16:00:13 <geppetto> #startmeeting fpc
16:00:13 <zodbot> Meeting started Thu Jul 14 16:00:13 2016 UTC.  The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:13 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:00:13 <zodbot> The meeting name has been set to 'fpc'
16:00:14 <geppetto> #meetingname fpc
16:00:14 <zodbot> The meeting name has been set to 'fpc'
16:00:14 <geppetto> #topic Roll Call
16:00:36 <orionp> hello
16:00:49 <geppetto> #chair orionp
16:00:49 <zodbot> Current chairs: geppetto orionp
16:00:52 <geppetto> hey
16:01:11 <geppetto> From what I remember a bunch of people can't make it this week, so it might be a short meeting
16:01:38 <geppetto> But apart from revisiting 637, ther'es not much on the todo
16:01:44 <geppetto> So not a big deal
16:02:01 <tibbs> Well, I'm actually around.
16:02:07 <geppetto> #chair tibbs
16:02:07 <zodbot> Current chairs: geppetto orionp tibbs
16:02:11 <geppetto> Ah, ha
16:04:36 <Rathann> hi, I'm actually around as well
16:04:49 <tomspur> hi
16:04:53 <geppetto> #chair tomspur
16:04:53 <zodbot> Current chairs: geppetto orionp tibbs tomspur
16:05:08 <tibbs> Oops.
16:05:12 <geppetto> #chair Rathann
16:05:12 <zodbot> Current chairs: Rathann geppetto orionp tibbs tomspur
16:05:25 * geppetto hates that C-w closes the tab in xchat now
16:05:42 <geppetto> Anyway ... we have 5 :-o
16:06:21 <geppetto> #topic Schedule
16:06:24 <geppetto> https://lists.fedoraproject.org/archives/list/packaging@lists.fedoraproject.org/message/ZBVHJMPOT55DZ2H6O4XD5B5EHYPR5IK5/
16:06:58 <geppetto> #topic #610  Packaging guidelines: Check upstream tarball signatures
16:07:01 <geppetto> .fpc 610
16:07:04 <zodbot> geppetto: #610 (Packaging guidelines: Check upstream tarball signatures) – fpc - https://fedorahosted.org/fpc/ticket/610
16:07:09 <geppetto> tibbs: Anything to discuss about this?
16:08:14 <tibbs> Well, if anyone has any opinions about what dwmw2 asked near the end of the ticket.
16:08:33 <tibbs> About where to store the signature files (lookaside or git).
16:08:45 <orionp> I tend towards git as well
16:09:02 <tibbs> I don't have an opinion, so really this is just waiting on me.  I doubt I'l lhave much time to worry about it until Flock.
16:09:09 <geppetto> tibbs: I'd heavily lean toward git as well
16:09:42 <tibbs> Is his asertion that we still use md5 for the lookaside checksums still true?
16:09:51 <tibbs> I thought we switched to SHA1.
16:10:02 <tibbs> Also, my ping is quite high at the moment.
16:11:21 <geppetto> just checked and it's still md5sum from what I can tell
16:12:11 <tibbs> I thought we had managed to get that fixed at some point.
16:12:19 <Rathann> I'm leaning towards git, too
16:12:26 <tibbs> It doesn't really matter anyway.  Stuffing them in git makes sense to me.
16:12:53 <tibbs> The lookaside should be used only for large stuff.
16:13:30 * geppetto nods
16:13:52 <geppetto> Getting diffs. and git blame etc. would be nice too
16:13:58 <tibbs> But that's just another thing on the pile.  I have my fingers in a few pies too many at this point.
16:14:21 * geppetto nods
16:14:23 <geppetto> #chair limburgher
16:14:24 <zodbot> Current chairs: Rathann geppetto limburgher orionp tibbs tomspur
16:14:34 * limburgher is late, and contrite
16:14:49 <limburgher> Where are we?
16:15:01 <geppetto> #topic #637  approval for a 'docker-latest' package on fedora
16:15:07 <geppetto> .fpc 637
16:15:08 <zodbot> geppetto: #637 (approval for a 'docker-latest' package on fedora) – fpc - https://fedorahosted.org/fpc/ticket/637
16:15:18 <geppetto> limburgher: just in time to start 637 :)
16:15:19 <tibbs> limburgher: Haven't really missed anything.
16:15:29 <geppetto> Although I've pinged Dan, and he isn't here
16:16:12 <tibbs> I think basically this is about two things:
16:16:26 <tibbs> how we name parallel-installable multi-version packages.
16:16:40 <tibbs> And whether we want to grant review process exceptions for them.
16:17:02 <tibbs> I'm in favor of the latter, and I think we whould at least try to fix something for the former.
16:17:55 <tibbs> Whether the maintainers want to make a "-latest" package, or always have the base one be the latest and split of versioned packages for the older ones, I would leave up to them.
16:18:39 <tibbs> is there anything else in that ticket that really involves us?  It's tough for me to keep context.
16:18:52 <orionp> I really don't like the precendent of a package named "-latest"
16:19:12 <geppetto> My objection to "-latest" is the same, it is at best a meaningless name but could easily be a very bad name N months from now
16:19:17 <orionp> And it would be nice to get some standardization of naming versioned packages
16:19:24 <Rathann> well, I think package named foo is considered "latest" in Fedora by default
16:19:40 <tibbs> Right, -latest seems to me to be not something we'd want most of the time.
16:19:40 <orionp> Rathann: ++
16:19:47 <geppetto> I'm somewhat happy for people to get exceptions for "usage" names ... Eg. this is the version needed by OBS or whatever
16:19:49 <Rathann> and we do have guidelines for naming parallel installable versions of the same
16:19:55 <tibbs> But I can see "foo" being stable and "foo-latest" being a git snapshot.
16:20:16 <tibbs> But then maybe another name woud be etter, like "foo-unstable".
16:20:24 <Rathann> well it really depends on what upstream considers stable
16:21:01 <Rathann> I have one non-Fedora example (FFmpeg), where upstream considers git HEAD stable and recommended
16:21:02 <geppetto> yeh, even foo-unstable is ok by me I think.
16:21:10 <geppetto> :)
16:21:19 <geppetto> If it works on my machine, ship it
16:21:43 * limburgher glares at geppetto
16:21:43 <Rathann> well, not in the case of FFmpeg, they have a testsuite which is run regularly
16:22:00 <Rathann> it doesn't provide 100% coverage, but it's better than nothing
16:22:10 * geppetto nods
16:22:28 <tibbs> Rathann: Right, which is why I'd just leave it up to the maintainers to decide how they want to run things.
16:22:40 <Rathann> anyway, I don't like -latest, but if it's well-founded, then I won't object
16:24:10 <limburgher> What if we did docker[numbers], and the most recent Provides: docker-latest?
16:24:25 <limburgher> Might be messy.
16:24:28 <geppetto> limburgher: yeh, that's what I suggested
16:24:59 <geppetto> I think Dan objected to the fact that there could be more than two versions at once that way
16:25:15 <geppetto> Maybe that the FS layout wouldn't match the package names either
16:25:18 <tibbs> Do they want there to be more than two versions at once, ever?
16:25:25 <limburgher> I JUST read that part. :)
16:25:31 <tibbs> If not, then I don't see why not -latest and -stable.
16:25:35 <geppetto> Or maybe just that he'd rather it just copied the RHEL solution, not sure.
16:25:56 <tibbs> I'm not entirely sure what the RHEL solution is, honestly.
16:26:01 <tibbs> But please don't say SCLs.
16:26:02 <geppetto> tibbs: No, AIUI they just want two seperate streams
16:26:08 <orionp> perhaps we should just wait until dwalsh or someone from the docker folks can discuss...
16:26:26 <geppetto> tibbs: they ship two packages in RHEL, docker which is the compat. one and docker-latest which is the newest
16:26:30 <orionp> Seems like the hangup is managing storage /var/lib/docker....
16:26:50 <tibbs> Are they intended to be parallel-installable?
16:26:54 <geppetto> yeh
16:27:21 <geppetto> One uses /var/lib/docker the other uses /var/lib/docker-latest
16:27:25 <geppetto> etc.
16:28:53 <tibbs> So personally I don't have a problem with that solution.
16:29:07 <orionp> though at the moment they seem to be the same version - 1.10.3-44
16:29:20 <tibbs> But I can't recall wat the objections to '-latest' were.
16:30:24 <Rathann> tibbs: the only technical objection might be that by the time a given Fedora version gets EOL, the -latest package might not be up-to-date with upstream anymore
16:30:36 <geppetto> <geppetto> My objection to "-latest" is the same, it is at best a meaningless name but could easily be a very bad name N months from now
16:30:47 <Rathann> for whatever reason (too old version of the dependencies, for example)
16:31:08 <geppetto> Yeh, roughly what Rathann said
16:33:28 <geppetto> ok, anyone want to vote on anything?
16:33:47 <tibbs> So if it's just semantics, what name is better?
16:34:04 <tibbs> And what would we recomend for versioned older packages?
16:34:06 <geppetto> I suggested a couple in the ticket
16:34:40 <tibbs> I guess that would go into the updated naming guidelines.
16:35:17 <Rathann> tibbs: either use what our current guidelines recommend (plain "docker" is latest and parallel installable other version is named docker1.x)
16:35:36 <Rathann> or, depending on what upstream recommends, docker-stable, docker-dev, whatever ;)
16:35:44 <Rathann> even docker-latest might be fine
16:36:44 <tibbs> I'm just trying to frame this in my head in terms of guidelines.
16:37:17 <tibbs> We'd recommend "foo" as the latest Fedora version, and fooX.YZ (no dash, but with the period).
16:37:46 <geppetto> Yeh, that's what we recommend now
16:38:14 <tibbs> Do we have that already?  I thought there was still some question about wheter you needed to have the period.
16:38:33 <orionp> https://fedoraproject.org/wiki/Packaging:Naming#Multiple_packages_with_the_same_base_name
16:38:46 <orionp> recommends the . now
16:39:17 <orionp> I still think we're just spinning without talking to the docker packagers...
16:39:30 <limburgher> agreed.
16:39:43 <orionp> I don't like -latest, but there may be issues with storage location migration
16:40:13 <orionp> going from /var/lib/docker1.10 to /docker1.11, etc.
16:40:26 <Rathann> orionp: docker (with no numbers) could use /var/lib/docker-latest and the versioned one could use /var/lib/docker, though that's not intuitive at all
16:40:42 <tibbs> Right, that's negatively intuitive.
16:41:08 <orionp> I just don't know docker and can't say what the issues are
16:41:31 <tibbs> I guess we just need to figure out a couple of different scenarios and recommend... something.
16:42:03 <orionp> I don't like -latest - it goes against our orientation, but I could be persuaded in this case if the issues are intractable
16:42:06 <tibbs> But for Fedora, I guess I still don't see the problem with '-latest'.
16:42:26 <tibbs> I mean, we don't care about what's in EOL or whether the package names represent reality.
16:46:30 <Rathann> after re-reading the ticket description it kind of makes sense to use -latest, but I'd still prefer docker and docker-stable
16:47:00 <geppetto> ok
16:47:21 <limburgher> +1.  We sort of assume a Fedora package to be the latest, usually, by default.
16:47:31 <Rathann> I mean, if openshift has a dependency on a specific docker version then let it have it (and have docker-stable provide docker = stable version)
16:47:38 <geppetto> #info No objections to using other namging than versioned, everyone but me seems ok with using docker-latest as the package name too if you want.
16:48:16 <geppetto> Ok, hopefully Dan will be here next week or update the ticket after reading the log.
16:48:21 <geppetto> #topic Open Floor
16:48:30 <geppetto> Anything anyone else wants to bring up?
16:48:53 <geppetto> I know we got a few trivialish tickets, I moved one or two to tibbs
16:49:12 <geppetto> another waiting on a real draft just so we aren't guessing at what they want
16:50:18 <tibbs> Yeah, I can write them up whenever I get the chance.
16:51:49 <geppetto> Cool, so anything else, if not I'll close in a couple of minutes
16:52:13 <tibbs> Hopefully I can get used to living on this laptop and try to get some of these actually written up.
16:52:28 <tibbs> BTW, does anyone know about pypi and the idiocy that has gone on there?
16:52:58 <misc> number80: ^
16:53:16 <tibbs> I just want to know what we should be using now.
16:53:30 <tibbs> But there appear to be at least two solutions for stable tarball links.
16:53:31 <limburgher> Nothing here.
16:53:59 <tibbs> I'll be happy to change the example spec in the guidelines and maybe toss something into SourceURL, but I don't know what we should actually use,
16:55:56 <geppetto> Nothing jumps out at me on a look at the website
16:56:29 <geppetto> we could just pick the one which looks more sane, and recommend that
16:56:42 <geppetto> someone will ping us if it stops working :)
16:56:57 <tibbs> Heh, the shorter one.
16:57:02 <tibbs> I can live with that.
16:57:06 <orionp> Apparently pypi.io is the "Next Generation Python Package Repository" for pypi, but I don't know if that url is stable
16:58:05 <Rathann> speaking of python, what happened to the automated dependency generator change?
16:58:22 <orionp> the links from the download pages contain hashes, but apparently there are non-hashed versions?
16:58:50 <tibbs> Rathann: I thought that was still being worked on.
16:59:01 <number80> tibbs: s/pypi.python.org/pypi.io/ according pypi admins
16:59:06 <tibbs> I hope that someone who understands what they're doing will have some guidelines for us soon.
16:59:35 <tibbs> number80: Ah, because the ticket someone opened suggested that we use something else.
16:59:51 <number80> it is stable, but it's unlikely to be the default pypi frontend for a while as they have a shortage of people
16:59:59 <tibbs> files.pythonhosted.org
17:00:41 <number80> works too
17:01:32 <tibbs> I'll take the shorter one.
17:01:38 <number80> *nods*
17:01:42 <geppetto> :)
17:01:57 <number80> I know that SUSE made the same choice (I'm working closely with them on a packaging project)
17:02:12 <tibbs> All the better.
17:03:29 <geppetto> ok, sounds like that's sorted
17:03:32 <geppetto> Anything else?
17:04:48 <geppetto> ok, closing
17:05:10 <geppetto> #endmeeting