun, Mar 07, 2010 at 01:06:13PM -0500, Jonathan Yu wrote: > Recently I've been working on adopting the arch-perl package under the > Debian Perl Group's umbrella. However, there are now some test > failures (which didn't surface before because tests were simply > disabled). So, long story short, my main issue with this is we are > redistributing software which does not pass its own tests (and which > is currently untested). Have you asked upstream? -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100307201428.ga11...@scru.org
Re: Proposed removal of arch-perl (libarch-perl)
On Sun, 07 Mar 2010 13:06:13 -0500, Jonathan Yu wrote: > 2. popcon score - see > http://qa.debian.org/popcon.php?package=arch-perl - does not appear > helpful. It has 0 all across, despite hundreds of submitters according > to the graph. The page for the binary package seems more helpful, at least for guessing numbers from the chart: http://qa.debian.org/popcon-graph.php?packages=libarch-perl If I look at http://qa.debian.org/popcon-graph.php?packages=axp http://qa.debian.org/popcon-graph.php?packages=archzoom http://qa.debian.org/popcon-graph.php?packages=archway the values for "vote" (~ actually used recently) is under 10, so they look like removal candidates. Cheers, gregor -- .''`. http://info.comodo.priv.at/ -- GPG Key IDs: 0x8649AA06, 0x00F3CFE4 : :' : Debian GNU/Linux user, admin, & developer - http://www.debian.org/ `. `' Member of VIBE!AT & SPI, fellow of Free Software Foundation Europe `-NP: Misha Alperin: Psalm No.1 signature.asc Description: Digital signature
Re: Proposed removal of arch-perl (libarch-perl)
On 07/03/10 19:06, Jonathan Yu wrote: > Who is using arch-perl? > > 1. It has many reverse-dependencies > Reverse Depends: > axp > archzoom > archway > axp > archzoom > archway Well, those are repeated so they are not that many. Maybe you can remove those three packages together with libarch-perl? They're all three orphaned... Cheers, Emilio -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4b93ede5.5020...@debian.org
Proposed removal of arch-perl (libarch-perl)
Hi: Recently I've been working on adopting the arch-perl package under the Debian Perl Group's umbrella. However, there are now some test failures (which didn't surface before because tests were simply disabled). So, long story short, my main issue with this is we are redistributing software which does not pass its own tests (and which is currently untested). One possible solution is to just keep tests disabled, but this seems like a bad idea. Who is using arch-perl? 1. It has many reverse-dependencies Reverse Depends: axp archzoom archway axp archzoom archway 2. popcon score - see http://qa.debian.org/popcon.php?package=arch-perl - does not appear helpful. It has 0 all across, despite hundreds of submitters according to the graph. 3. Last upstream upload was in 2005. However, there are no bugs filed against the request tracker. 4. CPAN Testers results are currently: FAIL (114) NA (1) UNKNOWN (2) I am sort of conflicted on the best way to solve this problem and am open to the Debian community's suggestions here. Presumably the Arch packages mentioned *really* need this module, which, as mentioned above, doesn't even pass its own tests. We can continue distributing this package in Debian with tests disabled, I suppose. Cheers, Jonathan -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/d1b732a71003071006j163ed062v7b1b71db2f06d...@mail.gmail.com