17:09:44 <abadger1999> #startmeeting fpc
17:09:44 <zodbot> Meeting started Wed Jan 30 17:09:44 2013 UTC.  The chair is abadger1999. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:09:44 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:10:02 <spot> abadger1999: are you trying to convince me to leave? ;)
17:10:04 <tibbs> Howdy.
17:10:14 <abadger1999> spot: hah! :-)
17:11:05 <abadger1999> #topic NodeJs: https://fedoraproject.org/wiki/PackagingDrafts:Node.js
17:11:33 <abadger1999> We had a list of questions last week and I added them to the Questions section at the bottom.
17:12:38 <abadger1999> Here's the changes from last week: https://fedoraproject.org/w/index.php?title=PackagingDrafts%3ANode.js&diff=320438&oldid=320362
17:14:52 <tibbs> So, just the one admon at the top?
17:15:30 <abadger1999> Right.
17:15:41 * Smoother1rOgZ here
17:15:46 <abadger1999> He linked to a spec template which should be easy enough to merge into its own section.
17:16:06 <Rathann> hm, one question - if nodejs is noarch, why does it use %{prefix}/lib instead of %{_datadir}?
17:16:16 <Rathann> same for modules
17:17:07 <abadger1999> Rathann: Well... we've had this discussion before... library code, even if its noarch can go in %{_prefix}/lib
17:17:29 <Rathann> ah, seems I forgot that one
17:17:30 <Rathann> sorry
17:17:45 <limburgher> Bah, sorry I'm late, where are we?
17:17:50 <abadger1999> Rathann: java, mono, python, perl all use %{_prefix}/lib
17:18:02 <Rathann> apologies, but I need to go away for a while, urgently
17:18:11 <rdieter> seems most of our questions and concerns have been addressed, minus the %nodejs_fixdep thing
17:18:22 <abadger1999> some of those split between %{_prefix}/lib and %{_libdir}, others have an exemption to put them all in %{_prefix}/lib
17:18:58 * abadger1999 agrees with rdieter.
17:19:02 <tibbs> Still, seems that this isn't quite done.
17:19:31 <abadger1999> tibbs: k.  Do we have additional questions/direction on what we want accomplished?
17:20:07 * abadger1999 would also be okay voting on this and then iterating improvments over time.
17:20:40 <tibbs> I think that diff isn't complete, or something.
17:20:54 <spot> "After some thought, I think it might be better to drop %nodejs_fixdep and provide a similar script that generates a patch instead."
17:21:20 <spot> so perhaps we can consider the updated draft - the fixdep section?
17:21:22 <tibbs> Yeah, I was expecting that to be gone.
17:21:48 <tibbs> Also, the spec template, but that's not a big deal.
17:22:00 <spot> yeah, -fixdep references. :)
17:23:11 * abadger1999 does that quickly
17:23:37 <racor> abadger1999: java etc. all are facing the same problem (IMO: design bug). FWIW: perl uses %_libdir /usr/lib64/perl5
17:23:59 <tibbs> Personally I didn't mind the fixdep thing, but if that's going to be replaced by some kind of patch generator that people are supposed to use then it would be nice to actually see that.
17:24:25 <tibbs> Otherwise we just have to come back here and vote on the revision.
17:24:55 <Smoother1rOgZ> ll /home/nboutin/ -a
17:25:16 <tibbs> Indeed.
17:25:29 <abadger1999> Okay, refresh https://fedoraproject.org/wiki/PackagingDrafts:Node.js  nodejs_fixdeps removal
17:26:30 <abadger1999> racor: right.  java and mono have the issue.  perl and python do the arch/noarch split.
17:27:31 <spot> FWIW, I'm +1 on the draft as it is now (minus the Questions and below, which aren't the draft)
17:27:44 <abadger1999> +1
17:28:56 <geppetto> +1
17:29:11 <limburgher> +1 also
17:29:13 <tibbs> +1; I guess we can look at the dependency fixing thing later.
17:29:18 <Smoother1rOgZ> +1
17:30:18 <abadger1999> We're at +6, racor, Rathann, rdieter you can weigh in for the record.
17:30:46 <rdieter> +1
17:31:31 <abadger1999> #info Nodejs packaging guidelines pass (+1: 6, 0: 0, -1: 0)   -- Note that a FESCo multilib exemption is needed.
17:31:51 <abadger1999> #topic Ruby Guidelines update: https://fedoraproject.org/w/index.php?title=PackagingDrafts%2FRuby
17:32:30 <abadger1999> Diff from what we looked at last week: https://fedoraproject.org/w/index.php?title=PackagingDrafts%2FRuby&diff=320523&oldid=320351
17:32:42 <abadger1999> Questions from FPC and FESCo are at the bottom.
17:34:16 <abadger1999> mitr brought up in the FESCo discussion that ENV variables were a more UNIXy  method and safer than modifying the commandline.
17:34:36 <abadger1999> Bkabrda added that but left the command line option in.
17:35:04 <abadger1999> We should decide if we want to keep that or mandate that only the ENV var method be used in packaging.
17:35:56 <abadger1999> The other half-answered question was about running unittests.  Bkabrda added information about how to run them under jruby but didn't want to mandate running them.
17:36:15 <abadger1999> He's right that we don't have an absolute rule that unittests must be run.
17:36:29 <tibbs> I kind of agree with the latter, but it should at least be suggested.
17:36:48 <abadger1999> However, the closest other example we currently have is python2 vs python3 where the guideline anticipates that you'll run the unittests for both.
17:36:49 <tibbs> In my package reviews I always at least ask why they didn't run tests if there are any they didn't run.
17:37:19 <abadger1999> ie: nothing says, "You MUST" -- but the spec template, etc have the %check section to do both.
17:37:21 <tibbs> As for the ENV var thing, I'm trying to understand what in the proposed guidelines would change depending on how we decide.
17:38:00 * spot has to go now (in europe, meeting people). ruby draft looks good to me, i'm ambivalent on ENV vs cmdline
17:39:15 <abadger1999> tibbs: There's two choices (maybe three): 1) Only use ENV in the examples, not the CLI possibility.
17:39:38 <abadger1999> 2)  Have bkabrda change the upstream rubypick to not allow CLI
17:40:00 <abadger1999> 2) Recommend to FESCo not to allow CLI when they approve the Ruby2/JRuby Feature
17:40:10 <tibbs> OK, so it's the rubypick section where this is important?
17:40:14 <abadger1999> Yeah.
17:41:21 <geppetto> My feeling is that the packaging policy examples should only show one way to do it, just to avoid confusion.
17:41:33 <geppetto> Not sure I care about ENV vs. CLI though.
17:41:59 <geppetto> So I guess abadger1999's #1
17:42:08 <tibbs> Indeed, not sure I think I really care.
17:42:33 <Smoother1rOgZ> me neither
17:42:35 <tibbs> At least with a variable you can export it and make a global change.
17:42:55 <tibbs> As opposed to endlessly repeating lots of underscores.
17:43:00 <abadger1999> <nod>  Now that I can read how they're planning on using it -- I think it's more a FESCo issue.  For packagers, it's just a choice of using /usr/bin/ruby (=> rubypick) vs /usr/bin/ruby-mri
17:44:38 <abadger1999> If both LCI and ENV stay, I'm okay with examples of both staying...
17:45:12 <tibbs> But the guidelines should at least be consistent.
17:45:40 <abadger1999> How about -- I'll watch whether FESCo prohibits the CLI usage and change the examples to ENV var as a clarification if that happens.
17:46:31 <tibbs> Do the guidelines even use any of that outside of the rubypick section?
17:46:49 <abadger1999> tibbs: well... there are two usages though.  I suppose they have two audiences... sitewide/user-wide setting would use env var.  One off could use env var but may find CLI easier to remember.
17:47:36 <abadger1999> tibbs: nope.  I had thought they were going to change shebangs to be #!/usr/bin/ruby __mri__    but the admon/note makes it clear that they want to use /usr/bin/ruby-mri
17:47:47 <tibbs> It appears the guidelines care about that exactly once outside of the rubypick section, in one call to testrb where it actually makes sense to use the command line version.
17:48:20 <tibbs> So at this point I'm inclined to say that what's there is fine unless somehow the non-ENV stuff gets banned.
17:48:22 <abadger1999> (In cases where the app needed to run on a specific interpreter)
17:48:29 <abadger1999> wfm.
17:48:57 <geppetto> I still think having two examples of doing one thing is going to be more confusing, and thus. worse.
17:49:11 <limburgher> There's certainly that possibility.
17:52:14 <tibbs> So....
17:52:25 <tibbs> I mean, I just don't care enough to care.
17:56:04 * abadger1999 rewriting a little
17:56:21 <Rathann> I'm a bit confused why rubypick is described
17:56:44 <Rathann> the draft doesn't say what to do with it, whether it's recommended to use it or not (and how)
17:58:46 <abadger1999> https://fedoraproject.org/wiki/PackagingDrafts/Ruby#Shebang_lines
17:59:01 <abadger1999> Does that look better?
17:59:10 <abadger1999> Rathann: That might help with your confusion as well :-)
18:01:16 <tibbs> Seems fine to me.
18:01:26 <Smoother1rOgZ> +1 on this
18:01:45 <abadger1999> Okay -- so unittests...
18:02:03 <tibbs> At this point I'm good with the whole draft.
18:02:39 <abadger1999> I see that the current guidelines have running unittests as: " it '''should''' be run in %check." and "You may skip test suite execution when not all build dependencies are met but this must be documented in the specfile. "
18:02:41 <tibbs> By the way, about tests,
18:02:55 <tibbs> http://fedoraproject.org/wiki/Packaging:Guidelines#Test_Suites
18:03:04 <tibbs> If the source code of the package provides a test suite, it should be executed in the %check section, whenever it is practical to do so.
18:03:05 <abadger1999> I think we should do something similar for running the tests under jruby
18:03:17 <abadger1999> ah, cool.
18:03:19 <tibbs> Which seems to be cover the question of whether tests should be run.
18:03:46 <tibbs> I think adding a thing about testing under both interpreters with the same strength as that statement makes sense.
18:03:58 <abadger1999> k
18:04:39 <tibbs> "practical to do so" gives pretty reasonable leeway in any case.
18:07:24 <abadger1999> <nod>
18:07:29 * abadger1999 triesto write this up
18:09:54 <Rathann> abadger1999: wfm, thanks
18:10:17 <abadger1999> https://fedoraproject.org/wiki/PackagingDrafts/Ruby#Testing_With_Different_Ruby_Implementations  <= reworded section
18:12:55 <Smoother1rOgZ> wfm.
18:13:02 <tibbs> Sure, seems fine.
18:13:07 * Smoother1rOgZ would bold "need"
18:13:56 <abadger1999> Let's vote if we still have quorum
18:14:00 <tibbs> +1
18:14:02 <abadger1999> +1
18:14:26 <Smoother1rOgZ> +1
18:14:27 <geppetto> +1
18:14:27 <Rathann> +1
18:14:46 <limburgher> +1
18:16:20 <abadger1999> Okay, this passes
18:16:47 <abadger1999> #info Ruby Guideline update passed with changes: (+1: 6, 0:0, -1: 0)
18:16:56 <abadger1999> And that's all the time we have this week.
18:17:24 <abadger1999> Thanks for coming everyone!
18:17:25 <abadger1999> #endmeeting