path: root/meeting-logs/20140624.txt

[15:04:39] <rich0> roll call :)
[15:04:48] -*- dilfridge here
[15:04:56] -*- WilliamH here
[15:05:00] <rich0> blueness, dberkholz, scarabeus, ulm?
[15:06:14] <blueness> rich0, png
[15:06:21] <blueness> sorry almost forgot!
[15:06:25] <rich0> np :)
[15:06:32] <rich0> 3rd time might be the charm for finishing our term!
[15:06:38] <blueness> yep
[15:06:48] <rich0> Ok, let's get started and hopefully dberkholz, scarabeus, and ulm will catch up.
[15:06:55] <rich0> GLEP 64
[15:06:56] <blueness> yep
[15:06:59] <rich0> Err, GLEP 62.
[15:07:43] <rich0> Thoughs?
[15:07:46] <rich0> Thoughts?
[15:08:00] <rich0> https://wiki.gentoo.org/wiki/GLEP:62
[15:08:00] -*- WilliamH is looking for the glep
[15:08:19] <blueness> reading
[15:08:37] <rich0> mgorny: ping, FYI - your GLEP is under discussion...
[15:09:21] <rich0> I like the fact that it avoids polluting @world 
[15:09:46] <dilfridge> I like the whole idea, I just think the detail description of the algorithm is not really making sense under "Reference implementation"
[15:10:15] <rich0> Well, it obviously isn't an actual implementation - more of a design.
[15:10:28] <dilfridge> yeah, sure,
[15:10:34] <rich0> I'd probably have this section updated once an actual implementation is done.
[15:10:44] <dilfridge> but I think this is more something that should come out of writing the code
[15:10:47] <dilfridge> exactly
[15:10:48] <blueness> dilfridge, yeah i don't see that as bad, its an outline of what IUSE_RUNTIME entails
[15:11:21] <rich0> So, we can approve the GLEP, except that the reference implementation can be updated once actually implemented.  :)
[15:11:24] <WilliamH> I'm sure this was discussed and I missed it, why is this a glep instead of a feature for a new eapi?
[15:11:26] <blueness> rich0, i guess the point is if a package installs a perl script, you and you add IUSE_RUNTIME=dev-lang/perl then perl is NOT pulled in?
[15:11:32] <blueness> err ... not exactly
[15:11:54] <blueness> but some flag which pulls in perl
[15:11:56] <rich0> WilliamH: it wasn't discussed.  Probably was just propsoed this way since GLEPs used to be the way these things were done.
[15:12:21] <dilfridge> ulm had something to say about this
[15:12:57] <dilfridge> as far as I can remember, his argument originally was "it's 100% backward compatible and could even in theory be added independent of eapi"
[15:13:04] <WilliamH> blueness: as an example, logrotate could be a use flag aas long as it were in IUSE_RUNTIME.
[15:13:28] <blueness> this is the point -> enabling or disabling any of the flags must not involve rebuilding the package,
[15:13:32] <dilfridge> exactly
[15:13:48] <blueness> okay i get it now, i'm okay with it
[15:13:56] <rich0> exactly - perl wouldn't be pulled in unless you enabled the perl USE flag.  If you enabled it later then perl would be pulled in, but the package wouldn't rebuild.
[15:14:14] <blueness> right
[15:14:17] <rich0> If you disabled it later then perl isn't pulled in, though it won't go away if something else pulls it in, obviously
[15:14:30] <rich0> Any issues with the actual proposal?
[15:14:54] <blueness> not really, i also don't see the 'reference implementation' as a problem
[15:14:56] <rich0> There is the GLEP vs EAPI6 bit - probably does make more sense as part of the EAPI, though I'm not opposed to having the GLEP as well.
[15:15:30] <dilfridge> the "Reference implementation" section is not a problem, I just would not see it as "close to final", more as a roadmap
[15:15:47] <WilliamH> Yes this is more of a roadmap.
[15:16:31] <dilfridge> one suggestion would be, if an implementation in portage is ready in time, consider it for EAPI=6, otherwise postpone for later (this should not hold up the EAPI)
[15:16:33] <rich0> Do we just want to approve it for EAPI6?
[15:16:53] <blueness> sure if we approve the gleb
[15:16:56] <blueness> glep
[15:17:03] <rich0> dilfridge: makes sense - I'd probably leave that up to the PMS/portage teams or the next council
[15:17:13] <blueness> unless you think it will hold up eapi=6 and we don't want that
[15:17:26] -*- WilliamH isn't in favor of retroactively applying it to older eapis either.
[15:17:42] <rich0> No, I'd just make it work for new EAPIs unless all the PMs chime in that it isn't an issue.
[15:17:53] <rich0> That's the whole point of PMS.
[15:17:59] <dilfridge> not retroactively
[15:18:38] <rich0> Granted, the proposal is basically designed to handle that - older PMs would ignore the IUSE_RUNTIME and just treat it as a use flag.
[15:18:46] <rich0> So, the dependency would work fine, it would just trigger a rebuild when it changes.
[15:18:51] <dilfridge> well, it's not an issue in terms of definition or functionality, just in matters of convenience (if a pm does not respect IUSE_RUNTIME people end up with an insane number of rebuilds)
[15:19:14] <rich0> So, in that sense it could potentially go into EAPI6 even if portage isn't ready.
[15:19:14] <dilfridge> (assuming that IUSE_RUNTIME is widely accepted)
[15:19:40] <dilfridge> hmm
[15:19:44] <rich0> If it gets rid of einfo spam, more power to it!  :)
[15:20:31] <rich0> Ok, do we just want to include it in EAPI6 then?  We can save appropving the GLEP until the rest of EAPI6 is ready to be approved?
[15:20:50] <dilfridge> we can add it to the list of planned EAPI6 features, yes
[15:20:55] <rich0> EAPI6 itself gets its real approval when PMS is submitted to the next council.
[15:21:26] <rich0> I think any of the items we included in EAPI6 are at the discretion of the PMS team if for whatever reason the PMs can't implement all of them.
[15:21:34] <WilliamH> We should probably find out from the portage guys a rough idea of whether it will hold up EAPI 6?
[15:21:38] <rich0> That is why this isn't a final approval.
[15:22:25] <rich0> WilliamH: I don't see this as a hard commitment to anything.  This is just a process to get rid of EAPI6 items that we don't want so that people don't implement them only to have the next council reject them.
[15:23:00] <rich0> Granted, I guess they could still do that.  :)
[15:23:00] <blueness> WilliamH, who are the portage guys these days?
[15:23:00] <blueness> ie who's the lead?
[15:23:09] <dilfridge> blueness: dol-sen
[15:24:03] <rich0> My personal recommendation would be to include, and then let the portage guys reject it later.  Anything not implemented would be dropped from the final EAPI6.
[15:24:10] <blueness> rich0, let's add it, they can always amend later
[15:24:17] <rich0> blueness: ++
[15:24:23] <blueness> <mid-air!>
[15:24:29] <rich0> Really we're just saying htat it is "approvable"
[15:25:11] <dilfridge> sounds good
[15:25:16] -*- dilfridge approves
[15:25:22] <blueness> call the vote!
[15:25:33] <rich0> Ok, so how about "The council accepts IUSE_RUNTIME for implementation in EAPI6 as outlined in GLEP 62.  The actual GLEP will be approved when EAPI6 is approved as part of PMS."
[15:25:37] <rich0> vote!
[15:25:43] -*- rich0 yes
[15:25:47] -*- blueness yes
[15:25:53] -*- dilfridge yes for glep 62 
[15:25:58] -*- WilliamH yes
[15:26:11] <rich0> ok, passes 4-0 with 3 absent.
[15:26:27] <rich0> That brings us to bugs.
[15:26:33] <dilfridge> yawn
[15:27:24] <rich0> dberkholz: write your summaries!
[15:27:54] <rich0> blueness: I think the last one is yours?
[15:28:02] <blueness> which bug?
[15:28:12] <WilliamH> I need to ask a quick question about the rules of the glep though.
[15:28:13] <rich0> bug 477030
[15:28:14] <willikins> rich0: https://bugs.gentoo.org/477030 "Missing summary for 20130611 council meeting"; Doc Other, Project-specific documentation; CONF; ulm:council
[15:28:26] <rich0> WilliamH: sure, we're just about at AOB anyway
[15:28:38] <blueness> rich0, all my summaries are posted afaik
[15:28:40] <blueness> let me look
[15:28:53] <rich0> you're right
[15:29:02] <WilliamH>  I guess my logrotate example is wrong, because of rule 3.
[15:29:09] <rich0> Unless you were responsible before you joined the council
[15:29:17] <rich0> 2013-06-11
[15:29:20] <rich0> It was the last council
[15:29:31] <rich0> I just was looking at your bug comment
[15:29:31] <dilfridge> betelgeuse
[15:29:50] <rich0> Ok, just say it three times and hopefully it will happen.
[15:29:56] <blueness> rich0, oh that was me being a newbie ;)
[15:29:57] <dilfridge> :D
[15:30:00] <WilliamH> I was thinking we could use this to reinstate use flags for things like logrotate, xinetd, etc.
[15:30:06] <dilfridge> brb
[15:30:29] <rich0> WilliamH: what is the problem with that?
[15:30:43] <rich0> Which rule 3?  flags listed in IUSE_RUNTIME must not be referenced in phase functions, DEPEND, LICENSE or SRC_URI,
[15:30:43] <rich0> ?
[15:30:53] <WilliamH> rich0: right.
[15:31:02] <rich0> why would logrotate be part of those phases?
[15:31:09] <rich0> It would be an RDEPEND.
[15:31:21] <WilliamH> if use logrotate; then
[15:31:29] <WilliamH> # install logrotate config files
[15:31:29] <rich0> You install the logrotate script unconditionally, and then pull in logrotate conditionally as an RDEPEND.
[15:31:30] <WilliamH> fi
[15:32:01] <rich0> This would only manage dependencies, not what actually gets installed.
[15:32:05] <rich0> That is the key to IUSE_RUNTIME.
[15:32:11] <WilliamH> rich0: logrotate isn't an rdepend, the package doesn't need it to run.
[15:32:28] <rich0> You don't use IUSE_RUNTIME with things the package NEEDS to run.
[15:32:39] <rich0> It is for things that optionally improve the package.
[15:32:48] <WilliamH> rich0: right, so you can't put logrotate in rdepend.
[15:32:49] <rich0> I need to dig up some examples.
[15:33:05] <rich0> WilliamH: why not?  It is a suggested dependency, basically.
[15:33:13] <floppym> I think it would be a bad idea to add IUSE_RUNTIME="logrotate systemd openrc ..." to every packages.
[15:33:29] <rich0> Myabe logrotate isn't a great example.
[15:33:34] -*- WilliamH agrees with floppym
[15:33:51] <rich0> I forget what the last package I installed was that had about 10 lines of einfo about useful stuff I could optionally install.
[15:34:07] <floppym> net-misc/netctl is a good example.
[15:34:53] <rich0> Good one
[15:35:49] <rich0> systemd has some
[15:35:56] <rich0> sys-apps/systemd-ui: for GTK+ systemadm UI and gnome-ask-password-agent
[15:36:05] <rich0> dracut has a laundry list
[15:36:07] <dilfridge> logrotate is a bad example, since in that case the idea originally was to control installation of a small file
[15:36:27] <dilfridge> (which is not possible if you dont reinstall on switching useflag)
[15:36:32] <rich0> in this case it wouldn't be about controlling the file installation, but about pulling in logrotate itself.
[15:36:36] <dilfridge> yep
[15:36:50] <rich0> dracut is actually a pretty good example
[15:37:14] <rich0> http://pastebin.com/4cfwVqSb
[15:37:27] <WilliamH> In that case, I would not suggest logrotate as a use flag still.
[15:38:10] -*- WilliamH still doesn't like that we force users to use install_mask if they want to get rid of small files like that.
[15:38:42] <rich0> Anything else on this?  I'm not hearing that we want to actually revisit the vote...
[15:38:54] <rich0> I'd definitely like to get to open floor this week!  :)
[15:39:22] <blueness> i'm good, this was more of a clarification for me
[15:39:42] <rich0> Sure, and the bonus of this is that it gives everybody more stuff to fight over in six months!
[15:39:42] <WilliamH> heh we can move on.
[15:39:57] <rich0> Thus ensuring we or our replacements still have a job to do...
[15:40:05] <rich0> Ok, open floor then.
[15:40:09] <rich0> And any other business.
[15:41:01] -*- rich0 enjoys the crickets...
[15:41:19] <dilfridge> let me just state that I think this was a productive year :)
[15:41:36] <blueness> i think so too
[15:41:44] <blueness> now i have to sit down and write my glep
[15:41:48] -*- WilliamH agrees
[15:41:54] <rich0> Yes, I was pretty happy with how things went.  
[15:42:49] <rich0> Ok, then I guess we'll wrap up.  It has been great serving with all of you!
[15:42:59] <WilliamH> Same here. :-)
[15:43:10] <blueness> Handshakes all around!
[15:43:16] <rich0> Who knows, maybe a few of us will get to repeat the adventure.  The lists are much quieter than they were last election.
[15:43:18] <dilfridge> Yep, thank you all! same from me!
[15:43:59] <blueness> well some of us are running again so maybe we'll see one another again
[15:44:08] <blueness> <mid-air!>
[15:44:14] <rich0> Ok, then we're officially done barring any emergencies, and anybody who can find me after the meeting is welcome to a free drink.  :)
[15:44:15] -*- WilliamH is curious how many have accepted their nomminations so far...
[15:44:16] <blueness> WilliamH, i was going to ask jmbsvicetto that very question
[15:44:21] <blueness> the business that anyone can nominate did not turn out so bad
[15:44:31] <blueness> the fact that you *have* to accept stopped the madness
[15:44:49] <dilfridge> https://www.gentoo.org/proj/en/elections/council/2014/council-201406-nominees.xml
[15:45:05] <rich0> blueness: yup - I raised the question just so that we didn't get any birthers after somebody gets elected and it is pointed out that no dev nominated them.  :)
[15:46:25] <rich0> Lots of incumbents this year - we're gluttens for punishment!
[15:46:30] <rich0> err, gluttons
[15:46:46] <rich0> Good thing I'm not running on a platform of being an ispell replacement
[15:48:01] <rich0> Ok, well, we're officially dismissed.  I'll post the logs/summary.