16:03:10 <spot> #startmeeting Fedora Packaging Committee
16:03:10 <zodbot> Meeting started Thu May  9 16:03:10 2013 UTC.  The chair is spot. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:03:10 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:03:13 <spot> #meetingname fpc
16:03:13 <zodbot> The meeting name has been set to 'fpc'
16:03:17 <spot> #topic Roll Call
16:03:39 * geppetto is here
16:04:12 * limburgher is here but will be AFK some portion of 1615-1645.
16:06:09 * rdieter can pretend to be here, esp if it helps make quorum
16:06:16 * abadger1999 hree
16:06:22 <Rathann> here
16:06:38 <limburgher> abadger1999 is apparently a pantomime horse.
16:07:04 <spot> tibbs: ping ?
16:07:23 <spot> SmootherFrOgZ: ping
16:08:06 * SmootherFrOgZ here
16:08:50 * SmootherFrOgZ need to update his calendar
16:08:51 <spot> Okay, lets get down to it.
16:09:03 <spot> #topic Bundling exception for agg in mapnik - https://fedorahosted.org/fpc/ticket/279
16:09:39 <spot> Summary: Mapnik used to use system agg, but they have made mapnik specific changes. agg is long dead upstream.
16:09:45 * abadger1999 has a recollection of agg coming up before.
16:09:58 <Rathann> no mention of any attempt to use the fork as replacement for the bundled library or to send mapnik specific changes to the new upstream
16:10:09 <Rathann> maybe they're not so specific
16:10:36 <Rathann> i.e. I'd like to see the standard questions answered with respect to the fork/new upstream
16:10:54 <SmootherFrOgZ> hmm...so, can prentend the bundled version could be the new update of the system one?
16:11:09 <spot> looks like there are a few things that depend on agg in fedora (desmume, gnash, mapnik, python-matplotlib)
16:11:16 <geppetto> Yeh, it seems like they should either merge to the fork … or at worst start their own fork and put that in Fedora.
16:11:38 <SmootherFrOgZ> can we pretend*
16:12:22 <rdieter> SmootherFrOgZ: yes, per Rathann's initial comment
16:12:49 <limburgher> Speaking as agg maintainer, I'm not opposed to that, if it doesn't blow up the other deps.
16:13:03 * rdieter would like the standard questions answered too
16:13:15 * SmootherFrOgZ nods
16:13:21 <spot> I'll ask the questions in the ticket we have posed: "Would it be possible to use the 2.4 fork? Have either you or mapnik tried to send changes to the fork upstream? Alternately, would it be possible to apply mapnik's changes to the system agg package?"
16:13:24 <rdieter> limburgher: I take it they didn't talk to you yet either?
16:13:55 <spot> Any other questions we want answered?
16:14:09 <Rathann> SmootherFrOgZ: I'd rather see the mapnik devs work on merging their changes with the fork than update current Fedora agg to the version bundled in mapnik
16:14:25 <limburgher> rdieter:  Tom ran it by me, but I didn't have details.
16:14:28 <SmootherFrOgZ> Rathann: sure thing.
16:15:49 <spot> okay, i'm not hearing other questions
16:15:53 <spot> so i'll ask them and move on. :)
16:16:20 <spot> #action Asked clarifying questions
16:16:32 <spot> #topic New Python Macros for Easier Packaging - https://fedorahosted.org/fpc/ticket/281
16:16:57 <spot> I believe the only macro up for discussion today is %python_default_filter
16:17:07 <abadger1999> I'm +1 to that.
16:17:18 <abadger1999> Yeah, the other portion needs discussion on the ml.
16:17:22 <geppetto> +1
16:17:23 <rdieter> +1
16:17:29 <abadger1999> (it's been started... I need to weigh in)
16:17:54 <spot> +1
16:18:14 <spot> (fwiw, all it does is exclude %{python*_sitearch}/.*\\.so)
16:18:27 <SmootherFrOgZ> +1
16:18:52 <Rathann> +1
16:19:23 <spot> okay, thats +5
16:19:48 <spot> #action %python_default_filter approved - (+1:5, 0:0, -1:0)
16:20:30 <spot> #topic Bundling exception for gpick (lua) - https://fedorahosted.org/fpc/ticket/282
16:21:07 <geppetto> -1, I understand their frustration with lua not having been updated yet … but just bundling lua in 666 packages isn't the answer.
16:21:07 <spot> It seems like we just need to update lua pretty badly.
16:21:26 <Rathann> -1 from me as well, for the same reason
16:21:26 * spot is going to provenpackager lua up to current in rawhide unless i hear a good reason not to
16:21:52 <geppetto> As long as rpm works, that seems fine by me :)
16:21:56 <Rathann> maybe gather all the interested parties and have them co-maintain lua
16:22:52 <SmootherFrOgZ> -1 as well and waht geppetto said.
16:22:55 <SmootherFrOgZ> what*
16:23:33 <limburgher> I'm back.
16:23:36 * limburgher reads
16:23:45 * limburgher applauds spot's PPing lua
16:23:57 <limburgher> +1 on 281 if it matters
16:24:03 <spot> I think it is probably too late for f19 though.
16:24:03 <limburgher> -1 on 282
16:24:32 <abadger1999> -1 282 ; +1 spot moving the update forward in rawhide
16:24:33 <limburgher> spot: For something of that magnitude, I agree, unless it puts out a major fire.
16:25:07 <spot> -1 on 282
16:25:42 <spot> #action Bundling exception denied, spot is going to update lua in rawhide (+1:0, 0:0, -1:6)
16:26:22 <spot> #topic Specify that systemd .service .socket .timer, should not be marked executable - https://fedorahosted.org/fpc/ticket/283
16:27:24 <spot> basically, the original request was to add this wording to the Systemd page:
16:27:31 <spot> "Unit files should be owned by user:group root:root, with permission mask of 0644."
16:27:33 <geppetto> I don't see the point of this … do we put this in policy anywhere else, when it doesn't really matter?
16:28:20 * spot thinks this is a bit pedantic, these files aren't going to do anything at all if they are exec'd
16:28:50 <geppetto> right, I mean in theory someone could be clever and make it do something … but in general it's a noop.
16:28:55 <spot> systemd will be able to exec services/sockets/timers no matter the ownership
16:29:35 <spot> so this is just a "cleanliness" issue. If they want to submit an rpmlint check, hey, knock yourself out. :)
16:29:55 * spot isn't opposed to this, i just don't see the reason to do it.
16:30:13 <abadger1999> yeah -- If rpmlint errors for it then it can fall under the "justify rpmlint output" guideline.
16:30:37 <geppetto> sure
16:30:49 <rdieter> rahul's followup comment (about permissions in general) seems like a better approach too
16:30:49 * abadger1999 would rather see a "files should default to 0644" style guideline than this
16:30:50 <limburgher> I agree that rpm lint is a better place for this.  I'm having a hard time conceptualizing the harm in someone ./foo.service ing.
16:30:55 <abadger1999> but that's already rpm's default
16:31:49 * spot proposes that we simply ask them to submit this as an rpmlint check, since ownership and exec perms do not prevent proper operation.
16:32:04 <Rathann> yeah, no point in adding "files must be root:root 0644" to every domain-specific guideline
16:32:14 <limburgher> spot: Or open up a security issue.
16:32:27 <abadger1999> <nod>
16:32:29 <limburgher> Rathann: Exactly.
16:32:30 <abadger1999> spot: +1
16:32:46 <spot> +1 from me
16:32:54 <Rathann> +1
16:32:56 <limburgher> +1
16:32:58 <SmootherFrOgZ> +1
16:32:59 <rdieter> +1
16:33:21 <spot> i see +6, if geppetto wants to get in on this... :)
16:34:00 <geppetto> We are voting on telling them to just submit the rpmlint check?
16:34:05 <geppetto> If so +1
16:34:05 <spot> yep.
16:36:56 <spot> #action rpmlint check fine, guidelines change no (+1:7, 0:0, -1:0)
16:37:18 <spot> #topic github source URL: improved tarball names - https://fedorahosted.org/fpc/ticket/284
16:37:54 <smani> (I'm here in case there are any questions)
16:38:18 * spot could swear i just did a github download the other day that worked
16:39:00 <spot> yeah, i did librecad
16:40:24 <limburgher> So. . .is this then not needed, or does librecad comply with the new format?
16:40:40 <spot> well, i'm just not seeing the failure
16:40:59 <spot> e.g. https://github.com/LibreCAD/LibreCAD/archive/f90eb457b36727374f3609d4b37d0f3c47c1c057/librecad-2.0.0beta5-f90eb45.tar.gz works
16:41:13 <Rathann> suggesting alternative ways is fine, mandating them, not so
16:41:16 <limburgher> Wasn't there, once upon a time, a script someone had that checked for Bad SourceURL?
16:41:20 <geppetto> I also though the last change was an rpm hack to specify <url>#<filename> … how can a server change break that?
16:41:54 <smani> spot: the name of the file which gets downloaded by the link above is $name-$commit-tar.gz, not librecad-2.0.0beta5-f90eb45.tar.gz
16:42:07 <spot> smani: are you sure?
16:42:09 <misc> limburgher: rpmlint does check , with some option (IIRC)
16:42:23 <spot> when i use wget, i get librecad-2.0.0beta5-f90eb45.tar.gz
16:42:30 <spot> 2013-05-09 12:40:14 (442 KB/s) - ‘librecad-2.0.0beta5-f90eb45.tar.gz’ saved [15199613/15199613]
16:42:36 <limburgher> misc:  Right, but I recall a "do the whole disto and mass file BZ" sort of thing.
16:42:36 <smani> use wget --content-disposition
16:42:40 <abadger1999> Just checked... wget https://github.com/fedora-infra/fas/archive/e64034ee729b241aa4521f3bc40b3e2593248153/fas-0.8.17-e6403e.tar.gz  works and outputs a file fas-0.8.17-e6403e.tar.gz ; the toplevel directory inside is
16:42:46 <abadger1999> fas-e64034ee729b241aa4521f3bc40b3e2593248153/
16:43:18 <smani> or i.e. if you download the file from firefox, the filename changes
16:44:00 <spot> so we can either tell people to get the tarball from github with wget (no --content-disposition) or we can adopt this proposal to eliminate naming confusion
16:44:38 <smani> (afaic content disposition just tells wget to not rename the file based on what's in the URL one provided to wget)
16:44:47 <spot> smani: yep
16:46:10 <spot> what's the "ahead" for?
16:46:22 <smani> how many commits ahead of the last tag
16:46:49 <spot> ugh, that number is going to be dynamic.
16:47:12 <abadger1999> If you're getting into ahead...
16:47:22 <abadger1999> Then you're supposed to use the snapshot guidelines...
16:47:39 <spot> abadger1999: its not that, its that github embeds ahead in there
16:47:50 <limburgher> It's both, really.
16:48:14 <abadger1999> Alternately, if you are using a specific revision from github that is either a pre-release revision or a post-release revision, you must follow the "snapshot" guidelines.
16:48:15 <spot> for librecad, the name off the tarball/ url ends up being: LibreCAD-LibreCAD-2.0.0beta5-0-gf90eb45.tar.gz
16:48:27 <spot> thats on the tag.
16:48:34 <abadger1999> although -- now that I read that in this context, I can see how people would interpret that as just being about naming...
16:48:36 <spot> see the ahead of 0.
16:49:38 <limburgher> Oh github, is there *anything* you can't complicate?  /me beams romantically
16:50:35 <abadger1999> /archive/ doesn't have ahead
16:50:53 <abadger1999> archive gives: fas-e64034ee729b241aa4521f3bc40b3e2593248153.tar.gz
16:50:59 <abadger1999> (with --content-disposition)
16:51:29 <spot> yeah, thats an option.
16:51:44 <limburgher> I feel like making up a PSA about a URL being a "forever" URL.
16:51:49 <spot> it does mean we don't get a tag embedded in it.
16:51:54 <abadger1999> <nod>
16:52:14 <spot> but the commit should be unique enough.
16:52:29 * abadger1999 notes that the tag was only in the filename before.. not in the toplevel directory's name.
16:52:53 <spot> I think just switching it to the $project-%{commit}.tar.gz is sufficient.
16:53:16 <spot> That said, I don't have any issue with what smani has done, just not what i'd recommend as the default.
16:53:41 <smani> personally I find it usefull to see how near we the archive is to a certain commit, but well, personal preference I guess :)
16:54:24 <spot> I think if we did go with a version that had ahead, we'd need to add wording to tell people that if ahead > 0, you need to handle it as a snapshot.
16:55:35 <spot> anyone else want to chime in on this?
16:56:02 * abadger1999 doesn't like ahead
16:56:27 * rdieter doesn't like ahead either
16:57:04 <abadger1999> I also think that we should push people towards using the Revision Control section http://fedoraproject.org/wiki/Packaging:SourceURL#Using_Revision_Control  unless they're actually retrieving a release.
16:57:30 <rdieter> agreed
16:58:02 <limburgher> Hear hear.
16:58:06 <spot> okay then, i propose we adjust the pathing in the spec from "%{name}-%{version}-%{shortcommit}.tar.gz" to "$PROJECT-%{commit}.tar.gz"
16:58:26 <spot> that way, no matter how the thing gets downloaded, it has the right name.
16:59:07 <abadger1999> +1
16:59:32 <limburgher> +1
16:59:36 <SmootherFrOgZ> +1
16:59:38 <geppetto> 0
16:59:43 <spot> +1
17:00:04 <Rathann> 0
17:00:32 <rdieter> i guess that sucks a little less, +1
17:01:02 <spot> #action url recommendation changed to $PROJECT-%{commit}.tar.gz (+1:5, 0:2, -1:0)
17:01:17 <spot> smani: thanks for your help
17:01:22 <smani> welcome
17:01:45 <spot> We're at the hour mark (and I am starving). :)
17:01:53 <spot> #topic Thanks rdieter
17:02:14 <spot> I believe we have pretty unanimously agreed on the new FPC member, I will notify them today
17:02:27 <spot> But I wanted to thank Rex for his time on the FPC. :)
17:02:38 <abadger1999> Thank you Rex!
17:02:42 * limburgher waves
17:02:50 * Rathann waves too
17:02:54 <Rathann> big thanks
17:02:59 <geppetto> Yeh, have fun :)
17:03:03 <rdieter> thanks, it's been fun.  looking forward to seeing the new blood
17:03:09 * SmootherFrOgZ hugs rdieter
17:03:10 <spot> #topic Open Floor
17:03:13 <abadger1999> it's been a pleasure having you with us -- I'd be happy to have you back if you have the time :-)
17:03:18 <limburgher> Oh, god, it won't be THAT bad. . .
17:03:47 <rdieter> abadger1999: i promise, if I ever get bored... :)
17:03:47 <abadger1999> limburgher: thanks for doing a lot for the .desktop file cleanup.
17:04:17 <limburgher> abadger1999:  No sweat, anytime.
17:05:07 <limburgher> Wish more people had played along. . .
17:05:51 <limburgher> Maybe I oversold it in my blog post. . .
17:06:32 <abadger1999> yeah, mether, mamoru, paragn, bruno wolff, and you did the majority, I think.
17:07:04 * spot hears no new topics, so I'm going to close this out and go find lunch.
17:07:06 <spot> thanks everyone!
17:07:08 <spot> #endmeeting