17:00:16 <geppetto> #startmeeting fpc
17:00:16 <zodbot> Meeting started Thu Dec  3 17:00:16 2015 UTC.  The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:16 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:00:16 <zodbot> The meeting name has been set to 'fpc'
17:00:16 <geppetto> #meetingname fpc
17:00:16 <geppetto> #topic Roll Call
17:00:16 <zodbot> The meeting name has been set to 'fpc'
17:00:24 <mbooth> Hi
17:00:33 <orionp> hello
17:00:36 <geppetto> #chair mbooth
17:00:36 <zodbot> Current chairs: geppetto mbooth
17:00:38 <geppetto> #chair orionp
17:00:38 <zodbot> Current chairs: geppetto mbooth orionp
17:01:12 <geppetto> limburgher racor Rathann SmootherFr0gZ tibbs|w tomspur: FPC ping
17:01:26 <geppetto> tibbs: You too ;)
17:01:37 <tomspur> pong
17:01:43 <geppetto> #chair tomspur
17:01:43 <zodbot> Current chairs: geppetto mbooth orionp tomspur
17:02:08 <tibbs> Howdy.
17:02:17 <geppetto> #chair tibbs
17:02:17 <zodbot> Current chairs: geppetto mbooth orionp tibbs tomspur
17:03:10 <geppetto> Ok, nobody else appears to be on IRC so I'll just start
17:03:17 <geppetto> #topic Schedule
17:04:30 <geppetto> https://lists.fedoraproject.org/archives/list/packaging%40lists.fedoraproject.org/message/5DDO4MBWMNOW2LX7POBKSGAOVEVYGBMO/
17:04:59 <geppetto> #topic #579     Mention abipkgdiff and abidiff in the Guidelines
17:04:59 <geppetto> .fpc 579
17:04:59 <geppetto> https://fedorahosted.org/fpc/ticket/579
17:05:01 <zodbot> geppetto: #579 (Mention abipkgdiff and abidiff in the Packaging Guidelines instead of abi-compliance-checker) – fpc - https://fedorahosted.org/fpc/ticket/579
17:05:05 <geppetto> Dodji: Hey
17:05:25 <Dodji> geppetto: hey
17:05:35 <geppetto> Anything you want to add?
17:05:50 <Rathann> hi
17:05:51 <geppetto> #chair Rathann
17:05:51 <zodbot> Current chairs: Rathann geppetto mbooth orionp tibbs tomspur
17:05:56 <geppetto> Hey
17:06:23 <Dodji> So, to summarize, I have made the changes that were requested
17:06:57 <Dodji> namely, mention in the packaging guidelines that people need to review the abi changes of their packages (when appropriate)
17:07:06 <Dodji> not being specific about the how
17:07:36 <Dodji> though the link points to a page that says how to do it, both with libabigail tools and abi-compliance-checker
17:08:25 <Dodji> and of course, beefed up the "how" pages in the wiki.  That should hopefully make them more useful than they were before
17:08:53 <Rathann> hm, the two sections in https://fedoraproject.org/wiki/How_to_check_for_ABI_changes_in_a_package are basically pointing to the same links
17:09:46 <geppetto> you think so?
17:09:49 <Dodji> Rathann: not exactly
17:10:00 <Dodji> Rathann: if you look carefully, in the case of libabigail, they are not
17:10:20 <Rathann> ah
17:10:22 <Rathann> right
17:10:27 <Rathann> abidiff and abipkgdiff
17:10:29 <Dodji> Rathann: one link is to compare the abis of two *binaries*, whereas the other is to compare the ABIs of two *packages*
17:10:44 <geppetto> given the programs are in the libabigal package … maybe it's better to use a provide or filename for the install instructions?
17:10:46 <Dodji> for abi-compliance-checker, sadly, there is no support for comparing packages
17:10:46 <Rathann> does it make sense to put that in two separate wiki pages?
17:11:11 <Rathann> personally, I feel there's too much clicking to be done
17:11:32 <Dodji> Rathann: but even for the a-c-c case, I made the effort to provide two separate links
17:11:54 <Dodji> <Rathann> personally, I feel there's too much clicking to be done
17:11:54 <Rathann> I'd put it all in the generic https://fedoraproject.org/wiki/How_to_check_for_ABI_changes_in_a_package and made sure it has a nice table of contents
17:12:09 <Dodji> the wiki page will certainly live on, and made better over time
17:12:32 <Dodji> I don't think that should be blocker to update the packaging guidelines as I am requesting
17:12:46 <Rathann> right, I'm not objecting to the change
17:12:47 <Dodji> or do you think otherwise?
17:12:52 <orionp> hmm, I'm more of a fan of smaller wiki pages
17:13:30 <Dodji> <geppetto> given the programs are in the libabigal package … maybe it's better to use a provide or filename for the install instructions?
17:13:35 <Dodji> geppetto: I am not sure to follow
17:13:53 <Dodji> geppetto: what install instructions?
17:14:52 <geppetto> Dodji: The bit talking about what to pass to yum to install the packages that contain the programs
17:15:08 <geppetto> Dodji: Just that I can see the programs moving out of the library package itself at some point
17:16:31 <geppetto> apart from that I think it's fine … I'm not that bothered if it's 1 or N pages
17:16:58 <geppetto> I think N is probably slightly harder to maintain, but probably slightly easier to use … meh
17:17:07 <Rathann> also, s/yum/dnf/ ;)
17:17:30 <Dodji> geppetto: ahhh, I see
17:18:16 <Dodji> geppetto: so what would be the easier yum invocation that you'd advise people to use then?
17:18:32 <Dodji> geppetto: (I am talking practically, sorry, to update the wiki page)
17:19:24 <geppetto> I guess: yum install /usr/bin/abidiff … assuming you don't have any virtual provides already to go
17:20:06 <racor> test, 1, 2, 3
17:20:13 <geppetto> hey racor
17:20:19 <geppetto> #chair racor
17:20:19 <zodbot> Current chairs: Rathann geppetto mbooth orionp racor tibbs tomspur
17:20:28 <racor> Sorry, folks I am having networking probs
17:20:30 <Dodji> geppetto: okay, it makes sense
17:20:41 <Dodji> geppetto: I don't have any virtual provides for that
17:20:53 * geppetto nods
17:21:09 <Dodji> geppetto: I hope there is no other binary with the same name, haha :-)
17:21:20 <geppetto> That would be difficult :)
17:21:42 <geppetto> Anyway, Dodji has to go soon … does anyone have other suggestions before we vote?
17:23:35 <tomspur> I find "that we shall name" confusing on the first read and might name the "library LIB" or so. Nothing of big importance though
17:23:46 <orionp> I'm +1
17:23:59 <mbooth> Me too +1
17:24:11 <tomspur> +1 either way
17:24:29 <geppetto> Yeh I thought L was a weird example name, but not enough to have to change anything :-o
17:24:41 <tibbs> Seems OK, though really I think the buildsys shoud just be doing this kind of thing.
17:24:46 <geppetto> +1
17:24:58 <Dodji> geppetto, tomspur: oh, I can name it lib, no problem :-)
17:25:05 <Rathann> +1 from me
17:25:09 <Dodji> I don't know myself why I said L ...
17:25:12 <tibbs> +1
17:25:28 <geppetto> FOO or BAR is better than L, and all of those are better than lib IMO
17:25:38 <Dodji> 'kay
17:26:48 <geppetto> racor: Did you want to vote for the record?
17:28:26 <tomspur> I have also networking problems with freenode today. Unsure if racor is still here...
17:28:32 <geppetto> #action Mention abipkgdiff and abidiff in the Guidelines (+1:5, 0:1, -1:0)
17:28:34 * geppetto nods
17:28:36 <Dodji> geppetto, tomspur: I changed the wording (s/L/FOO) and I removed the "that we shall name" part.
17:28:42 <geppetto> Cool
17:28:46 <Dodji> geppetto, tomspur: in https://fedoraproject.org/wiki/How_to_check_for_ABI_changes_in_a_package
17:28:50 <racor> +1
17:28:54 <tomspur> Dodji: thanks :)
17:28:57 <geppetto> #undo
17:28:57 <zodbot> Removing item from minutes: ACTION by geppetto at 17:28:32 : Mention abipkgdiff and abidiff in the Guidelines (+1:5, 0:1, -1:0)
17:29:05 <geppetto> #action Mention abipkgdiff and abidiff in the Guidelines (+1:6, 0:0, -1:0)
17:29:08 <Dodji> np
17:29:16 <geppetto> Ok, done :)
17:29:18 <geppetto> #topic #580     [Clarification] Policy on auto-generated code
17:29:18 <geppetto> .fpc 580
17:29:18 <geppetto> https://fedorahosted.org/fpc/ticket/580
17:29:19 <zodbot> geppetto: #580 ([Clarification] Policy on auto-generated code) – fpc - https://fedorahosted.org/fpc/ticket/580
17:29:23 <Dodji> thank you a lot people
17:29:49 <orionp> thank you
17:30:17 <geppetto> I'm mostly happy with tibbs proposal here
17:30:50 <tibbs> I had almost forgotten I had submitted that.
17:31:47 <tomspur> How strong do you wand to enforce Q3?
17:32:20 <geppetto> Oh, I'm sure we'll have our packaging police swat team at the ready ;)
17:32:28 <tomspur> In javascript land it might be difficult to package the tool, which upstream uses to create an (e.g.) minified version of some javascript source file
17:32:49 <geppetto> Hmmm
17:32:58 <tibbs> It's a really complicated issue.
17:33:11 <tomspur> I mean, will Q3 be a 'must' or 'do if possible' or something like that
17:33:22 <tibbs> However, I think we've dealt with the javascript issue already.
17:33:40 <tomspur> I'm sure there will be new ones ;)
17:33:43 <tibbs> I mean, the minified version isn't "the source".  We don't consider those source code.
17:33:47 <geppetto> As long as the original javascript too, I'm not sure I care ? … but maybe we should, as who is going to make sure that the minified version is the same
17:34:46 <tibbs> The fact that jsmin is non-free has ben an issue for a long time.
17:34:57 <tibbs> But we also had an issue with fonts, where we couldn't generate the source.
17:34:58 <Rathann> IMO we shouldn't ship the minified versions shipped by upstream
17:35:31 <tibbs> To be honest, Q3 might be more of an issue for  FESCo or even the Board (or whatever it's called now).
17:35:33 <Rathann> we should run one of the minifiers available in Fedora on the original JS
17:36:31 <Rathann> with autotools-generated files it's slightly a different issue, because the files are at least human readable and modifiable
17:36:37 <tibbs> Anyone agree with me?  Should I bump Q3 and related issues of what is "source" up to FESCo?
17:36:53 <Rathann> tibbs: +1
17:37:04 <geppetto> I'm not opposed to it … but I'm not sure we need to do so either
17:37:13 <Rathann> maybe fedora-legal as well?
17:37:30 <racor> tibbs: -1. Actually I do not think there is a generally applicable answer
17:37:30 <Rathann> it's partly a technical issue as well
17:37:33 <geppetto> I guess it's probably better
17:37:34 <tibbs> True, there are legal issues involved as well, at least for GPL stuff.
17:37:48 <tibbs> racor: There's not a generally applicable answer for asking fesco to take a look at something?
17:37:55 <tibbs> I'm afraid I don't understand.
17:38:29 <racor> tibbs: no. I mean Q3 is not answerable.
17:38:56 <tibbs> I think it is.  I just don't think it's within our purview to answer it.
17:39:20 <Rathann> I'd put the line at human-readable autogenerated "code"
17:39:39 <Rathann> i.e. if it's not, it MUST be regenerated from source during the build process
17:39:42 <tibbs> We can at least provide guidance; as with the rest of our guidelines, we can't cover all cases.
17:39:46 <Rathann> otherwise SHOULD
17:40:17 <racor> tibbs: AFAICT, the FSF's view on generated code in FSF owned code is "no non-free" generator.
17:40:26 <tomspur> If the file cannot be generated on the fly (due to missing tools), I'd consider it as bundled by upstream and would ask FESCo what to do in that case (Forbid it or allow it with some documentation?)
17:41:09 <tibbs> racor: Yes, the GPL is pretty clear on that case, actually.
17:41:24 <Rathann> tomspur: I don't agree that it's bundled code
17:41:51 <mbooth> It's not bundled code
17:42:16 <Rathann> I mean - the question of whether it's bundled is independent of whether it can be regenerated using free tools or not
17:42:22 <tibbs> Indeed, it's not bundled; it's closer to a prebuilt binary.
17:42:26 <Rathann> yup
17:42:53 <tibbs> I think the important thing is: can we reasonably patch it?
17:42:57 <tomspur> hmm...
17:43:21 <tibbs> You can patch autotools-generated stuff.  You can patch the output of bison or lex (though I don't know why you would).
17:43:32 <tibbs> You can't really patch minimized javascript.
17:43:40 <tibbs> You can't patch java bytecode.
17:43:48 <Rathann> tibbs: which is more or less what I was saying earlier
17:44:06 <Rathann> and it's called "minified", IIRC
17:44:09 <tibbs> I have no idea if you can patch the thing that sgallach was asking about in the ticket; he didn't provide that information.
17:45:14 <geppetto> I'm pretty sure you don't want to patch gdbus-codegen
17:45:21 <geppetto> Well, the output of that
17:46:49 <geppetto> So we all fairly happy to do: 1b. 2. yes. 3. ASK FESCO
17:47:31 <tomspur> +1
17:47:48 <mbooth> That's convenient because that's exactly what I did here: http://pkgs.fedoraproject.org/cgit/eclipse-pdt.git/tree/eclipse-pdt.spec#n16
17:48:02 <mbooth> Re: 1b and 2 yes
17:48:56 <geppetto> :)
17:49:00 <tibbs> mbooth: That's the kind of thing that makes sense.  Though obviously there's a bug there somewhere.
17:49:21 <mbooth> tibbs: The Fedora version is too new, so it's an upstream bug
17:49:45 <mbooth> I should make that clearer in the comments
17:49:58 <orionp> +1 to the propasal
17:50:28 <tibbs> I'll add "cook up a draft for 580" to my list.
17:50:55 <geppetto> Yeh, +1
17:51:11 <tibbs> And then we'll see if I can make something people will accept.
17:51:22 <tibbs> I'll also cook up a FESCo ticket.
17:51:34 <geppetto> #action tibbs Draft based on 1. b; 2. yes; and ask FESCO/legal about 3
17:51:45 * geppetto nods
17:51:56 <geppetto> #topic #581     Remove prohibition of SCL macros from Guidelines
17:51:57 <geppetto> .fpc 581
17:51:57 <geppetto> https://fedorahosted.org/fpc/ticket/581
17:51:58 <zodbot> geppetto: #581 (Remove prohibition of SCL macros from Packaging Guidelines) – fpc - https://fedorahosted.org/fpc/ticket/581
17:52:18 <tibbs> Still -1 on this.
17:52:41 <tibbs> Plus I'm still pissed that the SCL proponents just walked away from the process.
17:52:42 <geppetto> I guess I'm ok with this … we don't ban any other random macros people want to use to help them out personally
17:52:49 <Rathann> +1 for 580 FWIW
17:53:36 <geppetto> And we were only banning these because there was a significant chance a bunch of packages would be SCLized, and we didn't want that contaminating the main packages.
17:53:38 <tibbs> Plus Toahio's comment is 100% on point.
17:53:53 <tibbs> geppetto: Well, that's something of an oversimplification.
17:54:17 <geppetto> On the other hand, I'm not sure how much it really buys the packagers to have SCLized stuff in the main Fedora package
17:54:23 <tibbs> We ended up at "SCLs would be OK as separate packages" and they walked away from the process at that point.
17:54:38 <racor> -1  I consider SCLs to be maldesiged junk and in the ages of git do not see any reason to pollute Fedora with them.
17:54:51 <tibbs> That's actually why I cooked up the whole "exception to review process" things as well, but it didn't matter.
17:55:04 * geppetto nods
17:55:24 <geppetto> tibbs: Yeh, I meant more as the reason why we explicitly banned those macros in main Fedora packages
17:55:33 <tibbs> The fact is that Red Hat just came up with this without community input, then tried to push it through.
17:55:45 <tibbs> It didn't go well, and now we see complaints that it makes their lives more difficult.
17:56:22 <Rathann> -1 from me for the reasons above
17:56:46 <tibbs> Too bad it's all baked into RHEL7 at this point and it can't really evolve into something better.
17:57:10 <tibbs> Also too bad is the fact that nobody could ever provide an answer as to why compat packages don't work.
17:57:53 <geppetto> AIUI, they do … but SCL people think it's much easier to have SCL packages.
17:58:06 <geppetto> "they do" == compat packages do work.
17:58:32 <tibbs> And nobody would ever even bother to answer the question of how it's easier from a general standpoint.
17:58:39 <geppetto> Anyway, with 3 -1s, this isn't going to pass.
17:59:05 <geppetto> #action Remove prohibition of SCL macros (+1:1, 0:0, -1:3)
17:59:29 <tibbs> And my final rant, nobody has actually added anything new to the argument.
17:59:57 <geppetto> #topic #566     RPM file triggers
17:59:57 <geppetto> .fpc 566
17:59:57 <geppetto> https://fedorahosted.org/fpc/ticket/566
18:00:00 <zodbot> geppetto: #566 (RPM file triggers) – fpc - https://fedorahosted.org/fpc/ticket/566
18:00:20 <geppetto> Not much new here, although a bunch of bugs have been found by the systemd guys when trying to change systemd.
18:00:23 <tibbs> Seems there are unpleasant bugs in RPM surrounding some of this.
18:00:51 <tibbs> geppetto: In case it helps you, there is now an autogenerated tarball that just contains all of the package specs.
18:01:01 <geppetto> First bug got fixed fairly quickly, so hopefully everything will get fixed this year
18:01:19 <tibbs> When is F24 branching?
18:01:21 <geppetto> tibbs: Cool, how big is that? Where is it?
18:01:44 <geppetto> https://fedoraproject.org/wiki/Releases/24/Schedule
18:01:45 <tibbs> I have to find it.
18:01:59 <geppetto> says alpha freeze is feb. 2016
18:02:29 <geppetto> branch 2nd Feb.
18:03:05 <geppetto> #topic Open Floor
18:03:10 <geppetto> Anything else?
18:04:53 <tibbs> I'm spent.  Is there anything I've forgotten to write up?  It's been a busy couple of weeks.
18:05:16 <Rathann> nothing new from me, thanks
18:05:27 <Rathann> oh
18:05:35 <tibbs> geppetto: http://pkgs.fedoraproject.org/repo/
18:05:40 <Rathann> the trac shows #540 as waiting to be written up
18:05:44 <tibbs> rpm-specs-latest.tar.xz
18:06:10 <geppetto> 21M ??? really?
18:06:21 <geppetto> That has to be just the specfiles, right?
18:06:25 <tibbs> That's a hell of a lot smaller than the checkout seed.
18:06:27 <Rathann> geppetto: we have around 20k packages
18:06:28 <geppetto> The git-seed will be all of git
18:06:39 <tibbs> The specs tarball is just specs.
18:06:44 * geppetto nods
18:06:49 <tomspur> nice
18:07:07 <geppetto> Rathann: I was shocked it was so small, not that it was so big :)
18:07:50 <tibbs> Could probably try to tar up a shallow clone or something without the .git dirs if people really want to see that.
18:08:12 <geppetto> Not unless anyone wants
18:08:24 <orionp> half of it is texlive... :)
18:08:29 <geppetto> haha
18:08:48 <orionp> actually, probably %changelog
18:08:49 <Rathann> :)
18:08:56 <Rathann> ok, I'm off, thanks guys
18:09:02 <tibbs> Hmm, I did writeup 540.  I must not have moved the ticket status.
18:09:12 <geppetto> Yeh, I'll close in a couple of minutes
18:09:19 <geppetto> Rathann: see ya next week
18:09:28 <tibbs> Or, more likely, I must not have clicked the submit button after typing things in.  Because that's what always happens to me.
18:09:41 <Rathann> bye
18:09:50 <tibbs> https://fedoraproject.org/wiki/Packaging:Guidelines#Application_Specific_Guidelines
18:09:53 <tibbs> Yeah, it's in there.
18:10:01 <geppetto> tibbs: So do those tarballs get generated at 5:45 UTC every day?
18:10:09 <tibbs> geppetto: Pretty much.
18:10:19 * geppetto nods
18:10:36 <tibbs> Though it's probably 5:00 and it just takes that long to generatd the big tarball.
18:10:47 <geppetto> I assume it's somewhere between really hard and impossible to make the git-seed one also a torrent?
18:10:58 * geppetto nods
18:11:16 <tibbs> I would think you'd have to see it directly from pkgs, which means having more stuff running there.
18:11:29 * geppetto nods
18:11:43 <tibbs> Not sure who really uses torrents these days.  For something like that I'd only ever expect to see one seed anyway.
18:12:06 <tibbs> I mean, torrents are nice, but for content that a whole lot of people want.
18:12:08 <geppetto> Yeh, changing daily makes it hard … at least with current torrent infra.
18:12:36 <geppetto> Lots of people still use torrents for 1GB+ files :)
18:13:10 <tibbs> I'm not sure infra has ever been concerned with the bandwidth usage.
18:13:23 <tomspur> How often is the git-seed downloaded each day?
18:13:40 <tibbs> That would be a Q for the infra folks; I don't have access to that info.
18:13:48 <geppetto> Yeh, I was thinking more about people who might want to download it but can't easily do 6GB in a single day
18:14:05 <geppetto> But, meh, life sucks for them
18:14:18 <tibbs> I think at that point they might just check out the repositories they really want.
18:14:29 * geppetto nods, or write a script like I did
18:14:30 <tibbs> Or just download srpms.
18:14:36 * geppetto nods
18:14:46 <tibbs> I mean, it's a seed, so you're not supposed to download it every day.
18:15:03 <geppetto> Yeh, anyone doing that is crazy :)
18:15:21 <nirik> in the last month it's been downloaded 5 times
18:15:35 <tibbs> Which seems about right.
18:15:39 <geppetto> Were they all tibbs ?:)
18:15:59 <tibbs> Actually I don't think I've ever downloaded the seed.
18:16:09 <geppetto> Fair enough
18:16:38 <geppetto> Anyway … anyone have anything or I'll close in a couple of minutes
18:16:53 <tibbs> Nothing from me.
18:17:56 <tomspur> Thanks, see you next week.
18:18:16 <tibbs> tomspur: At some point we need to converse more about the python macros thing.
18:18:30 <tibbs> I have all of that swapped out completely and need to get back into it and try to finish it up.
18:19:00 * tomspur nods
18:19:10 <geppetto> Do it at the meeting in 3 weeks time ;)
18:19:44 <tibbs> Yeah, I guess we won't be meeting on the 24th.
18:19:52 <geppetto> Probably not :)
18:20:04 <geppetto> That would be a special kind of present :)
18:20:08 <tomspur> Same for 31th ;)
18:20:38 <tibbs> Just realized I haven't eaten in 43 hours.  Oops.  I almost forgot to come home from the office last night.
18:21:01 <geppetto> You should go do that
18:21:18 <geppetto> #endmeeting