16:00:43 <geppetto> #startmeeting fpc
16:00:43 <zodbot> Meeting started Thu Nov  1 16:00:43 2018 UTC.
16:00:43 <zodbot> This meeting is logged and archived in a public location.
16:00:43 <zodbot> The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:43 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:00:43 <zodbot> The meeting name has been set to 'fpc'
16:00:43 <geppetto> #meetingname fpc
16:00:43 <zodbot> The meeting name has been set to 'fpc'
16:00:43 <geppetto> #topic Roll Call
16:00:45 <mhroncok> hi
16:00:47 * limburgher here
16:00:51 <tibbs> Howdy.
16:00:52 <geppetto> #chair mhroncok
16:00:52 <zodbot> Current chairs: geppetto mhroncok
16:00:54 <geppetto> #chair limburgher
16:00:54 <zodbot> Current chairs: geppetto limburgher mhroncok
16:00:56 <geppetto> #chair tibbs
16:00:56 <zodbot> Current chairs: geppetto limburgher mhroncok tibbs
16:01:09 <geppetto> #chair decathorpe
16:01:09 <zodbot> Current chairs: decathorpe geppetto limburgher mhroncok tibbs
16:01:14 <decathorpe> hey guys
16:01:24 <geppetto> redi said he might be 5 minutes late
16:06:50 <geppetto> #topic Schedule
16:07:11 <geppetto> https://lists.fedoraproject.org/archives/list/packaging@lists.fedoraproject.org/message/ENFQ3JLQ7NDVWQ4QQPJPYCYF7DJINRQB/
16:07:57 <geppetto> So … no change again.
16:08:13 <tibbs> I guess I need to move the "tilde in version" ticket back to meeting.
16:08:36 <tibbs> There are three exception requests related to tildes.
16:08:37 <geppetto> Ahh, maybe … do we have anything to discuss on that?
16:08:48 <geppetto> Which one is it?
16:08:48 <tibbs> Well so I've typed in several walls of text.
16:09:39 <geppetto> #topic  #398 Tilde in version
16:09:43 <geppetto> .fpc 389
16:09:45 <zodbot> geppetto: Issue #389: bundled bootstrap binary exception for sbt - packaging-committee - Pagure.io - https://pagure.io/packaging-committee/issue/389
16:09:47 <tibbs> Starting down at https://pagure.io/packaging-committee/issue/398#comment-537897
16:10:27 <tibbs> The end result of that seems to be:
16:11:06 <tibbs> 1) My original proposal isn't workable in probably the most common "weird" versioning sequence.
16:11:22 <tibbs> 2) People want "tilde" but they don't agree as to how they would use it.
16:11:39 <tibbs> To me, #2 validates the need for real guidelines.
16:11:56 <tibbs> And #1 means I need to rework the proposal a bit.
16:13:17 <tibbs> So... it's progress but it looks like more work for me.
16:14:23 <redi> here now - sorry I was late
16:14:26 <geppetto> Well some good and some bad news, I guess
16:14:32 <geppetto> #chair redi
16:14:32 <zodbot> Current chairs: decathorpe geppetto limburgher mhroncok redi tibbs
16:15:18 <tibbs> I will keep working on it but the twin attitudes of "ignore the guidelines when I don't feel like following them" and "keep filing tickets until the committee changes its mind" really aren't helping.
16:15:46 <tibbs> And I still don't know if we would vote to pass anything involving tilde at this point.
16:16:00 <mhroncok> imho not at this point
16:16:17 <tibbs> Well obviously; we have no workable proposal.
16:16:40 <redi> I don't even see how to make one from the various suggestions, because of your #2
16:16:59 <geppetto> If we get to the point you are happy with it, I think I'd probably vote for it … ofc. reading through giant comments atm. :)
16:17:02 <limburgher> The whole thing makes my head spin.
16:17:05 <redi> everybody wants something different, but seems to think allowing ~ somehow will solve their issues
16:17:19 <tibbs> Well the problem is that many of the suggestions don't even take into account the requirements of such a thing.
16:17:30 <tibbs> So I just have to ignore those and press on.
16:17:42 <tibbs> In the end all that's clear is that someone will be dissatisfied regardless.
16:19:16 <tibbs> But it certainly is possible to come up with a workable proposal.
16:19:47 <redi> ok, glad to hear it
16:19:50 <mhroncok> shall we close all the exception seeking tickets then with something like: ...
16:19:54 <tibbs> It's just not possible to do so without making someone unhappy.  Some of the comments essentially say that any guideline will be too complicated for people to follow.
16:20:05 <tibbs> Including the existing guidelines.
16:20:22 <mhroncok> The FPC is wokring on a workable proposal of tilde in guidelines. No exceptions are granted at this point, juts please wait a bit longer.
16:20:26 <tibbs> But offer no real solution to that except for something which doesn't tell you what to do if things go wrong.
16:20:34 <ignatenkobrain> sorry for being late
16:20:37 <ignatenkobrain> .hello2
16:20:38 <zodbot> ignatenkobrain: ignatenkobrain 'Igor Gnatenko' <i.gnatenko.brain@gmail.com>
16:20:57 <geppetto> #chair ignatenkobrain
16:20:57 <zodbot> Current chairs: decathorpe geppetto ignatenkobrain limburgher mhroncok redi tibbs
16:22:38 <geppetto> mhroncok: I'm fine with that, maybe refer to the original ticket if they want to follow along?
16:23:00 <tibbs> I'm thinking that I should start another ticket when I've completed the updated proposal.
16:23:09 <tibbs> The current one is too long and pagure doesn't make it easy to comprehend.
16:23:15 <geppetto> Fair enough … tidies things up, I guess.
16:23:29 <geppetto> I'm sure someone will complain about that :)
16:24:02 <tibbs> I will certainly reference the new ticket in the old ticket so that nobody accuses us of trying to do something in secret in an open ticket tracker.
16:24:32 <tibbs> I also did a little work on some R packaging stuff which I will need to present at some point.
16:24:42 <mhroncok> a new ticket with a new proposal that obsoletes the odl ticket. that's juts fine and if they complain, wel..
16:24:55 <tibbs> That R stuff relates to 823.
16:28:15 <tibbs> Nothing really to add on that, though if anyone is following the discussion in devel, I'd appreciate input on the acceptability of the "hide everything behind a macro" case.
16:28:32 <geppetto> .fpc 823
16:28:34 <zodbot> geppetto: Issue #823: Proposal: Use immutable CRAN URLs - packaging-committee - Pagure.io - https://pagure.io/packaging-committee/issue/823
16:29:11 <geppetto> tibbs: I thought we were generally in favour of that?
16:29:21 <tibbs> See https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/message/DXFG3IBOPEKKJSHNR3BGIU7DNQRURN7L/ for what I'm talking about.
16:29:26 <geppetto> tibbs: Is there something special that's different in this case?
16:29:38 <tibbs> geppetto: Well, we've sort of gone back and forth on how much is OK to hide.
16:30:34 <tibbs> Basically, that package has no %prep, %build, %install or %check and the %files list doesn't list anything explicitly.
16:30:46 <tibbs> Just %_r_simple_archful_package and the rest is automatic.
16:31:35 <geppetto> I'm not sure hiding %files is a good idea
16:31:41 <nim> tibbs, in my experience you're better of with one macro per rpm section
16:31:46 <geppetto> apart from that, it seems fine to me
16:31:51 <tibbs> nim: You can do that as well.
16:31:51 <nim> tibbs, that's easier to evolve and maintain
16:32:17 <nim> tibbs, but that depends if the thing you package is composable with something else in a single spec or not
16:32:30 <tibbs> But there is no technical need or reason to list everything out since for R packages the majority of them have exactly the sections.
16:32:59 <mhroncok> tibbs: it lets you stuck custom commands in between
16:32:59 <tibbs> And if you don't, less magical macros are also provided.  Further down in the thread is an example of that as well.
16:33:09 <tibbs> mhroncok: Yes, look at both examples.
16:33:22 <tibbs> For most of R packages you don't need to add anything else.
16:33:44 <tibbs> For those that do, you have the option of using %r_prep, %r_install and such just like we do with python.
16:34:21 <tibbs> The file list is still automatically generated, though you can certainly ignore it and list the files manually, or add to it by listing files explicitly.
16:34:32 <tibbs> R packaging is so uniform that it's rare to need to do this, though.
16:34:32 <mhroncok> tibbs: I think that this is easier to digest for a packager that wlaks by such spec
16:34:37 <mhroncok> *walks
16:34:56 <mhroncok> hovever I'm not agains "hide everyhting ebhind one macro"
16:35:07 <tibbs> I think it depends on what the common case.
16:35:09 <tibbs> is.
16:35:22 <mhroncok> it's just that if we start with the second proposal, people get used to it and thay might want more abstraction later
16:35:47 <tibbs> For R (or CRAN, at least), it appears that the common case is extremely common.
16:35:49 <mhroncok> but if we say: here's a macro that does 3 sections, well it might be to hard to fgrasp
16:36:40 <tibbs> Depends on how it's documented.
16:36:45 <mhroncok> that said, I hate %files magic
16:36:53 <tibbs> So you don't use %find_lang?
16:37:07 <mhroncok> I use it, but I hate it anyway :)
16:37:23 <mhroncok> There is always a certain level of magic
16:37:27 <tibbs> All R packages will include exactly the same nine lines in %files.
16:37:45 <mhroncok> I mean even %{python3_sitearch}/%{srcname}/ is magical in a way
16:37:52 <tibbs> Some may include extras.
16:38:26 <tibbs> Anyway, this was an experiment to see how far you could go.
16:39:01 <tibbs> It butts up against the vagaries of the RPM specfile parser at this point but doesn't do anything "crazy".
16:39:26 <mhroncok> I feel that: %{_libdir}/R/library/%{packname}/ is far more readable if that is the way to inslude the DESCRIPTION, INDEX, etc files
16:39:41 <tibbs> That doesn't work.
16:39:55 <tibbs> Misses %doc and %license bits for things that live under there.
16:40:17 <mhroncok> aha!
16:41:15 <mhroncok> in that case I guess it has a point
16:41:20 <tibbs> I just think we've not really done as much as we could if we have hundreds of completely identical packages.
16:41:28 <geppetto> #topic Open Floor
16:41:51 <tibbs> Let the easy packaging be as easy as possible and leave the complexity for the things that need it.
16:42:10 <geppetto> tibbs: yeh, I'm maybe fine with an extra line for %files … but if it really is so common then meh, can just hide that too I guess.
16:42:12 <tibbs> The problem is that you have to document it properly, so that if someone uses the least complicated method and it fails, they know why.
16:42:23 * geppetto nods
16:42:31 <tibbs> Well right now I'm not proposing to actually remove the %files line.
16:42:47 <tibbs> I experimented with it but it was just a bit too magical.
16:43:17 <geppetto> Ahh, ok, cool.
16:43:22 <geppetto> Anyway … we've kind of been open floor since the start, but is there anything else anyone wants to talk about before we close?
16:43:41 <nim> tibbs, it's not really much more magical than the things I'm working on now
16:43:48 <geppetto> We can probably run over this week as there's no "go/no-go" meeting … but I'd rather not.
16:44:09 <nim> tibbs, I really had the feeling you were closing up on the  finish while I was stuck fixing Go bugs
16:44:47 <nim> just FYI, I'm still working on the forge templates as promised
16:45:18 <nim> I got distracted when I hit the git release part and the v thing github injects
16:45:41 <nim> into trying to convince git upstream once again to make releases first-class git objects
16:46:04 <nim> I hate documenting breakage instead of fixing it
16:46:32 <nim> nim, but I'll post a draft this week probably
16:47:17 <nim> nim, and I'm still hoping https://github.com/rpm-software-management/rpm/issues/104 progresses
16:47:30 <nim> so I can upstreaming the Go macros that use it
16:47:35 <nim> in Fedora
16:48:49 <tibbs> I wouldn't hold my breath on that, but I do hope that something like it gets implemented at some point.
16:49:05 <mhroncok> it will
16:49:15 <mhroncok> but it might take a lot of time
16:50:07 <tibbs> If it existed, the R stuff I'm talking about could certainly use it.
16:50:19 <nim> mhroncok, the alternative it to fix up pm request into a deployable state
16:50:36 <nim> mhroncok, basically just filter properly the commands it passes to dnf
16:50:48 <tibbs> One other interesting case is that R modules can know if they need to be archful or noarch, but of course you have to unpack the source tarball to find that out.
16:52:04 <geppetto> interesting
16:52:14 <geppetto> Anyway … we good?
16:52:17 <tibbs> Yeah.
16:52:22 <geppetto> Ok :)
16:52:33 <geppetto> #endmeeting