16:00:17 <geppetto> #startmeeting fpc
16:00:17 <zodbot> Meeting started Thu Jul 26 16:00:17 2018 UTC.
16:00:17 <zodbot> This meeting is logged and archived in a public location.
16:00:17 <zodbot> The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:17 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:00:17 <zodbot> The meeting name has been set to 'fpc'
16:00:17 <geppetto> #meetingname fpc
16:00:17 <zodbot> The meeting name has been set to 'fpc'
16:00:17 <geppetto> #topic Roll Call
16:00:21 <mhroncok> hi
16:00:23 <tibbs> Hey folks.
16:00:27 <geppetto> #chair mhroncok
16:00:27 <zodbot> Current chairs: geppetto mhroncok
16:00:30 <geppetto> #chair tibbs
16:00:30 <zodbot> Current chairs: geppetto mhroncok tibbs
16:00:33 <decathorpe> .hello2
16:00:33 <zodbot> decathorpe: [hellomynameis decathorpe]
16:00:44 <geppetto> #chair decathorpe
16:00:44 <zodbot> Current chairs: decathorpe geppetto mhroncok tibbs
16:00:47 <redi> .hello2
16:00:47 <zodbot> redi: [hellomynameis redi]
16:00:52 <geppetto> #chair redi
16:00:52 <zodbot> Current chairs: decathorpe geppetto mhroncok redi tibbs
16:01:19 <ignatenkobrain> .hello2
16:01:19 <zodbot> ignatenkobrain: [hellomynameis ignatenkobrain]
16:01:23 <geppetto> #chair ignatenkobrain
16:01:23 <zodbot> Current chairs: decathorpe geppetto ignatenkobrain mhroncok redi tibbs
16:03:47 <tibbs> Down to 33 tickets.  So that's something.
16:04:10 <geppetto> 10 active
16:04:25 <geppetto> well, for the meeting
16:04:33 <geppetto> which is the one I care about ;)
16:04:48 <tibbs> I think most of the remaining ones just need to be closed at this point, but some have just fallen through the cracks.
16:05:09 <geppetto> We could try that, see what happens :)
16:05:17 <geppetto> #topic Schedule
16:05:20 <geppetto> https://lists.fedoraproject.org/archives/list/packaging@lists.fedoraproject.org/message/3F6PPWY22VAWUBZ5ITUJIBD6SN3RLVE2/
16:05:25 <tibbs> We should try to talk about everything tagged as meeting, or at least untag the ones we aren't going to talk about.
16:05:36 <geppetto> #topic #784 forbid globs for shared libraries as it conceals sonames
16:05:44 <geppetto> .fpc 784
16:05:44 <zodbot> geppetto: [showticket https://pagure.io/packaging-committee/issue/%s 784]
16:05:56 <geppetto> Ok, we talked about this a little in the open floor last week
16:05:56 <mhroncok> zodbot: go home, you are drunk
16:05:57 <ignatenkobrain> zodbot seems to be tired today
16:06:07 <tibbs> Nice, adding tags and adding a comment at the same time... doesn't always add the tags.
16:06:21 <tibbs> That might explain why some of the tickets never got tagged for a meeting.
16:06:45 <tibbs> https://pagure.io/packaging-committee/issue/784
16:07:00 <tibbs> I can imagine this one taking the whole meeting.
16:07:07 <ignatenkobrain> so for this one I tend to agree that packagers should explicitly specify SONAME and not glob it
16:07:15 <ignatenkobrain> heh
16:07:16 <tibbs> Well, "SHOULD".
16:07:24 <mhroncok> I also think it should be a SHOULD
16:07:25 <decathorpe> I have a question: does RPM extract the "real" version from the library for automatic provides and requires, or does it use the file name?
16:07:37 <mhroncok> decathorpe: filename AFAIK
16:07:43 <decathorpe> then the last comment is moot
16:07:45 <geppetto> decathorpe: extracts it
16:07:46 <redi> ???
16:07:50 <redi> it must get the real soname
16:07:53 <redi> not use the filename
16:07:59 <redi> or it's completely broken
16:08:05 * decathorpe shrugs
16:08:08 <ignatenkobrain> it extracts it
16:08:13 <geppetto> It does, it does if you rename to just .so
16:08:13 <redi> it MUST
16:08:15 <mhroncok> ok, sorry in that case
16:08:24 <ignatenkobrain> tools/elfdeps.c
16:08:24 <ignatenkobrain> 219:	    case DT_SONAME:
16:08:25 <ignatenkobrain> 318:     * For DSOs, add DT_SONAME as provide. If its missing, we can fake
16:08:30 <redi> phew
16:08:45 <geppetto> mhroncok: causing redi a minor heart attack ;)
16:08:50 <ignatenkobrain> if DT_SONAME is empty, the filename it is
16:08:57 <mhroncok> :D
16:09:00 <tibbs> Note: the draft in the initial comment is not the actual draft.
16:09:06 <decathorpe> okok, so ... we can't rely on the file name at all?
16:09:16 <redi> right
16:09:23 <decathorpe> yeah I updated it, and nobody seemed to notice ...
16:09:25 <tibbs> decathorpe: You should be able to edit the initial comment to either update the draft there or indicate that it's not current.
16:09:29 <mhroncok> well filenames with different versions in them, that's just wrong
16:09:37 <redi> right, and it's probably very rare
16:09:55 <redi> but it's incorrect to assume the filename is the SONAME, or the filename and the first digit  is the SONAME
16:09:59 <ignatenkobrain> that was the case for glibc until recently
16:10:15 <ignatenkobrain> libcrypt.so had SONAME for libcrypt-nss
16:10:17 <ignatenkobrain> or vice-versa
16:10:18 <ignatenkobrain> or something similar
16:10:22 <geppetto> wow
16:10:23 <ignatenkobrain> but yes, that's rare
16:10:57 <decathorpe> ok, I've updated the issue text directly
16:11:00 <mhroncok> and whether it's ok or not is for a different dicussion I guess
16:11:18 <redi> it's not Fedora's decision to make
16:11:27 <redi> imho
16:11:36 <tibbs> So I'm of two minds here.
16:12:05 <tibbs> On one hand, stricture here could have helped with several unannounced soname bumps just in the past several days.
16:12:22 <tibbs> On the other hand, I see no way to clearly state what people should and shouldn't do.
16:12:49 <mhroncok> I liked what somebody said on the ticket
16:12:55 <mhroncok> about requirement of rebuilds
16:13:05 <decathorpe> (that was me)
16:13:12 <tibbs> I'm actually leaning towards inverting the sense of the guideline:
16:13:25 <mhroncok> decathorpe: "What about something like: "don't use globs for file names or parts thereof, when a change in the matched part of the file name requires dependent packages to be rebuilt"?"
16:14:09 <tibbs> "Packagers SHOULD, if possible, structure the patterns in the %files section such that those patterns will need to be updated when the library version changes in such a way that rebuilds of other packages will be needed."
16:14:28 <mhroncok> tibbs: I like that, if examples are given
16:14:28 <redi> that's at least not incorrect
16:14:48 <tibbs> Right, examples are the key.
16:15:12 <tibbs> But also a certain vagueness; attempting to specify how the globs should look just leads down a rabbit hole of sorrow and flames.
16:15:27 <ignatenkobrain> I would go with something which references the SONAME word
16:15:44 <redi> yes
16:15:48 <ignatenkobrain> we already have SONAME described in guidelines somewhere
16:15:52 <tibbs> Oh, and what about libraries which don't reflect the soname in the filename?
16:15:58 <ignatenkobrain> we even have examples how to create library with soname
16:16:30 <decathorpe> tibbs: I don't think that's in the scope of this ticket ... one would have to use rpmabidiff(?) for that?
16:17:14 <ignatenkobrain> "Packagers SHOULD, if possible, structure the patterns in the %files section such that those patterns will need to be updated when the library changes in such a way that rebuilds of other packages will be needed (change of SONAME for instance)."
16:17:15 <tibbs> Well, the problem is that we're trying to prevent some specific effect (unannounced soname bumps).
16:18:11 <tibbs> And one important question (that doesn't necessarily need to be answered in the guideline) is: When is that not possible?
16:18:20 <mhroncok> and this particular guideline will only prevent it in some cases, probably in the majority of them
16:18:35 <decathorpe> upstream can always do nasty things
16:18:38 <mhroncok> which is fine
16:18:40 <tibbs> I don't know enough to say, but the case of a library which doesn't reflect the soname in the filename would probably be one such case.
16:18:41 <ignatenkobrain> decathorpe: abipkgdiff
16:19:08 <decathorpe> ignatenkobrian: right
16:19:10 <mhroncok> I don't think we need to design it an a way that would prevent all cases
16:19:21 <mhroncok> at least not now
16:19:24 <tibbs> It would probably be worth a sentence just after this guideline that suggests that people do run the relevant tool to look for such changes.
16:19:39 <geppetto> tibbs: nss just ships libnss3.so
16:20:00 <geppetto> that's the only example I can think of … but there are probably more
16:20:04 <decathorpe> absolutely. I didn't know about abipkgdiff until recently. It should be mentioned in the Update Guidelines or somewhere
16:20:58 <tibbs> I honestly don't even know how how to see what the soversion of a library is.
16:21:09 <ignatenkobrain> we should not try to solve this problem entirely, we just can't
16:21:10 <redi> readelf -d libfoo.so | grep SONAME
16:21:23 <ignatenkobrain> we should just tell people to be very careful when update packages
16:21:32 <mhroncok> ignatenkobrain: +1
16:22:12 <geppetto> ignatenkobrain: It might be cool if someone added something more usable … having both the old package and new one sucks a bit.
16:22:36 <geppetto> ignatenkobrain: I mean, I think we do … mostly. But automated checks better than telling people things :)
16:22:43 <tibbs> But SONAME of libnss3.so is just libnss3.so
16:22:50 <tibbs> Is ther no field that contais the version?
16:23:15 <redi> there's no such thing
16:23:22 <ignatenkobrain> tibbs: no
16:23:27 <ignatenkobrain> there is no such thing as version
16:23:28 <redi> the linker only cares about the SONAME
16:23:31 <redi> it's an opaque string
16:23:33 <tibbs> OK, so it's all more fail than I thought.
16:23:48 <geppetto> tibbs: They use symbol versioning … objdump -T
16:23:57 <redi> you could have a SONAME of "this_is_my_soname" and it would work fine
16:24:25 <redi> cmake's "soversion" is just a convention
16:24:54 <tibbs> So there's this thing which we try to conflate with a version and then conflate it again with a filename and we really hope that people will change it when somehow they're supposed to change it if they even realize that it needs to change.
16:24:58 <ignatenkobrain> I think we should make some agreement and move on 😉I almost liked tibbs' proposal
16:25:08 <redi> as I kept saying in the ticket, the filename is not the soname
16:25:27 <redi> they might be related, and usually are
16:25:44 <redi> but trying to define a guideline in terms of the filename along is not wise
16:25:50 <redi> s/along/alone/
16:25:56 <geppetto> yeh, tibbs thing that says "make sure it blows up when deps. need to rebuild" is the least objectionable
16:26:17 <tibbs> Except that.... the filename is all we have here.
16:26:43 <decathorpe> well, the only fail-safe way would be to run abipkgdiff?
16:27:06 <tibbs> Right, which we can of course mention.
16:27:10 <ignatenkobrain> yes, and I believe there are some automated checks which ru nit
16:27:18 <mhroncok> or if we can define a macro with the provide and the automatic provides would compare it
16:27:23 <geppetto> in fedora?
16:27:27 <ignatenkobrain> but it doesn't stop people from breaking things
16:27:30 <ignatenkobrain> geppetto: yes
16:27:33 <tibbs> The automated checks only run _after_ you have done your rawhide build and broken everything.
16:27:50 <mhroncok> ignatenkobrain: automated checks don't stop anyone. nobody reads the results, except maybe in bodhi
16:27:54 <decathorpe> (well, even bodhi checks didn't prevent a faulty update to be pushed to f27)
16:27:58 <mhroncok> tibbs: and thagt
16:27:58 <geppetto> tibbs: 👍
16:28:18 <ignatenkobrain> tibbs:  that's true, but you still have time to untag it until it hit compose 😃
16:28:20 <tibbs> I'm sure we all recognize that this is just a small bandage on top of a larger problem.
16:28:39 <mhroncok> yes
16:28:56 <tibbs> The people who will flame us for trying to put a solution in the guidelines will just ignore it anyway.
16:29:21 <mhroncok> can we maybe proceed on the bandage at least, acknowledging it's only partial?
16:29:28 <decathorpe> I don't think anyone would classify my suggestion as a "solution". But at least it would be something
16:29:36 <tibbs> Yes, so what do we do from here?
16:29:59 <tibbs> We have a couple of proposals, but mine was just tossed out there and isn't complete.
16:30:03 <geppetto> someone make a least worst proposal and we can all vote on it ?
16:30:04 <mhroncok> we add examples to tibbs's propsal modified by ignatenkobrain?
16:30:12 <decathorpe> I'm fine with that
16:30:14 <ignatenkobrain> "Packagers SHOULD, if possible, structure the patterns in the %files section such that those patterns will need to be updated when the library changes in such a way that rebuilds of other packages will be needed (change of SONAME for instance)."
16:30:36 <geppetto> Meh, +1
16:30:41 <tibbs> I can +1 that.
16:30:47 <tibbs> geppetto: But why "meh"?
16:31:14 <mhroncok> +1 provided that examples are added
16:31:26 <geppetto> tibbs: As I said, I'm not a fan of this bandage … but it says should
16:31:28 <redi> +1 modulo examples
16:31:35 <decathorpe> +1 with some examples
16:31:44 <tibbs> Certainly we can work on examples in the ticket if we accept the basic concept.
16:31:51 <decathorpe> And maybe a wall of shame of packages that do this? ;)
16:31:57 <mhroncok> tibbs: that's what I had in mind
16:31:58 <ignatenkobrain> +1 with some examples to tibbs' proposal modified by me :)
16:32:10 <ignatenkobrain> decathorpe: should be special wiki page
16:32:13 <redi> my suggestion's already in the ticket:  Example: if the SONAME is libfoo.so.11 then %_libdir/libfoo.so.* SHOULD NOT be used, but %_libdir/libfoo.so.11{,.*} is OK.
16:32:15 <ignatenkobrain> and then give up a badges ;)
16:32:25 <tibbs> decathorpe: Well, if there is a basic detection script then I will add it to the set of reports that I do occasionally.
16:32:27 <redi> (although that's phrased as SHOULD NOT and the new suggestion inverts it)
16:32:46 <tibbs> Just send it to the misc-fedora-package-utilities repo.  I'll give you commit access if you want.
16:32:53 <geppetto> #action forbid globs for shared libraries as it conceals sonames (+1:6, 0:0, -1:0)
16:33:05 <decathorpe> I pushed the script to a pagure repo and linked to it in the ticket
16:33:17 <geppetto> #action Need examples added to the proposal voted on.
16:34:02 <tibbs> Well, OK, but it's easier for everyone if these things are in as few random repos as possible.
16:34:34 <ignatenkobrain> geppetto: so what's next?
16:34:36 <ignatenkobrain> on our radar
16:34:41 <geppetto> #topic #782 Forbid %{pythonX_site(lib|arch)}/* in %files
16:34:45 <geppetto> .fpc 782
16:34:45 <zodbot> geppetto: [showticket https://pagure.io/packaging-committee/issue/%s 782]
16:34:52 * limburgher here finally
16:34:56 <geppetto> #chair limburgher
16:34:56 <zodbot> Current chairs: decathorpe geppetto ignatenkobrain limburgher mhroncok redi tibbs
16:35:04 <tibbs> I guess zodbot broke that when they updated it recently.
16:35:14 <mhroncok> https://pagure.io/packaging-committee/issue/782
16:36:14 <geppetto> mhroncok: what does you repoquery thing show?
16:36:45 <mhroncok> geppetto: about ownling / ?
16:36:48 <tibbs> So last week we agreed to basically saying "MUST NOT own the toplevel __pycache__"
16:37:00 <geppetto> ahh
16:37:04 <mhroncok> or about pycache?
16:37:32 <geppetto> yeh, the __pycache__ thing … but I see now
16:37:36 <mhroncok> plenty of packages own pycache currently, I cannot do the qery now
16:37:38 <tibbs> I believe we have mostly fixed the fallout from %python_sitelib and _sitearch being empty.
16:37:56 <mhroncok> tibbs: I found ~10 more, PRs open
16:38:07 <mhroncok> will merge on Friday
16:38:44 <tibbs> Honestly we should just be fixing these.
16:39:05 <tibbs> 61 binary packages currently own /usr/lib/python3.7/site-packages/__pycache__
16:39:06 <mhroncok> I just open PRs and say I'll merge them in 3 days
16:39:40 <mhroncok> for those 61 packages, I'll try to invent a fixer script
16:39:42 <tibbs> Since they can possibly be breaking the distro, I've been fixing them as I notice them.
16:39:59 <tibbs> Another 22 packages own /usr/lib64/python3.7/site-packages/__pycache__
16:40:03 * geppetto nods … AFAIK there isn't anything to discuss here, right? Just BZs/PRs to file?
16:40:18 <tibbs> Well, mhroncok indicated that he originally wanted other changes.
16:40:32 <tibbs> I believe a prohibition on using %{python2_sitelib}/*
16:40:37 <mhroncok> yes
16:41:21 <mhroncok> I don't want to fix all the packages with that, but I'd like to forbid it for the future ones
16:41:24 <tibbs> And I may try to restructure slightly so that the current restriction on owning __pycache__ isn't in the section on byte compilation.
16:41:42 <mhroncok> it tends to get updated to %{python3_sitelib}/* and that often leads to the __pycache__ thing
16:42:10 <tibbs> Personally I'm just not sure that's a good reason to ban it.
16:42:53 <mhroncok> it it dangerous in a similar way to the soname thing. random stuff keeps popping there
16:42:59 <tibbs> In most cases a ban wouldn't lead to much additional complexity in the %files section, I guess, but saying "It's not wrong, but it could lead to you doing something that's wrong in the future" isn't the greatest of reasons.
16:43:05 <mhroncok> such as %{python2_sitelib}/tests
16:43:30 <tibbs> The soname thing leads to distro breakage, though.  This... doesn't.
16:43:35 <mhroncok> right
16:43:47 <mhroncok> I'm ok if we say shouldn't
16:43:58 <mhroncok> or that it's not necouraged
16:44:08 <mhroncok> or something like that
16:44:14 <tibbs> So what would this look like?  Need something concrete to vote on.
16:44:28 <ignatenkobrain> SHOULD NOT works for me as well
16:45:13 <mhroncok> Packages should not list %{pythonX_sitelib}/* (and sitearch) in %files section, as it can easily lead to unnoticed errors.
16:45:48 <ignatenkobrain> I would do s/list/use/, but either way I'm +1
16:45:56 <tibbs> I can +1 that.
16:46:00 <decathorpe> +1 from me too
16:46:14 <ignatenkobrain> and should not should be upper case ;)
16:46:16 <mhroncok> A more specific globs are recommended, such as %{pythonX_sitelib}/modname*.
16:46:20 <limburgher> +1
16:46:39 <redi> +1
16:46:58 <mhroncok> I'm +1 obviously
16:46:59 <limburgher> PythonX: For Xtreme Development
16:47:02 <geppetto> +1
16:47:19 <tibbs> But when writing this up, I would want to lift these restrictions up to either a separate section or merge with the "Files to include section".
16:47:38 <tibbs> And don't worry about case of should and must; I can fix that all up when doing writeups.
16:47:56 <mhroncok> (whatever +1 even on that)
16:48:09 <geppetto> #action Recommend not listing  %{pythonX_sitelib}/* (and sitearch) in files (+1:7, 0:0, -1:0)
16:48:40 <geppetto> #topic #719 Simplify packaging of forge-hosted projects
16:48:48 <geppetto> https://pagure.io/packaging-committee/issue/719
16:49:05 <geppetto> So tibbs had some confusion here, or stuff didn't work out?
16:49:17 <decathorpe> No, the draft isn't up to date
16:49:35 <mhroncok> and the submitter seems not responsive
16:49:38 <decathorpe> I asked for an updated one two months ago
16:49:44 <mhroncok> the PR on redhat-rpm-macros is stalled
16:49:59 <tibbs> I'm going to merge the PR today.
16:50:05 <mhroncok> https://src.fedoraproject.org/rpms/redhat-rpm-config/pull-request/24
16:50:06 <tibbs> Was going to do it yesterday but got distracted.
16:50:39 <tibbs> Note that I have actually used the forge macros in a couple of packages and... it's certainly much nicer than doing it all by hand.
16:51:17 <tibbs> Note that nim posted to fedora-devel just a few hours ago.
16:51:25 <ignatenkobrain> so what are we voting on?
16:51:26 <geppetto> yay! … success? improvements? :)
16:51:33 <tibbs> Nothing to vote on, I think.
16:51:45 <geppetto> Ok, with your update I wasn't sure if we needed to change something
16:51:48 <tibbs> I just tossed in a comment because I had actually used the thing and wanted to give feedback.
16:52:07 <redi> which I said I'd do weeks ago and never did
16:52:12 <tibbs> I will probably copy the stale draft and start hammering it into shape.
16:52:29 <ignatenkobrain> sounds good to me
16:52:32 <geppetto> ok
16:52:41 <mhroncok> tibbs: that would be great, thank you
16:53:01 <tibbs> But I'm going to be away on vacation soon so I don't know how much progress I'll make in the near term.
16:53:16 <tibbs> But after using these things, they saved me time and so I like them.
16:53:22 * geppetto nods
16:53:38 <geppetto> Ok, then …
16:53:40 <mhroncok> it's stalled for months, I don't think it's time sensitive
16:53:42 <tibbs> Too bad the vim specfile mode doesn't handle them.
16:53:46 <geppetto> #topic Open floor
16:54:02 <tibbs> There was one old ticket which never got tagged meeting for some reason.
16:54:18 <tibbs> https://pagure.io/packaging-committee/issue/703
16:54:27 <tibbs> Had a draft an everything, but has sat there for a year.
16:54:56 <ignatenkobrain> I'm leaning towards preferring upstream's name
16:55:17 <mhroncok> I'm leaning towards lowercase
16:55:18 <tibbs> Should definitely be lower case except for specific things.
16:55:40 <tibbs> Where the specific thing is basically Perl because we've always done it that way.
16:55:45 <mhroncok> exactly. I mean for example Flask is uppercase on pypi, yet I'd prefer python-flask over python-Flask
16:55:58 <tibbs> The case contributes nothing useful except to make it harder to type when you're typing.
16:56:06 <mhroncok> exactly
16:56:21 <mhroncok> and I think more people would try lowercase if they cannot find CamelCase
16:56:24 <tibbs> Since packages with differing case can't coexist anyway, there is absolutely zero benefit.
16:56:27 <ignatenkobrain> sudo dnf repoquery --qf="%{NAME}\n" | grep "[[:upper:]]" | wc -l                                                                                                         18:56:08
16:56:27 <mhroncok> than the other way around
16:56:27 <ignatenkobrain> Last metadata expiration check: 2:05:16 ago on Thu 26 Jul 2018 04:50:58 PM CEST.
16:56:28 <redi> I think if upstream is Upper the fedora pkg should be "upper" but if it's UPper or UPPer UpPer the package can use that
16:56:28 <ignatenkobrain> 5053
16:57:01 <mhroncok> ignatenkobrain: grep -v perl
16:57:18 <tibbs> In any case, note that the whole "Package names should be in lower case" is already there; I just failed to drop the second part when I wrote this up ages ago for whatever dumb reason.
16:57:51 <ignatenkobrain> sudo dnf repoquery --qf="%{NAME}\n" | grep "[[:upper:]]" | grep -v perl | wc -l                                                                                          18:56:15
16:57:52 <ignatenkobrain> Last metadata expiration check: 2:06:46 ago on Thu 26 Jul 2018 04:50:58 PM CEST.
16:57:53 <ignatenkobrain> 2028
16:57:54 <ignatenkobrain> mhroncok: ^
16:58:07 <mhroncok> ignatenkobrain: thanks. still a lot
16:58:16 <ignatenkobrain> why don't we just make dnf to be case-insensitive instead? ;)
16:58:22 <tibbs> I don't think anyone is talking about mass-renaming these things.
16:58:36 <mhroncok> ignatenkobrain: it would be nice
16:58:38 <tibbs> If dnf is case insensitive then great, we should definitely force the actual package name to lower case then.
16:59:13 <mhroncok> dnf being ci would also produce a lot of inconsistency in requires specs
16:59:29 <mhroncok> somebody would use the actaul package name, somebody lowercase
16:59:50 <tibbs> Not really, no.
17:00:14 <tibbs> You make the dnf command line be case insensitive, not the dependency resolver.
17:00:17 <mhroncok> we'd need more guidelines for that
17:00:23 <mhroncok> tibbs: oh, ok
17:00:32 <mhroncok> that'll be nice
17:00:40 <mhroncok> and wouldn't need a change from our part
17:00:42 <mhroncok> I guess
17:01:03 <tibbs> But anyway, we have the dnf we have, and we still have this bad contradiction in the guidelines.
17:01:26 <geppetto> we'd need dnf to reject packages that are the same name but differ in case only … if it did that I'd happily +1
17:01:58 <mhroncok> ok, what are we voting? that we say lowercase and be done with it?
17:02:19 <ignatenkobrain> what would be the exceptions?
17:02:23 <tibbs> Well we already do say lowercase, but we do it in a confusing way.
17:02:27 <geppetto> I think so … that's generally been best practice forever … but there were a few outliers
17:02:29 <mhroncok> what zbyszek says - +1
17:02:50 <tibbs> I could draft a more comprehensive change if folks would like that.
17:02:54 <geppetto> sure
17:03:03 <geppetto> we are technically over
17:03:19 <tibbs> My penance for not fixing this properly back when we first agreed on "just use lower case".
17:03:35 <geppetto> #action tibbs will draft a more comprehensive change to say "use lowercase"
17:03:58 <mhroncok> and we vote on that once ready, right?
17:04:06 <geppetto> yeh
17:04:08 <geppetto> #undo
17:04:08 <zodbot> Removing item from minutes: ACTION by geppetto at 17:03:35 : tibbs will draft a more comprehensive change to say "use lowercase"
17:04:25 <geppetto> #action tibbs will draft a more comprehensive change to say "use lowercase" for 703
17:04:59 <geppetto> Ok, unless anyone really needs to talk about something else I'll end the meeting at 5 past.
17:05:30 <tibbs> I'm pretty much done.  There are some other tickets tagged as meeting; I will either ping them, untag them or just close them.
17:05:46 <geppetto> #endmeeting