17:00:16 <geppetto> #startmeeting fpc
17:00:17 <zodbot> Meeting started Thu Jan 28 17:00:16 2016 UTC.  The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:17 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:00:17 <zodbot> The meeting name has been set to 'fpc'
17:00:17 <geppetto> #meetingname fpc
17:00:17 <geppetto> #topic Roll Call
17:00:17 <zodbot> The meeting name has been set to 'fpc'
17:00:27 <tibbs|w> Howdy.
17:00:41 <geppetto> Hey
17:00:43 <geppetto> #chair tibbs
17:00:43 <zodbot> Current chairs: geppetto tibbs
17:00:44 <tibbs|w> It has been a horrible week for me.
17:00:50 <smdeep> .hello smdeep
17:00:51 <zodbot> smdeep: smdeep 'Sudeep Mukherjee' <smdeep@gmail.com>
17:01:11 <tibbs|w> smdeep: Welcome.
17:01:12 <geppetto> smdeep: Hey Sudeep
17:01:17 <smdeep> Hey all
17:03:20 <orionp> barely here - major firewall work going on...
17:03:32 <geppetto> Ok, you want me to chair you or not?
17:03:46 <geppetto> #chair orionp
17:03:46 <zodbot> Current chairs: geppetto orionp tibbs
17:03:54 <geppetto> Well, count you as a chair for quorum
17:04:29 <tibbs|w> Yeah, I'm still straightening out my RHEL7.2 NFS problems.
17:04:35 <tibbs|w> So I'm in and out too.
17:04:44 <geppetto> Ok, well we can just cancel this week
17:04:56 <geppetto> Just a reminder that I won't be here next week
17:05:20 * racor is here
17:05:24 <geppetto> #chair racor
17:05:24 <zodbot> Current chairs: geppetto orionp racor tibbs
17:05:42 <geppetto> I don't see anything super important on the schedule
17:05:56 <tibbs|w> I'm around enough if we have something to discuss.
17:06:00 <geppetto> I'm assuming the openshift thing doesn't need to happen Right Now(tm)
17:06:17 <tibbs|w> I'm not entirely sure I'll be able to chair next week.
17:06:32 <tibbs|w> And definitely not on the 11th (as the 10th is my birthday and I'll be out of town).
17:06:54 <geppetto> Ok
17:07:18 <tibbs|w> I'd hate to cancel all three meetings so if we can do anything today....
17:07:39 * geppetto nods … well we don't even have technical quorum yet
17:07:43 * tomspur is here, sorry got distracted
17:07:47 <geppetto> #chair tomspur
17:07:47 <zodbot> Current chairs: geppetto orionp racor tibbs tomspur
17:07:52 <geppetto> Ok, we have technical quorum now
17:08:00 <geppetto> We'll see how it goes.
17:08:11 <geppetto> #topic Schedule
17:08:13 <geppetto> https://lists.fedoraproject.org/archives/list/packaging@lists.fedoraproject.org/message/AJ43MQ6THTTICHXIFTYQXYUDAU2UNXX5/
17:08:19 <geppetto> #topic #591 Description of filtering macros in Perl is outdated
17:08:24 <geppetto> .fpc 591
17:08:26 <zodbot> geppetto: #591 (Description of filtering macros in Packaging:AutoProvidesAndRequiresFiltering#Perl is outdated) – fpc - https://fedorahosted.org/fpc/ticket/591
17:08:53 <geppetto> We spoke about this last week … but wanted to wait a week for more votes
17:08:56 <tibbs|w> Nothing happened as far as I can see.
17:09:36 <geppetto> Yeh, last edit was by you last week
17:09:55 <geppetto> #topic #593 use RPM tags for langpacks description
17:09:57 <racor> I have been investigating, but haven't come with a conclusion, yet
17:10:03 <geppetto> .fpc 593
17:10:04 <zodbot> geppetto: #593 (use RPM tags for langpacks description) – fpc - https://fedorahosted.org/fpc/ticket/593
17:12:15 <tibbs|w> Sorry, on 591, is there anything we actually _need_ to change now?
17:12:40 <paragan> Hi, I am here to answer any queries about langpacks guidelines. This guideline is to ask langpack providing packages to add Supplements: tag in their langpack subpackages. That
17:12:40 <paragan> way when basic language meta-package like langpacks-ja is installed, any base package installation will pull ja langpacks if they are available.
17:12:41 <geppetto> I mean … I don't think there was anything that was wrong
17:12:48 <tibbs|w> I mean, if something is terribly wrong and people are copying it or arguing that they're right because they're the guidelines....
17:12:52 <geppetto> But it might not be what people were actually doing
17:13:16 <geppetto> But if so that's been true for a while now, and won't matter a couple more weeks
17:13:38 <tibbs|w> True.  So no big deal.
17:13:47 <tibbs|w> Anyway, sorry for distracting.  593....
17:14:18 <geppetto> I think the main part of 593 is this wanting to go somewhere: https://fedoraproject.org/wiki/PackagingDrafts/Langpack
17:14:31 <geppetto> Note that this would be the first use of "rich deps" in Fedora.
17:15:05 <tibbs|w> Yeah, something got horribly screwed up with overlapping wiki edits.
17:15:32 <tibbs|w> There are a bunch of things I didn't change that ended up being shown under my change.
17:15:37 <tomspur> We requested manual mutexes online, when we had a similar problem in #fedora-python
17:16:08 <tibbs|w> I really should try to sort it out but I just haven't had the time.
17:16:27 <tomspur> I would prefer "%package langpack-bs" over "%package bs"
17:16:47 <tomspur> So it is easier to search for a langpack
17:16:54 <geppetto> tomspur: That's the current draft
17:16:54 <Rathann> hi
17:16:55 <tibbs|w> tomspur: I tried to add that, but I think it got reverted or something.
17:16:57 <paragan> it will be good to identify packages by using langpack
17:16:59 <geppetto> #chair Rathann
17:16:59 <zodbot> Current chairs: Rathann geppetto orionp racor tibbs tomspur
17:17:25 <tibbs|w> Also, the current draft has macroized things.
17:17:29 <tibbs|w> Which I like.
17:17:38 <tomspur> geppetto: in comment #5 the question arises again anyway, so maybe it is worth discussing about this first?
17:17:49 <tibbs|w> I can add that to redhat-rpm-macros once we approve something.
17:18:02 <tibbs|w> Assuming that's necessary.
17:18:07 <geppetto> Well, add me as a +1 on prefering the langpack-bs variant
17:18:22 <tibbs|w> Yes, +1.
17:18:45 <geppetto> Yeh, so at that point it's pretty much decided :)
17:19:05 <racor> I am for langpack-<lang>, too.
17:19:11 * geppetto nods
17:19:20 <tibbs|w> We don't want to do a mass renaming, though.
17:19:32 <geppetto> We don't?
17:19:42 <tibbs|w> Original draft said we didn't.
17:19:43 <geppetto> These should all be subpackages
17:19:49 <paragan> I don't think existing package maintainers will do renaming
17:19:54 <Rathann> +1 to langpack-<lang> as well
17:20:00 <paragan> so we just thought let's apply this for any future langpack pacages
17:20:02 <tibbs|w> They will if they convert their packages to use nice macros.
17:20:08 <jsilhan> geppetto, actually hugspell package is SRMP for each package
17:20:27 <tibbs|w> Yeah, hunspell is split up finely.
17:20:36 <tibbs|w> There are a few other packages which are that way.
17:20:46 <paragan> so does hyphen-* and mythes-*
17:20:48 <geppetto> Fair enough, but I assume that's an outlier and most others are subpackages?
17:21:00 <racor> tibbs|w: How many existing packages are we talking about?
17:22:10 <tibbs|w> Probably close to 100, though I haven't looked.
17:22:39 <paragan> yes they are many
17:23:04 <tibbs|w> dnf list aspell-\* autocorr-\* drascula-\* hunspell-\* hyphen-\*|
17:23:16 <tibbs|w> Says 274, but includes some other stuff (-devel).
17:24:01 <tibbs|w> Sorry, that also has "installed" stuff in  it as well.  Still, it's over 200.
17:24:06 <paragan> we have these packages marked langpacks in comps -> https://fedoraproject.org/wiki/I18N/Langpacks/BasePackages
17:24:20 <racor> tibbs: My rough estimate is ~50 ;)
17:24:54 <jsilhan> (that's why naming "langpack-<lang>" would be better for langpack management)
17:25:13 <tibbs|w> dnf list aspell-\* autocorr-\* drascula-\* hunspell-\* hyphen-\*|grep fedora|grep -Ev 'devel|music'|
17:25:26 <tibbs|w> 219 packages.  All of those really are langpacks.
17:25:49 <tibbs|w> Anyway, we know there are many.
17:26:30 <tibbs|w> I might suggest augmenting the macro or adding a new one which automatically does provides/obsoletes.
17:26:51 <tibbs|w> I could work on that as I'm sadly rather familiar with doing RPM macro magic these days.
17:27:04 <geppetto> ok
17:27:05 <tibbs|w> But that's not essential for a first pass anyway.
17:27:16 <geppetto> yeh
17:27:24 <tibbs|w> So: we all like foo-langpack-XX.
17:27:58 <tibbs|w> And a renaming would be up to the packagers, unless we break with our usual policy of not doing retroactive changes.
17:28:03 * jflory7 is here idling -- would like to mention something at your open floor when the time comes
17:28:14 <racor> tibbs|w: You counted binary package - The number of src.rpm is much smaller.
17:28:22 <tibbs|w> That's true.
17:28:38 <racor> In the order of 10-20, I'd guess ;)
17:28:38 <tibbs|w> It's still a whole lot of obsoletes/provides someone has to add.
17:28:46 <tibbs|w> Yes, not many srpms.
17:28:57 <tibbs|w> But a whole lot of packages which have to change names.
17:29:09 <paragan> racor, all hunspell-\*, hyphe-\*, mythes-\* , aspell-\* are individual src packages
17:29:28 <paragan> they are not coming from single srpm
17:29:29 <tibbs|w> Anyway, besides those two things I listed, what is up for discussion?
17:29:55 <racor> paragan: OK, I wasn't aware about this.
17:30:10 <paragan> we need Supplements tag as recommendation in langpacks packages as per given in https://fedoraproject.org/wiki/PackagingDrafts/Langpack
17:30:21 <tibbs|w> paragan: Did you get your changes from comment 5 back into the draft?
17:30:39 <paragan> yes see above link
17:30:54 <tibbs|w> Just making sure that's in the current draft.
17:30:55 <paragan> langpack-%{1}
17:31:02 <tibbs|w> I think supplements is fine.
17:31:04 <paragan> okay
17:31:14 <tibbs|w> You also had comments about the definition of "langpacks".
17:31:30 <racor> paragan: Well, that's a aspect, I am not comfortable with.
17:31:37 <paragan> tibbs|w, yes they are not just language translation packages
17:31:45 <tibbs|w> Well, make sure that's reflected in the draft.
17:32:00 <tibbs|w> I just want to make sure we're looking at the actual draft we're supposed to consider.
17:32:20 <paragan> tibbs|w, yes included as "The idea behind "langpacks" is to separate translations or language specific content into subpackages"
17:32:20 <tibbs|w> racor: I think the choice is between Supplements: in one package and Suggests: in another.
17:32:35 <tibbs|w> Where does your discomfort lie?
17:32:54 <paragan> no need of Suggests: tag
17:33:26 <tibbs|w> paragan: Well, you have to have either a forward weak dependency, or a reverse weak dependency.
17:33:34 <paragan> right
17:33:43 <tibbs|w> Hence, Supplements: in the langpack packages, or Suggests: in the "main" package.
17:34:13 <tibbs|w> I'm trying to figure out if racor preferred one over the other, or if he had a problem with weak deps in general, or something else.
17:35:32 <jsilhan> tibbs, the forward and reverse deps are not the same. The reverse one perfectly suits there
17:36:00 <jsilhan> the weak deps of already installed packages are not considered
17:36:03 <tibbs|w> I agree, but racor said he was uncomfortable and I'm trying to understand why.
17:36:24 <tibbs|w> I mean, we have to use some kind of weak dep here.
17:36:51 <racor> A simultaneous renamer (Obsoletes/Provides/...) with using a little test feature in the light of dnf's immaturity doesn't seem wise, IMO.
17:37:23 <paragan> racor, we are not asking to rename existing packages
17:37:52 <paragan> its just recommendation for any future coming langpack packages in Fedora.
17:37:59 <racor> tibbs|w: Let me put it verbosely: renamers alone are error-prone, ... and my trust in dnf is NULL
17:38:01 <tibbs|w> As I said earlier, FPC doesn't do that in general.
17:38:10 <tibbs|w> I don't know how many more times I can say that.
17:38:37 <tibbs|w> But someone may want to rename; it's up to them.
17:38:59 <paragan> right if existing package owners like to rename, they can
17:38:59 <tibbs|w> I only suggested we consider that case.  And someone has to test dnf at some point anyway.
17:39:00 <jsilhan> we request only to add suggests tag into existing langpacks
17:39:09 <jsilhan> * supplements
17:39:31 <tibbs|w> Otherwise any bugs just won't be found.  You can't just say "we'll never touch these features because they might be broken".
17:39:48 <racor> tibbs|w: I consider the renamer (*langpack*) to be progrees.
17:39:53 <geppetto> Names of weak deps. still confusing all the devs. N years later :(
17:40:36 <geppetto> But, yeh, while the rich deps. usage is new … and supplements, I guess … we can't easily do it in the packages without that.
17:41:07 <Rathann> considering the number of langpacks for a particular package can increase or (sometimes) decrease over time, revdeps are much better here
17:42:15 <paragan> Rathann, I am not familiar with revdeps approach but isn't that mean we need to Suggests: every langpacks in base pack, that will look horrible.
17:42:26 <geppetto> Kind of … I'm not sure how DNF works, but I'm guessing it's the same as yum and only considers the weak deps. on installs.
17:42:35 <tibbs|w> Supplements: is a reverse dependency.
17:42:39 <tibbs|w> I think we're all aggreeing.
17:42:46 <jsilhan> geppetto, yes
17:42:47 <geppetto> So adding langpacks as rev weak deps. isn't that ideal, from that POV
17:43:06 <tibbs|w> But, yes, we really need to know how dnf handles these things.
17:43:09 <geppetto> But I'm still fine with it.
17:43:34 <tibbs|w> If only so that we don't stick something out there, then find out it doesn't work after people start using it.
17:43:35 <Rathann> paragan: why horrible?
17:43:42 <tibbs|w> Is any of this implemented in any package.
17:43:58 <Rathann> the alternative is to have Suggests for every langpack in each package that has langpacks
17:44:03 <tibbs|w> Rathann: I think using forward dependencies for this means that the main package will have a huuge list of Suggests:.
17:44:07 <paragan> I thought we need to write Suggests: langpacks-as or langpacks-bs or langpacks-ja or .......
17:44:19 <paragan> in the same line
17:44:27 <geppetto> paragan: ?
17:44:30 <Rathann> exactly, it's better to have Supplements IMHO
17:44:33 <jsilhan> tibbs|w, it's in copr repo, you can try it for test set of packages
17:44:43 <jsilhan> tibbs|w, https://bugzilla.redhat.com/show_bug.cgi?id=1114422#c22
17:44:43 <paragan> right Supplements is better
17:44:49 <tibbs|w> But how does DNF handle it?  And renaming?
17:45:02 <Rathann> at least it's one per langpack and you don't have to modify the main package if there's a new langpack
17:45:15 <tibbs|w> Someone who's working on this could hopefully try out a few cases and paste the results so we can see.
17:45:34 <geppetto> Rathann: paragan: If you do normal weak deps. instead of reverse … it's the same thing just change s/supplements/suggests/ and put it in the main package.
17:45:38 <paragan> I have tested the Supplements: and its working fine
17:45:47 <tibbs|w> I think that's all anyone is asking for here.  Does dnf handle the issue as we'd expect?  If so, the discomfort should mostly go away.
17:46:22 <geppetto> Rathann: Advantage is you get newer langpacks as they are released … disadvantage is that the main package needs to know about all langpacks
17:47:03 <geppetto> Also, a minor thing is that deps. on main package will now be "big".
17:47:04 <paragan> I like Supplements: usage, its easier
17:47:31 <tibbs|w> paragan: I agree.
17:48:07 <tibbs|w> I suggest linking to some specs in the ticket.  Maybe some specs that show how it would look either way if people want to see that.
17:48:28 <geppetto> No need to do the later, unless anybody wants to argue hard for non-reverse deps.
17:48:41 <tibbs|w> And then doing some dnf tests, including how you would do a rename, and showing how all of that works.
17:49:04 <tibbs|w> This would be the first major use of weak reverse deps that I know of, so it would be good to be extra-certain here.
17:49:08 <paragan> sorry but I don't understand we are not asking to rename the packages
17:49:39 <tibbs|w> Because someone might want to do so, even if it's not mandatory.
17:49:48 <tibbs|w> And if it doesn't work then we need to tell them not to do so.
17:49:49 <racor> tibbs|w: In a real world project, I now would insist on a public demo to stress-test supplements/suggests.
17:50:05 <tibbs|w> racor: Which is kind of exactly what I'm asking for.
17:50:12 <geppetto> We already have suggests usage
17:50:19 <tibbs|w> And, also, there is the copr which is about as public of a demo as you can get.
17:50:30 <tibbs|w> geppetto: Yeah, I just don't think it's as widespread as this might be.
17:50:42 <geppetto> The main things here are supplements, if that's different … and "rich deps." which is the boolean operations.
17:51:18 <tibbs|w> I'm just suggesting that someone show us how it works in practise, with dnf output, so that we can see without everyone on the committee having to spin up a test vm.
17:51:20 <geppetto> jsilhan: Do you have any external test repos. for DNF you can post for either of those?
17:51:26 <racor> tibbs|w: OK, so we are heading the same road.
17:51:50 <tibbs|w> I don't think that's too much to ask, and someone familiar should be able to do all of this in half an hour.
17:51:57 <jsilhan> geppetto, dnf copr enable jsilhan/langpacks
17:52:04 <jsilhan> see https://bugzilla.redhat.com/show_bug.cgi?id=1114422#c22
17:52:09 <jsilhan> how packages are called
17:52:26 <geppetto> #info Test repo. is available at: copr enable jsilhan/langpacks
17:53:21 <jsilhan> the only difference is that the main langpacks are called "tl-langpacks-cz" instead of "langpacks-cz"
17:53:35 <jsilhan> I didn;t want to introduced any conflict
17:53:59 <tibbs|w> If you named them as they would be named, we could see how updates work.
17:55:58 <racor> jsilhan: Would it be possible for you to set up a page with detailed instructions on how to use this?
17:56:23 <jsilhan> racor, ok, no problem
17:56:52 <jsilhan> how to use demo or how to use rich deps?
17:57:06 <jsilhan> the demo is actually described in that bug comment step by step
17:57:18 <racor> jsilhan: background: I've never used copr before ;)
17:57:26 <jsilhan> dnf copr enable jsilhan/langpacks
17:57:29 <jsilhan> that's it
17:57:47 <jsilhan> then you have that repo file locally
17:58:07 <jsilhan> and "dnf install tl-langpacks-cz" will install package from that repo
17:59:25 <jsilhan> if you follow exact sequence of commands from: https://bugzilla.redhat.com/show_bug.cgi?id=1114422#c22 you will see how it works
17:59:35 <jsilhan> then you can play around by yourself
17:59:50 <jsilhan> installing / removing another config langpacks
18:01:50 <jsilhan> should I explain steps better? or is everybody playing with it?
18:02:36 <Rathann> the example looks good, though I wish it were placed somewhere more accessible than a bugzilla comment
18:03:28 <geppetto> jsilhan: I assume the last part only works with cleanup_leaf=true?
18:03:28 <jsilhan> Rathann, would you like it to be included in the draft?
18:03:50 <jsilhan> geppetto, yes
18:03:51 <geppetto> No, although maybe post it in the ticket?
18:04:02 <jsilhan> it's referenced
18:04:05 <jsilhan> there
18:04:12 <geppetto> It should be in the notes from this meeting anyway, so it'll be a bit more findable after today
18:04:17 * geppetto nods
18:06:00 <geppetto> Ok, is there anymore that people want to say about this feature at this meeting?
18:06:49 <tibbs|w> I'm pretty much +1 at this point, provided that it actually works properly with dnf, which I'm sure jsilhan will ensure anyway.
18:07:04 <geppetto> Sure, me too
18:07:06 <tibbs|w> But I would like to address the rename issue one way or the other.
18:07:16 <tibbs|w> We aren't asking anyone to rename.
18:07:28 <tibbs|w> But at some point we'll need to accommodate them if they decide they want to do so.
18:07:31 <geppetto> But I think a few others want to play with it more before voting, or am I wrong?
18:07:40 <tibbs|w> Or tell them they shouldn't rename at this point.
18:08:01 <geppetto> I don't think we should do that, if anything we should encourage them.
18:08:18 <tibbs|w> And then we should tell them how to do so.
18:08:23 * geppetto nods
18:08:29 <tibbs|w> And make sure that actually works.
18:08:36 <tibbs|w> It doesn't have to be in a first pass, though.
18:08:48 <tomspur> +1
18:08:53 <tibbs|w> Which is why I'm fine with this as is (though I need to go through it all again).
18:09:01 <geppetto> racor: Rathann: orionp: mbooth: You want to vote?
18:09:30 <racor> No, don't want to vote.
18:09:41 <orionp> I'm afraid i haven't been able to follow
18:09:43 <tibbs|w> I'l toss in a +1 just for the record.
18:09:54 <geppetto> +1 ftr
18:09:59 <tibbs|w> But we won't get to +5 anyway.
18:10:13 <geppetto> But, yeh, loooks like it'll be another week or 2 for people to look at it and vote then.
18:10:25 <tibbs|w> This is a really good example of why folks submitting things should give as much information as possible up front.
18:11:26 <paragan> Can people vote in the ticket and see if this can get approved by next week if there is no meeting next week?
18:11:29 <geppetto> #info Use RPM tags for langpacks (+1:3, 0:0, -1:0) … more time needed for people to see what happens/why/etc.
18:11:52 <paragan> okay
18:12:05 <Rathann> hm dnf remove langpacks-foo will only remove all langpacks for language foo if the option clean_requirements_on_remove is set to 1 (which is the default)
18:12:11 <Rathann> but if it's 0, then it won't
18:12:17 <geppetto> Rathann: Yeh
18:12:31 <Rathann> I have it set to 0 because dnf was overzealous in removing stuff IMHO
18:12:43 <tibbs|w> Me, too.
18:12:48 <geppetto> Yeh, me too
18:12:49 <racor> For what's it worth: dnf install tl-libreoffice-core-langpacks-es doesn't install tl-libreoffice-core
18:12:59 <geppetto> In theory a requires on the base package could be added in each langpack
18:13:11 <tibbs|w> So, perhaps some hard deps are needed somewhere.
18:13:14 <tibbs|w> Ninja'd.
18:13:17 <geppetto> And that would also "fix" racro's observation.
18:13:22 <jsilhan> yeah there shoudl be require
18:13:31 <paragan> racor, that is not it supposed to work
18:13:59 <Rathann> ok, I'm +1 as well at this point
18:14:08 <geppetto> #undo
18:14:26 <geppetto> jsilhan: You want to update the draft to add the requires?
18:14:38 <racor> paragan: but that's what needs to be exercised: The new rpm tags interaction with traditional requires/provides.
18:14:39 <geppetto> We are at +4 now … if anybody else wants to vote today
18:14:54 <Rathann> the expectation is, I guess, that who unsets that option will know what they're doing and deal with the fallout
18:15:08 <jsilhan> geppetto, in the draft it's actually there
18:15:09 <tibbs|w> paragan: Might be worth knowing why it's not supposed to work.
18:15:19 <racor> jsilhan's copr apparently doesn't test this.
18:15:59 <tibbs|w> Yeah, in the draft, the base package is hard-required.
18:16:03 <paragan> racor, those are just test packages and I think there is missing Requires: on tl-libreoffice-core
18:16:05 <geppetto> jsilhan: Ahh, you're right the example does have it … maybe add something with the supplements text at the top?
18:16:08 <tibbs|w> It's down in the macro definition.
18:16:18 <jsilhan> I've made that CORP minimal example and forgot about requires
18:16:25 <Rathann> yup the example does have it
18:16:25 * geppetto nods
18:16:43 <tibbs|w> Are you all referring to the macro in the draft as the example?
18:16:47 <Rathann> I mean Requires: on the main package it supplements
18:16:47 <geppetto> yeh
18:16:48 <Rathann> yes
18:17:05 <tibbs|w> Personally I would like for the definition of macro to _not_ be in the draft
18:17:20 <tibbs|w> We don't want people pasting it in.  It should be defined for them.
18:17:39 <tibbs|w> In redhat-rpm-config if there's no other base dependency these packages will have.
18:17:57 <Rathann> tibbs|w: but the macro in example isn't generic, is it?
18:18:10 <jsilhan> tibbs|w, as I wrote. I have not found any general macro. Some packages may add additional tags or files
18:18:22 <Rathann> I mean the file path may vary from package to package
18:18:33 <tibbs|w> True, but you could pass some of that.
18:18:41 <jsilhan> langpacks are not just for translations
18:18:48 <tibbs|w> I would have to see how it works out with a couple of packages.  Someone's done glibc.
18:19:05 <jsilhan> and so it can carry any file within any directory with any extension
18:19:12 <tibbs|w> It's just that if you want people to do this, that much macroization is beyond a lot of people.
18:19:17 <Rathann> I agree it would be nice to have a generic macro in rpm-config
18:20:04 <tibbs|w> Anyway, I can look at something.  How many sets of converted packages are available?
18:20:16 <paragan> for translations langpacks we can think of rpm macro but not for all kind of langpack packages
18:20:17 <tibbs|w> Also, it doesn't have to be a macro which covers _everything_.
18:20:27 <geppetto> #info Use RPM tags for langpacks (+1:4, 0:0, -1:0) … more time needed for people to see what happens/why/etc.
18:20:51 <geppetto> Only one more vote needed in the ticket, if people want to vote there before the next meeting.
18:21:27 <geppetto> Anyway … unless anyone really wants to talk about this more I'm going to move to 594
18:21:29 <paragan> When this will get approved I need to ask base package maintainers to add Supplements: tag in their langpack subpackages
18:22:11 <paragan> geppetto, tibbs Rathann tomspur racor thanks all for this discussion
18:22:21 <geppetto> no problem
18:22:29 <geppetto> #topic #594 uid/gid allocation requested for openshift origin
18:22:33 <geppetto> .fpc 594
18:22:34 <zodbot> geppetto: #594 (uid/gid allocation requested for openshift origin) – fpc - https://fedorahosted.org/fpc/ticket/594
18:23:02 <tibbs|w> BTW, a billion lines ago someone had something to ask in open floor.
18:23:12 <geppetto> If they did I missed it :-o
18:23:35 <tibbs|w> Just making sure that person is still around, since it's been a while.
18:23:48 * tomspur needs to go in 7 mins
18:23:52 * geppetto nods
18:24:01 <geppetto> I think this ticket is fairly simple
18:24:20 <geppetto> It's somewhat annoying that everything cloud related wants it's own uid/gid
18:24:30 <geppetto> But it is what it is.
18:25:03 <tibbs|w> I'm starting to wonder why we're caring so much.
18:25:17 <tibbs|w> If we fill up the range, well, it won't be our problem.
18:25:17 * geppetto shrugs … I'm +1
18:25:54 <tibbs|w> BTW, why do they ask for a soft allocated UID?
18:26:01 <tomspur> Hmm...
18:26:08 <tibbs|w> Have I forgotten the definitions?
18:26:32 <tibbs|w> soft allocated is just what everyone does when they add a user.  Doesn't need to go through us.
18:26:51 <geppetto> tibbs: I thought soft allocated meant they got to pick a number too
18:27:01 <geppetto> So package install ordering wouldn't affect the uid they got
18:27:06 <tibbs|w> Then what would hard allocated mean?
18:27:34 <tibbs|w> What else can be harder than fixing the name to the particular number?
18:27:36 <geppetto> If I remember correctly hard == always allocated on an install, soft == only allocated on a machine if package is installed.
18:28:27 <geppetto> So, in theory, it's possible to install "foo" and it request number XYZ but get XXX instead.
18:28:56 <geppetto> Whereas root or news or whatever is just assumed to be there, and there is no failure case.
18:29:00 <tibbs|w> OK, thy mean "soft static".
18:29:06 <tibbs|w> https://fedoraproject.org/wiki/Packaging:UsersAndGroups
18:29:25 <geppetto> Ok :)
18:29:52 <tibbs|w> So, assuming they really mean "soft static" then +1.
18:29:54 <geppetto> I'm still +1 with correct wording ;)
18:30:12 <geppetto> tomspur: orionp: Rathann: racor: mbooth: vote?
18:30:25 <tibbs|w> We should add a section on "Hard allocation" to that, for the case where the UID/GID does in %setup.
18:30:26 <Rathann> one moment
18:30:55 <tibbs|w> The ticket isn't the clearest thing.  By "volume" I'm assuming they mean some remote shared filesystem.
18:31:20 <geppetto> Yeh, I assume so.
18:31:24 <tibbs|w> I wish I had time to revisit my "fixing UIDs at kickstart" thing.
18:31:30 <tibbs|w> It's completely possible now.
18:31:31 <Rathann> +1
18:31:59 <geppetto> Ok, that's +3, anymore?
18:32:03 <tomspur> +1
18:32:06 <orionp> +1
18:32:18 <geppetto> And +5 …
18:32:26 <tibbs|w> BTW, it was jflory7 who wanted to ask something.
18:32:31 <geppetto> racor: mbooth: Want to vote for the record?
18:32:36 <jflory7> tibbs|w: Yes :)
18:32:36 <tibbs|w> Since I think we're getting close.
18:32:39 * tomspur needs to go. Sorry
18:33:00 <tomspur> Then see you in 2-3 weeks or so...?
18:33:11 <jflory7> tibbs|w: Okay to go ahead?
18:33:11 <geppetto> #action uid/gid allocation requested for openshift origin (+1:5, 0:0, -1:0)
18:33:18 <geppetto> #topic Open Floor
18:33:24 <geppetto> jflory7: Ok, go ahead now
18:34:22 <jflory7> The Fedora CommOps team is helping sub-projects and groups in Fedora put together a 2015 "Year in Review" article sharing a few highlights of the past year and few goals for 2016. With an important group like the FPC, it would be awesome to have a Year in Review shared on the Community Blog. :)
18:34:24 <jflory7> https://communityblog.fedoraproject.org/share-your-year-in-review-with-fedora/
18:34:37 <jflory7> That's the original announcement with a few templates to make the writing part easy.
18:34:49 <jflory7> Here's a few other live examples that have already been published: https://communityblog.fedoraproject.org/tag/year-in-review-2015/
18:35:36 <geppetto> And that can go out at any time?
18:36:04 <tibbs|w> Heh, our "year in review" seems to have ended up with a big backlog of writeups and announcements for me.
18:36:12 <tibbs|w> I should have time today, though.
18:36:18 <tibbs|w> "should".
18:36:36 <geppetto> Nah, I think one point of our year in review would be the huge reduction in the backlog, and new members.
18:36:44 * Rathann still has a truckload of bugs and package reviews to go through
18:36:54 <jflory7> geppetto: Ideally, we'd like to have it before too late in February -- maybe middle of the month at the latest would be a good deadline?
18:37:17 <Rathann> I can help edit/review the article though
18:37:17 <jflory7> We're hoping to write up a bigger "Fedora Year in Review" later with the contributions of others too. :)
18:37:24 <geppetto> jflory7: Right, I meant more was there anything it should wait on … or should someone just do it and publish asap.
18:37:34 <jflory7> geppetto: Oh, no, the sooner the better :)
18:37:51 <tibbs|w> I'm going to come right out and say that I probably won't have time to do this.
18:38:10 <geppetto> Well, that probably leaves me as volunteer then
18:38:20 <tibbs|w> Though the announcements I've sent over the year should be a reasonable base for some of it.
18:38:27 <geppetto> Yeh
18:38:43 <tibbs|w> I think this was the year we got much better organized and cleaned out our ticket backlog, though.  Wasn't it?
18:38:44 <geppetto> Dito. 50ish meeting minutes :)
18:38:52 <tibbs|w> I can't remember that far back.
18:38:57 <geppetto> tibbs: Yeh, see above :)
18:39:00 <tibbs|w> Well not this meeting.
18:39:41 <geppetto> I'll try and have a look at it, the big problem is that I've got tomorrow and then I'm busy until around the 11-12th of Feb.
18:40:14 <tibbs|w> geppetto: Can you throw something on the wiki and then if I have time I can add to it?
18:40:35 <jflory7> geppetto++ That would be awesome if you could get a start on it!
18:40:35 <zodbot> jflory7: Karma for james changed to 1 (for the f23 release cycle):  https://badges.fedoraproject.org/tags/cookie/any
18:41:26 <geppetto> Well … that's me volunteered then ;)
18:42:02 <geppetto> I'm not sure we have any goals for 2016 … but I can probably do an initial draft of top 3 highlights from 2015 and some links.
18:42:21 <jflory7> Also, for the record, feel free to ping me any time to help with this too -- especially once it's finished, so I can help get it scheduled :)
18:42:27 * geppetto nods
18:42:28 <tibbs|w> Not do worse than we did in 2015?
18:42:39 <tibbs|w> Address SCLs again?
18:42:48 <geppetto> hey, we did pretty well in 2015 … that's a pretty high bar.
18:43:00 <geppetto> tibbs: Don't you know the new new thing is containers
18:43:03 <tibbs|w> We strive for excellence.....
18:43:41 <geppetto> 2017 will be the year for containers using remote btrfs snapshots, for sure.
18:43:52 <tibbs|w> I'm hoping the container thing will blow over before I have to get too involved with it.
18:43:53 <geppetto> or 2016, whichever year we are in now :-o
18:44:13 <tibbs|w> You said btrfs.  Should probably change that to 2027.
18:44:55 <geppetto> I was just repeating the amazing systemd devs. who all had a good think and decided on btrfs snapshots
18:45:32 * geppetto should probably close soon, before I throw down any gauntlets.
18:46:06 <geppetto> #action geppetto Try to get a draft for FPC year in review.
18:46:11 <geppetto> Anything else?
18:46:19 <jflory7> That's all I wanted to add :)
18:46:20 <jflory7> Thanks!
18:46:23 <geppetto> no problem
18:47:51 <tibbs|w> I guess I'll wander off.  I still have to sort out the bad interaction between SCLs and epel-rpm-macros-6.
18:47:59 <tibbs|w> Oh, here's a random question:
18:48:08 <tibbs|w> ban useless uses of %defattr?
18:48:16 <geppetto> why?
18:48:28 <tibbs|w> Because useless....
18:48:41 <tibbs|w> It's just cargo culted crap.
18:48:57 <tibbs|w> I'm going to script them out of every Fedora spec after F24 branches anyway.
18:49:42 <tibbs|w> Probably not worth a guideline, but we have all of this junk that persists in packages for no good reason.
18:49:58 <geppetto> I'm not against it if you want to go gungho prooven packager on it.
18:50:11 <tibbs|w> I posted a huge list on devel@ the other day.
18:50:16 <tibbs|w> 4500 packages....
18:50:38 <tibbs|w> We suggest that new packagers look at existing packages, but so many of them are just terrible.
18:50:50 <tibbs|w> Are you up with what's in epel-rpm-macros currently?
18:51:09 * geppetto nods … I bet all mine have unneeded defattrs in them :-o
18:51:17 <geppetto> I'm not
18:51:19 <tibbs|w> Several of mine did too.
18:51:50 <tibbs|w> In EPEL5, I made things so that you no longer need BuildRoot:, %clean, Group: or the buildroot cleaning in %install.
18:52:13 <geppetto> interesting
18:52:17 <tibbs|w> I rebuilt all of epel and have no breakages due to it.
18:52:21 <Rathann> alright, I have to drop off now
18:52:27 <geppetto> cool
18:52:33 <Rathann> thanks guys
18:52:36 <geppetto> Rathann: Ok, thanks for coming
18:52:37 <Rathann> see you later
18:58:14 <tibbs|w> Crap, dgilmore has asked me to bring something up.
18:58:20 <geppetto> oh?
18:58:24 <tibbs|w> I should have filed a ticket.
18:58:47 <geppetto> Might be best way, file ticket and deal with it at a future meeting
18:58:52 <tibbs|w> He wanted FPC's opinion on whether koji should install soft deps when making the buildroot.
18:59:23 <geppetto> I'd guess not
18:59:31 <tibbs|w> My suggestion was no, you should actually specify the dependencies you need rather than relying on how soft deps manage to work.
18:59:46 <tibbs|w> But right now koji does, or doesn't, depending on which arch is building.
18:59:47 <geppetto> Yeh
18:59:52 <geppetto> :-o
19:00:03 <geppetto> That is the least useful way to do it
19:00:46 <tibbs|w> Anyway, I will try to remember to file a ticket.
19:01:13 * geppetto nods … after thinking a couple of minutes I'm pretty firmly in the no weak deps. … just add stuff to build deps.
19:01:27 <geppetto> But a ticket is good, maybe someone can come up with a reason for it.
19:01:35 <geppetto> Ok, really going to close now :)
19:01:41 <geppetto> #endmeeting