15:33:21 <abadger1999> #startmeeting fpc
15:33:21 <zodbot> Meeting started Thu Aug 15 15:33:21 2013 UTC.  The chair is abadger1999. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:33:21 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:33:39 <abadger1999> #chair geppetto limburgher tibbs|w Smoother1rOgZ racor
15:33:40 <zodbot> Current chairs: Smoother1rOgZ abadger1999 geppetto limburgher racor tibbs|w
15:35:07 * abadger1999 pulls up the agenda from geppetto
15:35:29 <abadger1999> #topic https://fedorahosted.org/fpc/ticket/323    Web Assets
15:35:38 <abadger1999> tchol: Are you around today?
15:36:34 <tchol> yup, sorry didn't notice it was time already
15:36:54 <abadger1999> No problem.
15:37:08 <abadger1999> So does anyone have any questions about these three related drafts?
15:37:15 <abadger1999> https://fedoraproject.org/wiki/User:Patches/PackagingDrafts/JavaScript
15:37:29 <abadger1999> https://fedoraproject.org/wiki/User:Patches/PackagingDrafts/Web_Assets
15:37:33 <abadger1999> https://fedoraproject.org/wiki/User:Patches/PackagingDrafts/FontsPolicy
15:37:56 <rathann2> Hi
15:38:28 <limburgher> The only thing that comes to mind is the precompiled Flash bit.
15:38:39 <tibbs|w> Are these actually ready?  It appeared to me that there was still a lot of ongoing discussion.
15:39:40 <geppetto> I thought we'd just delayed them, waiting for the reporter to be here
15:39:54 <tchol> limburgher:  i'm with you on not liking it, but it seems that it's already being done :-(
15:40:05 <tchol> limburgher: <https://fedoraproject.org/wiki/Web_Assets/Flash/Packages>
15:40:17 <limburgher> tchol:  An excellent reason to spell out forbidding it and root it out.
15:40:22 <tibbs|w> Then that stuff needs to come out; it violates other Fedora guidelines.
15:40:37 <limburgher> tibbs|w: <nods>
15:40:39 <tibbs|w> It's never been acceptable.
15:40:46 <abadger1999> #chair rathann2
15:40:46 <zodbot> Current chairs: Smoother1rOgZ abadger1999 geppetto limburgher racor rathann2 tibbs|w
15:41:44 <tchol> tibbs|w: the only thing really still under discussion is the final directory that will be exported by HTTP servers
15:42:05 <tchol> we're chatting with debian since they've run into lots of conflicts by using /javascript
15:42:16 <tibbs|w> Yeah, that's what I was thinking about.
15:42:17 <racor> i am confused about 2-3 things: somewhere the draft talks about contents and mentions "flashplayers" in the same sentence, I guess I am facing a nomenclature language barrier. To me a flashplayer is an application and so are js-applications.
15:42:25 <tchol> hopefully we can agree on something that works for everyone
15:42:58 <tchol> that's mostly tangential to the draft guidelines though.  the filesystem directories should be set in stone
15:43:20 <racor> somewhere, somebody mentioned letting /usr/share/fonts be a symlink to /usr/share/webassets/fonts. I do not agree with this. If, it should be conversely.
15:43:57 <tchol> racor: it's about flash applets that play in browsers, not the player itself
15:44:20 <racor> finally, could somebody explain the difference between webassets and traditional "html" pages /var/www etc.
15:45:05 <tchol> the backwards symlink was a typo in the draft I fixed weeks ago ;-)
15:45:20 <abadger1999> <nod> The fonts portion looks good to me now.
15:45:21 <tchol> it's defintely going to be `ln /usr/share/fonts /usr/share/web-assets/fonts`
15:46:18 <racor> tchol: OK, so far. Thanks for explaining.
15:46:28 <tchol> well for starters we're not supposed to be touching /var/www/html from packages ;-)
15:46:57 <tibbs|w> Basically I think that this is generally an improvement; the flash exception should just go and I think the font bit could be fleshed out a bit more
15:47:17 <tibbs|w> And of course someone has to pick something besides /javascript.
15:47:25 <tchol> tibbs|w: what's missing from the font section?
15:47:27 <abadger1999> racor: I think a difference is that html pages are very very seldom shared.  They're hardcoded to be used for a specific purpose.  The webassets being discussed are things that can be shared between applications -- javascipt libraries, css libraries, common icons, etc.
15:48:07 <tibbs|w> tchol: Well, it links to a non-guideline page, it kind of repeates itself ("you can use ttname" twice) and if this is common an example would be nice.
15:48:30 <tchol> I can roll the examples on that page into the guideline draft if you'd like
15:48:47 <limburgher> That might be helpful.
15:48:55 <tchol> I kind of thought it was more of an implementation detail.  you can fix the metadata any way you like, that script just makes it easy
15:51:45 <racor> wrt. html: We have many packages shipping html docs, installing them into all kind of places. How about the web-assets draft's relation to them? Are you aiming at handling them as well?
15:52:22 <tibbs|w> racor: It would be a stretch to think that these guidelines apply to that kind of thing.
15:52:34 <abadger1999> racor: I would apply it to the javascript that they install -- like jquery in python's sphinx generated docs.
15:52:43 <abadger1999> But not to any of the htmll itself.
15:52:56 <abadger1999> that's not a sharable asset.
15:53:19 <abadger1999> (javascript where that javascript is in multiple places.
15:53:20 <abadger1999> )
15:53:38 <racor> tibbs|w: aggreed, but I want to understand where to draw the line, rsp. what web-assets actually are. It's not clear to me and I guess it's not clear many others, also.
15:53:57 <tchol> exactly.  sphinx accounts for 75% of the copies of jquery in the distro.  definitely want to clean that up
15:54:33 <limburgher> tchol:  <nods>
15:54:50 <abadger1999> I guess we'd want to clarify in this section then: https://fedoraproject.org/wiki/User:Patches/PackagingDrafts/Web_Assets#Scope
15:55:20 <tchol> ok, so I added some info to the fonts draft:  https://fedoraproject.org/wiki/User:Patches/PackagingDrafts/FontsPolicy
15:55:21 * Smoother1rOgZ brb
15:55:28 <tchol> that look better?
15:56:12 <limburgher> Yes.
15:56:31 <abadger1999> +1
15:57:25 <geppetto> seems fiune to me. +1
15:57:33 * Smoother1rOgZ is back
15:58:25 <Smoother1rOgZ> +1 with info added.
15:58:55 <tchol> so I elided a chunk from scope about the scope of the directory and moved in into the section about the directory:
15:58:58 <tchol> https://fedoraproject.org/w/index.php?title=User:Patches/PackagingDrafts/Web_Assets&diff=347650&oldid=346224
15:59:18 <tchol> and the reason for that was that we really want the other policies not surrounding the directory to apply even to web app packages
15:59:28 <limburgher> That's clearer.
16:00:09 <tchol> if you ship app-specific JS along with your webapp package, it needs to not bundle
16:00:21 <tchol> if your CSS requires compiliation, it still needs to be compiled during %build
16:01:03 <abadger1999> tchol: yeah that makes sense.
16:01:05 <tibbs|w> I'm still trying to wrap my head around CSS needing compilation, but sure.
16:01:38 <tchol> tibbs|w: http://lesscss.org/
16:02:10 <abadger1999> maybe a slight wording change though: "All packages that contain static content that is only useful within the package in which they are shipped should continue to ship that content in the application's directory structure".
16:03:29 <tibbs|w> OK, that's just text processing; anything could use some intermediate language that gets turned into HTML, CSS, javascript, whatever; obviously that needs to be built like anything else.
16:04:20 <tchol> abadger1999: yeah that makes a lot more sense.  patch applied.  :-)
16:04:55 <tchol> tibbs|w:  indeed, but that's another thing that has slipped through the cracks thus far, so I really wanted to make it clear
16:06:01 <abadger1999> <nod> .
16:08:17 <tibbs|w> So, what can we actually vote on at this point?
16:08:29 <tibbs|w> Looks like there was agreement on the fonts stuff.
16:08:58 * tchol is killing the flash exception right now
16:09:42 <limburgher> I think so.
16:10:09 <tibbs|w> I'm just thinking that it would be nice to get to at least a couple of other things today if possible.
16:10:49 <abadger1999> tchol: In case it's useful -- here's a potential rewritten flash section: http://www.fpaste.org/32369/84853137/
16:10:55 <tchol> how's thiz:  https://fedoraproject.org/wiki/User:Patches/PackagingDrafts/Web_Assets#Flash
16:11:34 <geppetto> seems fine, to me.
16:11:37 <limburgher> Golden.
16:11:46 <abadger1999> tchol: +1 to your version.
16:12:08 <abadger1999> I can't find anything else wrong with any of these drafts.  Should we vote on them?
16:12:17 <limburgher> I'm game.
16:12:36 <tibbs|w> Except for the choice that directory.
16:13:05 <tibbs|w> But I'm prepared to say that as long as reasonable people can agree on something, we can just go ahead.
16:13:18 <tibbs|w> But we can't expect people to actually use much of this until that gets picked.
16:13:29 <tchol> abadger1999: i just combined the two
16:15:07 <abadger1999> Okay:  Proposal: Accept all three of the web asset related guidelines.  We'll be waiting for an update that takes care of the name of the javascript directory exported to web servers.
16:15:08 <abadger1999> +1
16:15:32 <limburgher> +1
16:15:39 <tibbs|w> +1
16:15:52 <Smoother1rOgZ> +1
16:16:07 <geppetto> +1
16:16:51 <abadger1999> Okay that's 5,  racor want to vote for the record?
16:18:25 <abadger1999> #info Web assets guidelines passed (+1:5, 0:0, -1:0)
16:18:55 <tchol> thanks, everyone!
16:19:01 <abadger1999> #topic https://fedorahosted.org/fpc/ticket/326 Bundling exception for python-kapteyn
16:19:17 <limburgher> tchol:  Thank you!
16:19:21 <abadger1999> tchol: Thanks for your work!
16:19:45 <tchol> i've added the flash stuff to the list of cleanup tasks so it doesn't get forgotten:  https://fedoraproject.org/wiki/Changes/Web_Assets#Flash
16:20:06 <limburgher> I see no answers to questions, and rathann's comment makes me question the necessity of bundling.
16:20:13 <abadger1999> <nod>
16:20:45 <limburgher> I can reiterate our request for info, saying if no answer we close this in a week?
16:20:50 <abadger1999> limburgher: wfm
16:21:18 <Smoother1rOgZ> limburgher: +1
16:21:26 <abadger1999> #topic https://fedorahosted.org/fpc/ticket/327 Python shebangs should not point to /usr/bin/python
16:21:49 <limburgher> Done.
16:22:59 <limburgher> <sigh>
16:23:28 <abadger1999> Okay, I think we should defer this a week -- I like what I think bkabrda and I have kinda hammered out but it's a bit messy and a few python packagers talked about the python guidelines at flock.
16:23:35 <abadger1999> Some of that will affect this.
16:24:12 <limburgher> +1 to letting that all get sorted out and solidified a bit.
16:24:16 <abadger1999> Especially these two pieces of discussion: https://fedoraproject.org/wiki/User:Toshio/Flock2013_Python_Guidelines#Shebang_lines
16:24:31 <abadger1999> https://fedoraproject.org/wiki/User:Toshio/Flock2013_Python_Guidelines#Wheel:_the_new_upstream_distribution_format
16:25:09 * Smoother1rOgZ is ok to defer
16:25:26 <abadger1999> That also leads to -- I'll be trying to turn those notes into one or more drafts to update the guidelines in the next week or two.  Will probably pass the drafts by the python-sig or packaging mailing list before they come to FPC.
16:25:48 <abadger1999> #topic https://fedorahosted.org/fpc/ticket/328 Soft-static uid/gid allocation for Performance Co-Pilot
16:27:14 <abadger1999> Looking at this request, I'm not sure if this has a use case or not.
16:27:21 <limburgher> I see nothing here that makes static allocation necessary.  Even if pmlogger is getting remote info, it's still typically writing to local storage, no?
16:27:49 <abadger1999> It doesn't look like the data-in-files are exported via a shared filesystem; they're served by one of the daemons instead.
16:28:01 <abadger1999> If that's the case, then it should not need a static id.
16:28:06 <geppetto> yeh, nothing requires shared uids across systems
16:29:44 <abadger1999> Want me to confirm on the ticket to make sure that's the case?  Then we can vote next week pretty easily (or the submitter might close it themselves).
16:30:02 <limburgher> abadger1999:  Please do.
16:30:24 <abadger1999> #action abadger1999 to confirm that our understanding of Performance co-pilot is correct
16:30:30 <abadger1999> #topic Open Floor
16:30:50 <abadger1999> We're at the end of our hour, is there anything (ticketed or unticketed) that anyone wants to bring up?
16:31:11 * Smoother1rOgZ looks
16:31:27 <racor> abadger1999: sorry, was distracted on the phone. My vote on the web-assets: +1, I still do not understand what web-assets are, do not understand what this would be useful for, but I don't want to hold up things.
16:31:42 <abadger1999> racor: Thanks
16:32:06 <abadger1999> One thing I wanted to bring up from fesco and flock -- we should reconsider scl guidelines.
16:32:25 <Smoother1rOgZ> ticket 332, this should be a quick one
16:32:31 <abadger1999> .fpc 332
16:33:15 <tibbs|w> https://fedorahosted.org/fpc/ticket/332
16:33:40 <abadger1999> #topic Bundling exception for IQmol https://fedorahosted.org/fpc/ticket/332
16:33:56 <abadger1999> So -- I haven't looked at those libs yet -- are they copylibs?
16:34:10 <tibbs|w> I went to the site, where it's not obvious.
16:34:12 <Smoother1rOgZ> they are.
16:35:08 <tibbs|w> Smoother1rOgZ: That seems to conflict with what you wrote in the ticket.
16:35:39 <limburgher> I was thinking that as well.
16:36:13 <Smoother1rOgZ> unless I missunderstood copylibs here
16:36:24 <Smoother1rOgZ> -s*
16:36:46 <tibbs|w> A library not intended to be packaged separately.
16:36:56 <tibbs|w> You're supposed to copy the code into your project.
16:37:05 <limburgher> https://fedoraproject.org/wiki/Packaging:No_Bundled_Libraries#Copylibs
16:37:29 <limburgher> tibbs|w: An irritating practice but a reality nonetheless.
16:37:49 <Smoother1rOgZ> tibbs|w: ha...well then, they're not.
16:38:21 <limburgher> Then I concur, -1.
16:39:29 <abadger1999> Hmm... Looks like they're code, don't have any makefiles or other build scrupts at all.
16:39:48 <abadger1999> But it doesn't look like they're intended to be modified per se.
16:40:35 <abadger1999> otoh, I don't think gnulib or egglib are intended to be modified.
16:40:40 <Smoother1rOgZ> abadger1999: they do have *.pro file which build Makefile
16:40:44 <abadger1999> ah
16:40:48 <limburgher> So it's less like they're copylibs and more like they're just poorly released?  Or are they more like some of these head-only packages?
16:41:46 <abadger1999> Smoother1rOgZ: What do you get out of running make ?  Are they .o files or a .so?
16:42:13 <Smoother1rOgZ> you got both at the end
16:42:19 <racor> -1. At least qmsgbox looks like a usually qmake-based library, which (qmake typically) lacks the proper system-integration.
16:42:24 <limburgher> Versioned or unversioned?
16:43:17 <abadger1999> k  It's looking more like it's just poorly released.
16:44:23 <limburgher> Might this be an upstream education moment?
16:44:29 <abadger1999> Smoother1rOgZ: you or I might be able to help the packager figure out how to build the code... should we look at that this week and report that on the ticket?
16:44:52 <racor> abadger1999: qmsgbox's github appears dead ;)
16:44:56 <Smoother1rOgZ> we should, yep
16:45:42 <abadger1999> Okay.
16:46:11 <abadger1999> #action Smoother1rOgZ and abadger1999 will help the packager figure out how to package these other pieces of code as libraries.
16:46:25 <abadger1999> We can close out the ticket next week if that's successful.
16:46:31 <abadger1999> #topic Open Floor
16:46:57 <abadger1999> #topic Revisit SCLs
16:46:59 <limburgher> abadger1999:  So what did you mean by scl?
16:47:01 <limburgher> jinx
16:47:06 <abadger1999> Software Collections
16:47:20 <abadger1999> There was a lot of talk about these at flock.
16:47:33 <limburgher> I should start going to stuff.
16:47:35 <geppetto> where is mattdm?
16:48:39 <tibbs|w> So I'm guessing this isn't about actually banning the macros from Fedora specs.
16:48:54 <abadger1999> tibbs|w: Right... It's about making them more official.
16:48:58 <tibbs|w> Ugh.
16:49:19 <geppetto> abadger1999: So current Fedora is "latest version of everything, in one place, if you want change we have it"
16:49:40 <abadger1999> Talking with mattdm I realized that I had misunderstood the concept before... I thought they were to be used for bundling but htey're really just a means of parallellizing installs.
16:49:41 <geppetto> abadger1999: SCLs are a solution to the problem of "my stuff doesn't change fast enough, so I need another version that is newer to install at the same time"
16:49:56 <geppetto> These are not compatible, IMO.
16:50:07 <abadger1999> geppetto: ah -- they can be used for the opposite as well:
16:50:26 <geppetto> abadger1999: Sure, but we frown _heavily_ on compat-* packages in Fedora too.
16:50:31 <abadger1999> my stuff hasn't kept up with Fedora's default libraries.  I need a backwards compat option until my upstream catches up.
16:50:41 <racor> sorry folks, I have to leave. bye.
16:51:03 <abadger1999> racor: thanks for coming!  This is more of an fyi -- I doubt we'll vote/have a draft today.
16:51:45 <abadger1999> geppetto: I suppose.  However, we do have compat package precedents.
16:52:01 <geppetto> also there's a fine line between "I need old versions until my upstream catches up" and "I'm bundling python because that's easier"
16:52:08 <geppetto> abadger1999: sure.
16:52:39 <abadger1999> python26  in epel.  python2 and python3 packages (sorta) in Fedora.  individual python-foo$VER packages.
16:53:13 <geppetto> Yeh, I'm not saying we've never done parallel installable versioned packages in Fedora … but it's far from the norm.
16:53:15 <abadger1999> I think that in the past it was fesco that established how strictly we frowned on compat-* packages.
16:53:43 * abadger1999 remembers the python compat to enable zope ruling came from fesco.
16:53:47 <tibbs|w> And now fesco seems to be going off the deep end and thinking it's OK all of a sudden.
16:54:42 <limburgher> geppetto:  I think it's a great thing to do in RHEL/EPEL land, but I question it's utility in Fedora.
16:55:00 <tibbs|w> But I guess whatever they tell us they want, we'll figure out how to implement.
16:55:20 <limburgher> At the same time, I question the harm.  If there's someone willing to maintain a large stack and it doesn't interfere. . .
16:55:56 <tibbs|w> But we've been here before.
16:56:24 <tibbs|w> Remember when the python maintainer didn't want someone else maintaining their own separate python stack because they'd get all the bugs?
16:56:46 <tibbs|w> Or the kernel maintainers having to deal with bugs from someone else maintaining their own modules?
16:56:48 <limburgher> Right.
16:57:46 <tibbs|w> I wonder if we'd still have httpd 2.2 packages around to support that MPM thing from a few weeks back.
16:58:09 <tibbs|w> Anyway, the point is that it's not as simple as "someone wants to maintain this old stack; let's let it in".
16:58:12 <limburgher> User confusion will always be with us.  I just experienced it myself, where I tried to assign a bug to another component, where the other owner and I were in agreement that that's where the problem was, and the user objected.
16:58:21 <geppetto> limburgher: So, Eg. you'd be happy letting the httpd-tik stuff happen as SCL?
16:58:27 <limburgher> tibbs|w: Right.
16:58:52 <limburgher> geppetto:  Not at all.  httpd22, on the other hand, maybe.
16:58:54 <geppetto> Yeh, tibbs beat me to it … but that
16:59:16 <limburgher> Honestly I'm not thrilled by any of it.
16:59:20 <abadger1999> geppetto: possibly... I think we would have the opportunity to define "parallelizable" a bit more strictly than httpd-itk takes advantage of, though.
16:59:34 <limburgher> this. ^^^^^^^^^^^^^^^^^^^^^
17:00:36 <geppetto> I doubt it … I think the best we could do is have a big "we have a giant ban hammer, so you better have a really good reason to use SCL and the resources to maintain it"
17:01:12 <geppetto> Kind of like bundling
17:01:46 <limburgher> Yeah.  Inelegant, but it may come to that.
17:02:56 <abadger1999> Does httpd-itk build without bundling against httpd-2.2?  If so, it probably could go in an scl but there's other factors too -- things that take a reserved port can't run in parallel even if they can be installed in parallel... I'm kinda shying away from taking advantage of that but we could.
17:03:40 <abadger1999> The python-zope precedent was also that the maintainer had to agree to the parallel stack... that's a possibility here too.
17:03:54 <abadger1999> I think the latter is a fesco question though.
17:04:15 <abadger1999> The former (conflicting ports) would be up to us.
17:05:04 <limburgher> That doesn't worry me so much, as long as it's not something enabled by default, which most things aren't anyway.
17:05:13 <abadger1999> Anyhow -- I wanted to make sure everyone knew this was going to be coming up soon and that fesco is generally in favor of us coming up with something.
17:05:31 <limburgher> And if you install httpd and httpd22 and expect them both to bind to port 80. . .well. . .you're silly.
17:05:32 <abadger1999> (for a broad range of something)
17:05:53 <Smoother1rOgZ> heh
17:06:04 <abadger1999> limburgher: round robin! j/k ;-)
17:06:25 * limburgher looks around for large mallet
17:06:38 <abadger1999> I think we need to wait for mattdm to give us a draft to actually vote on something.
17:06:44 <abadger1999> limburgher: Ouch!
17:06:56 <limburgher> abadger1999:  Oh, you found it!
17:07:22 <abadger1999> Alright, anything else for open floor?
17:07:23 <limburgher> Agreed.  Plus, I'm starting to see E_BLOODSUGAR, so it might good to wrap up for the day.
17:07:34 * abadger1999 will close in one minute
17:08:04 <abadger1999> 30s
17:08:38 <abadger1999> #endmeeting