fpc
LOGS
17:02:54 <abadger1999> #startmeeting fpc
17:02:54 <zodbot> Meeting started Thu Mar  6 17:02:54 2014 UTC.  The chair is abadger1999. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:02:54 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:03:00 <abadger1999> #meetingname fpc
17:03:00 <zodbot> The meeting name has been set to 'fpc'
17:03:04 <abadger1999> #topic Roll Call
17:03:15 <abadger1999> RemiFedora said he might not be able to make it
17:03:22 <abadger1999> #chair geppetto
17:03:22 <zodbot> Current chairs: abadger1999 geppetto
17:03:34 * RemiFedora here for 30-40' max
17:03:35 <willb__> howdy FPC
17:03:37 <abadger1999> I think racor said the same last week
17:04:11 * limburgher is here but has been sick the last 2 days and is up to my larynx in backlog. . .
17:04:33 <abadger1999> tibbs, SmootherFrOgZ, spot: FPC meeting time if you're around
17:04:40 <abadger1999> #chair limburgher RemiFedora
17:04:40 <zodbot> Current chairs: RemiFedora abadger1999 geppetto limburgher
17:05:09 * Rathann here
17:06:17 * nonamedotc is here to follow discussion on his ticket. :)
17:06:25 <abadger1999> #chair Rathann
17:06:25 <zodbot> Current chairs: Rathann RemiFedora abadger1999 geppetto limburgher
17:06:34 <abadger1999> Okay, that's temporarily quorum.
17:06:38 <tibbs|w> Howdy.
17:06:45 <abadger1999> #chair tibbs
17:06:45 <zodbot> Current chairs: Rathann RemiFedora abadger1999 geppetto limburgher tibbs
17:07:36 <abadger1999> mattdm: You around for Go Guidelines?
17:08:06 <abadger1999> #topic #385     workarounds for rpm symlink <-> directory issue
17:08:11 <abadger1999> https://fedorahosted.org/fpc/ticket/385
17:08:15 <abadger1999> I updated the scripts.
17:08:54 <abadger1999> panu had some more comments about using .rpmsave and the description.
17:09:03 <abadger1999> I'll update those this week.
17:09:17 <abadger1999> If others would like to review the updated lua script logic, that would be great.
17:09:29 <abadger1999> #topic #391     Exception for bundled libraries in icecat
17:09:50 <abadger1999> This is likely to be a long topic.
17:10:41 <abadger1999> So... I wrote up the points I thought we raised at the last meeting.
17:10:46 <abadger1999> Where are we now?
17:12:52 <tibbs|w> I guess basically they disable some things or we allow them to bundle.
17:12:59 <abadger1999> <nod>
17:13:08 <geppetto> I thought we were mostly in agreement
17:13:16 <geppetto> let them do what firefox is doing
17:13:17 <abadger1999> I don't really see any standard reasons to bundle.
17:13:35 <abadger1999> There's the fairness aspect.
17:15:00 <abadger1999> I think I'm 0 for that being grounds to accept bundling... firefox is speshul because of the trademark issue... that doesn't exist here.
17:15:45 <abadger1999> If others like that though (spot liked the fairness angle too) then maybe we should decide on whether firefox should get to bundle?
17:15:54 <Rathann> I'm -1 for permanent exception, +1 for temporary exception while unbundling is being worked on
17:15:59 <tibbs|w> I could go for a temporary exception.
17:16:01 <Rathann> (for icecat)
17:16:11 <abadger1999> Yeah, I could definitely go for a temporary exception as well.
17:16:41 <tibbs|w> I assume that the bundling is necessary because of changes that haven't yet been upstreamed.
17:17:13 <abadger1999> tibbs|w: For some of the libraries I know that to be true.  I'm not sure if it covers all of them, though.
17:17:45 <tibbs|w> Otherwise there's no point in a temporary exception.  It's not entirely clear if there's any benefit.
17:18:03 <abadger1999> firefox upstream seems to take the position that they want to control the libraries that they're using because they're more likely to care about issues than the library upstreams.
17:18:17 <abadger1999> Okay -- so that may be something to ask.
17:18:49 <abadger1999> #info ask whether all the libraries requesting an exception have a non-upstreamed patch to unbundle.
17:20:21 <geppetto> yeh, I mean that's probably true when they are distributing it from their website … not so much when we are distributing it in Fedora
17:20:58 <tibbs|w> It's just that attitude breeds the proliferation of other non-upstreamed patches.
17:21:03 <RemiFedora> let icecat prove it can do a better job than firefox ;)
17:21:28 <tibbs|w> I don't think icecat aspires to be anything other than a trademark-free version, though.
17:21:44 <abadger1999> Okay, anything else to ask the ticket owner?
17:23:04 <abadger1999> yeah... which kinda seems to defeat the purpose... (if the problem with the trademark licensing is that it prevents diverging from mozilla then why remove the trademark problem with no intention to fix oustanding issues?)
17:23:33 <geppetto> abadger1999: don't start with all that logic stuff :)
17:23:36 <abadger1999> #topic #397     Please, choose an ID number for a new group called
17:23:42 <abadger1999> https://fedorahosted.org/fpc/ticket/397
17:24:15 <abadger1999> jcapik  replied to our questions from last week.
17:24:30 <abadger1999> but I guess I'm behind the times.
17:25:23 <abadger1999> Anyone know how  he envisions "Accessing the devices remotely" without software translation?
17:25:26 <geppetto> I don't understand his answer
17:26:37 <abadger1999> Alright -- I guess we're all lost as to how he sees that working.
17:26:46 <tibbs|w> Yeah, I don't quite get it.
17:26:57 <abadger1999> Does someone else want to try asking for clarification this time?
17:26:58 <tibbs|w> The ticket basically asks us to do the impossible.
17:27:31 <tibbs|w> And when did we get into the business of actually choosing the numbers anyway?
17:28:00 <abadger1999> I was going to forward the request on to ovasik/setup owner.
17:29:30 <geppetto> I will add a comment
17:29:48 <abadger1999> geppetto: Thanks!
17:30:15 <abadger1999> Okay, new business
17:30:30 <abadger1999> nonamedotc, willb__: are either of you here for a specific ticket?
17:30:40 <willb> I am
17:30:44 <nonamedotc> I am here for #399
17:30:48 <willb> #401
17:31:14 <rsquared> I'm here for #392
17:31:37 <abadger1999> #topic #399     request for bundled library exception - clustal omega
17:31:42 <abadger1999> https://fedorahosted.org/fpc/ticket/399
17:34:08 <RemiFedora> seems to be a fork
17:34:52 <abadger1999> yeah.
17:35:09 <abadger1999> nonamedotc: Would upstream be interested in maintaining this as a true fork?
17:35:20 <nonamedotc> nope. it's an academic lab.
17:35:24 <abadger1999> rename the library and  release it for others to use?
17:35:25 <abadger1999> k
17:35:36 <nonamedotc> they have no interest in changing it any way other than their needs
17:35:45 <nonamedotc> their answer was "it is gpl"
17:36:37 <abadger1999> nonamedotc: Did they attempt to send changes to the squid author and they were refused or did they just assume that the squid author wouldn't take them?
17:36:53 <RemiFedora> hard stop for me, bye
17:36:53 <nonamedotc> i contacted the squid author - he was not interested
17:37:10 <nonamedotc> clustal upstream did not send at all
17:37:24 <nonamedotc> they told me they did not send since changes were specific to clustal
17:37:42 <abadger1999> k
17:37:51 <tibbs|w> This is one of those academic things where nobody cares about anything other than getting their code to run.
17:38:07 <abadger1999> yeah and both upstream's are in that category.
17:38:07 <nonamedotc> tibbs|w: exactly.
17:40:06 <tibbs|w> I guess there are only two choices: allow bundling permanently or not include the software.
17:40:30 <abadger1999> nonamedotc: for the record, if it's dealing with user input, there are security implications; otoh, the software itself might not be used in a place where that matters much.
17:40:53 <abadger1999> So I don't think we have good precedent for this but we could set some.
17:41:04 <abadger1999> I'd say this library falls under the idea of copylib.
17:41:31 <nonamedotc> abadger1999: sorry - i should not have made that statement.
17:41:33 <abadger1999> But in the past we've tried not to blanket allow cpoylibs because they're still full of fail.
17:41:46 <Rathann> sorry guys I have to leave
17:42:04 <geppetto> doesn't anything else ship this?
17:42:14 <abadger1999> Rathann: bye!
17:42:15 <tibbs|w> Did Europe have a DST transition already?
17:42:36 <geppetto> I guess I'd be ok with them "bundling" it until something else uses it too, and then we'd make them turn it into a real lib. and share it.
17:42:38 <abadger1999> I think that may have dropped us below quorum.
17:44:25 <abadger1999> geppetto: We could... otoh, that does set the stage for future failure (ie: it's unlikely that hte upstream positions will change in the meantime so everyone will just be hoping that no other software we want in Fedora makes use of that library).
17:44:30 <geppetto> we had 6 at one point
17:45:18 <abadger1999> nonamedotc: For this questeion "    Is upstream keeping the base library updated or are they continuously one or more versions behind the latest upstream release? "  was your answer related to squid upstream or to clustal omega upstream?
17:45:44 <nonamedotc> squid upstream - they do not have releases
17:45:56 <nonamedotc> they are providing a single tarball
17:45:59 <abadger1999> nonamedotc: k.  Do you know how that appies to clustal omega?
17:46:13 <abadger1999> nonamedotc: Do they resync their code whenever squid produces a new tarball?
17:46:26 <nonamedotc> clustal omega is actively developed, as far as I know.
17:47:00 <nonamedotc> I doubt it but I can ask - clustal omega's efforts are in optimizing the sequence alignment algorithm
17:47:17 <abadger1999> k
17:48:22 <abadger1999> If they are good about keeping their code up-to-date wrt the code they're bundling it makes us less concerned that bugs fixed in the library upstream will stick around in the application doing the bundling.
17:49:19 <nonamedotc> I am not sure if the (squid) upstream even does releases anymore
17:49:37 <abadger1999> I'd like to add something to the guidelines that allows this... but I'm not sure what... "It's a copylib and XXXX.  Together those reasons make us comfortable with allowing bundling".
17:50:43 <nonamedotc> just for information - squid upstream last "release" was in 2002.
17:51:39 <tibbs|w> Oh, my.
17:51:39 <tibbs|w> At least it was this century.
17:52:11 <abadger1999> Ohhh -- here's a precedent: "TexStudio contains a forked copy of qcodeedit 2, which is at least two years dead. Since TexStudio is the only user, there is no benefit to a separated library, and permission to bundle has been granted. "
17:52:31 <geppetto> this millenium!
17:52:51 <abadger1999> So that could go along with geppetto's idea that we allow bundling unless someone else starts using squid as well.
17:54:53 <abadger1999> Proposal: "clustal omega has a forked copy of squid[LINK].  Since squid was last updated in 2002, is a copylib, *and* clustal omega is the only user, there is currently no benefit to a separate library.  Permission to bundle has been granted unless some of those criteria change."
17:55:06 <abadger1999> +1
17:55:11 <geppetto> +1
17:55:14 <tibbs|w> We need to clarify "the only user" or "the only user in Fedora".
17:55:20 <abadger1999> tibbs|w: <nod>
17:55:43 <tibbs|w> I'm not sure I'm entirely comfortable with either of those, honestly.
17:55:51 <abadger1999> Proposal: "clustal omega has a forked copy of squid[LINK].  Since squid was last updated in 2002, is a copylib, *and* clustal omega is the only user in Fedora, there is currently no benefit to a separate library.  Permission to bundle has been granted unless some of those criteria change."
17:56:29 <tibbs|w> "the only user which is likely to ever be part of Fedora" seems more reasonable, I guess, but even that's splitting hairs.
17:56:57 <abadger1999> yeah, I'm not entirely comfortable with it either (as it leaves the door open to later having to remove clustal omega)... but now that I've found precedent, I'm not against it.
17:58:47 <abadger1999> Okay, we're at +2 and don't have quorum besides.
17:59:01 <abadger1999> I'll put the proposal i nthe ticket and we can see if we can collect more votes there.
17:59:22 <tibbs|w> I'll drop in my +1 either way, but if we're going to make a general rule I think my last proposal makes the most sense.
17:59:44 <nonamedotc> Can i provide the link for squid for the sake of completeness? :)
18:00:03 <abadger1999> nonamedotc: Sure
18:00:13 <abadger1999> #topic 401 (reverse) bundling exception for stream-lib
18:00:17 <abadger1999> https://fedorahosted.org/fpc/ticket/401
18:00:23 <abadger1999> willb: Okay, you're up.
18:00:25 <nonamedotc> squid link - http://selab.janelia.org/software.html
18:00:31 <nonamedotc> thanks all!
18:00:38 <willb> ok, https://fedorahosted.org/fpc/ticket/401
18:01:05 <willb> stream-lib has forked a data structure used internally in Apache Cassandra
18:01:19 <willb> and made it suitable as a reusable component for other thigns
18:01:22 <willb> things, even
18:01:51 <willb> I have a package under review now that depends on stream-lib for some functionality (that I've had to disable as a result), so I'd like to package stream-lib
18:02:21 <willb> The code has basically diverged from Cassandra, but they've added in some algorithmic improvements since forking
18:03:45 <abadger1999> I think we'd consider this forking.
18:03:58 <tibbs|w> I agree.
18:04:23 <abadger1999> Proposal: stream-lib's use of Cassandra code is considered forking and therefore allowed.
18:04:27 <abadger1999> +1
18:05:16 <tibbs|w> +1
18:06:46 <limburgher> +1
18:07:59 <abadger1999> geppetto: Want to vote before I take this to the ticket?
18:08:38 <geppetto> +1
18:08:52 <geppetto> abadger1999: sorry, in another meeting as well … I'm around though
18:08:59 <abadger1999> Cool
18:09:08 <willb> thanks, FPC!
18:09:19 <abadger1999> #info stream-lib's use of Cassandra code is considered forking and therefore allowed.  (+1:4, 0:0, -1:0) => Taking to ticket for the remaining vote.
18:09:32 <willb> I'll start packaging and put it up for review after the last vote comes in
18:10:16 <abadger1999> #topic 392 bundling expception for greenmail
18:10:19 <abadger1999> https://fedorahosted.org/fpc/ticket/392
18:12:25 <rsquared> What a recap of greenmail situation?
18:13:19 <abadger1999> rsquared: sure.
18:14:47 <rsquared> greenmail is a fake e-mail server written in java.  It is useful for applications that need to interface with an e-mail server to test against greenmail instead of a real e-mail server.
18:15:25 <rsquared> greenmail bundles code from foedus and james for some of it's protocol implementations, but modifies them heavily for its purpose.
18:15:46 <rsquared> The foedus bundling was previously approved as a fork.
18:16:59 <abadger1999> I'm inclined to say this is a fork as well.
18:17:16 <rsquared> greenmail hasn't updated the james based source since the initial fork except for minor changes related to greenmail functionality.  The james code that is bundled is pretty heavily modified.  I diffed the files and got a 9000+ line diff, which I pared down to 7000+ by removing comments and code style changes
18:18:48 <abadger1999> james is an application (so wouldn't be directly usable).  james is a real imap server, greenmail is a fake imap server for testing (so there's a difference in purpose).  greenmail based work on an older release of james and has been making code changes to it ever since (so there's a large divergence).
18:20:46 <abadger1999> Anyone else have thoughts or should I just make the proposal?
18:21:44 <geppetto> yeh, just do it
18:21:57 <abadger1999> Proposal: greenmail's use of james is considered a fork and therefore allowed.
18:21:59 <abadger1999> +1
18:22:03 <tibbs|w> +1
18:22:06 <geppetto> I'm mostly fine about people forking to do new things
18:22:11 <geppetto> +1
18:23:17 <limburgher> +1
18:23:43 <abadger1999> #info greenmail's use of james is considered a fork and therefore allowed. (+1:4, 0:0, -1:0) => Taking to ticket for the remaining vote
18:23:49 <rsquared> Thanks everyone!
18:24:01 <abadger1999> Back to the agenda items
18:24:03 <abadger1999> #topic #398     Tilde in version
18:24:08 <abadger1999> https://fedorahosted.org/fpc/ticket/398
18:24:33 <tibbs|w> "not supported by all Fedora versions" was only part of the reason this terrible thing was kept out.
18:24:55 <abadger1999> I'm still -1
18:25:06 <geppetto> You just think it looks ugly?
18:25:33 <abadger1999> geppetto: It's like list comprehensions in python.
18:25:41 <abadger1999> If you see it, you have no idea what it does.
18:25:50 <abadger1999> And you can't google for it.
18:26:03 <abadger1999> without already knowing something about what it is.
18:26:05 <limburgher> I don't see the benefit.
18:26:20 <abadger1999> Since our guidelines already handle this I don't see why we should add it.
18:28:23 <geppetto> ok, that makes sense … less magic is bad. I don't mind it after you understand the magic, and it's what debian does … but I'm not a big advocate.
18:29:34 <abadger1999> Proposal: Change guidelines to allow tilde in version
18:29:35 <abadger1999> -1
18:29:47 <limburgher> -1
18:31:04 <abadger1999> tibbs|w, geppetto: care to vote and then I'll ask for remaining votes in the ticket.
18:31:11 <tibbs|w> -1
18:31:25 <geppetto> seems kind of pointless :)
18:31:37 <geppetto> 0
18:32:12 <abadger1999> #info Change guidelines to allow tilde in version (+1:0, 0:1, -1:3) => Will go to the ticket to seek remaining votes.
18:32:31 <abadger1999> #topic #400     Exception for bundled library FoX in exciting
18:32:35 <abadger1999> https://fedorahosted.org/fpc/ticket/400
18:34:13 <tibbs|w> I think the last commend here is on point.
18:34:24 <tibbs|w> Haven't had much time to look over this one, though, since it just came in yesterday.
18:36:09 <abadger1999> <nod>
18:36:33 <tibbs|w> I'm generally against bundling just to change some compile-time option, though.
18:36:40 <abadger1999> also, thinking that the library won't *read* non-pretty-printed xml seems.... fishy.
18:38:39 <abadger1999> Hm... although the packager apparently has found that the program crashes if the unmodified library is used.
18:44:15 <abadger1999> heh.  fortran... not one of the languages I'm currently fluent in.
18:44:19 <abadger1999> k.
18:45:07 <abadger1999> I'm going to update the ticket to say that we lean towards just fixing the code (look into using the invalid-pretty-print option that FoX upstream is proposing) or comment#1's idea of two compilations of the library.
18:45:13 <abadger1999> but haven't taken a vote yet.
18:45:18 <abadger1999> #topic Open Floor
18:45:39 <abadger1999> One thing of note is that we're probably going to need to revisit the Conflicts guidelines soon.
18:45:50 <abadger1999> I'm going to talk to sgallagh tomorrow about it.
18:46:00 * sgallagh is around now, actually
18:46:10 <abadger1999> okay.
18:46:22 <abadger1999> sgallagh: If you want to talk briefly about it, we can start thinking about it.
18:46:31 <sgallagh> Sure
18:46:43 <abadger1999> we have 10 minutes and ~4 FPC members still here.
18:46:58 <sgallagh> So, one thing that several of us discussed at DevConf was the need for some of the products to have differing default configurations
18:48:21 <sgallagh> One potential solution to this would be for the "default" configuration settings to be broken out into subpackages that Provides: package-config
18:49:00 * limburgher here about that much longer
18:49:02 <sgallagh> And the base package would be able to satisfy a requirement from either default config
18:50:10 <sgallagh> Naturally, these config subpackages would need to Conflict explicitly, since they'd be providing the same files.
18:51:58 <abadger1999> sgallagh: if we can guarantee that nothing depends on the precise subpackages this sounds doable.
18:52:15 <sgallagh> It may also need to lead into a discussion of "soft deps" to deal with mixing and matching the fedora-server-release and fedora-workstation-release packages on a system
18:53:41 <abadger1999> sgallagh: having something depend on server-httpd-config would be a problem as it would mean you could have a conflict that propogates up a stack of packages.
18:53:57 <sgallagh> Yes, I realize that
18:53:58 <abadger1999> sgallagh: what would soft deps solve in this case?
18:54:21 <sgallagh> fedora-server-release Recommends: httpd-config-server
18:54:24 <geppetto> sgallagh: what is a default configuration?
18:54:24 <abadger1999> It seems like you'd be able to mix and match fine just by virtue of hte virtual provides.
18:54:34 <sgallagh> If httpd-config-workstation was already present, that would be ignored
18:54:34 <geppetto> sgallagh: Have you thought about alternatives?
18:55:11 <sgallagh> abadger1999: The main question remains: "If I start with a Server install, can I also install a meta-package to become Workstation?"
18:55:32 <sgallagh> geppetto: These are default configs that *must* be made up entirely of %config files
18:55:36 <sgallagh> So alternatives would be a waste
18:55:40 <abadger1999> sgallagh: I don't think soft deps helps you there.
18:56:19 <sgallagh> abadger1999: It depends on the implementation. One that I've been pushing for is allowing a dep "language"
18:56:27 <abadger1999> the tool that transitions from a server to a workstation just needs to know which of several package choices it should prefer.
18:56:39 <sgallagh> I.e. When installing this package, if $other_package is installed, Require this one.
18:56:46 <sgallagh> Or "Recommends" this one
18:56:50 <abadger1999> that preference can come from a soft dep in the meta-package, but it could also come from something like comps.
18:57:10 <sgallagh> Well, my goal was for the "tool" to just be "yum", if possible
18:57:50 <abadger1999> The tool also has the chance of doing somethings wrong... it's going to be "reseting" the config to a default state in some ways rather than fully accomodating what the sysadmin might have set up.
18:58:03 <geppetto> Well in theory you can create a new tool that's alternatives like so all the packages put their config. files somewhere and the tool copies them into /etc or whatever
18:58:10 <sgallagh> Yeah, I do *not* want a package installation to be able to change existing config
18:58:33 <abadger1999> (say I had a servre install and I manually replaced srver-httpd-config with cloud-httd-config.  The tool would likely replace my cloud-httpd-config with workstation-httpd-config)
18:58:34 <geppetto> I mean you kind of want/need that anyway, right? As the config. package can't ship a new version of a config. file owned by another package.
18:59:06 <sgallagh> abadger1999: I'd say outright that this would be incorrect behavior.
18:59:36 <sgallagh> But I can see where it would be difficult to address that programmattically
19:00:45 <abadger1999> alright... our hour is up.
19:01:01 <sgallagh> abadger1999: Continue tomorrow, then?
19:01:03 <abadger1999> So on conflicts -- I guess we need to hash out conflicts vs alternatives for this use case.
19:01:12 <abadger1999> soft deps might be a red herring.
19:01:26 <abadger1999> sgallagh: yeah -- we can continue tomorrow and see if we can make a proposal
19:01:31 <sgallagh> Sounds good
19:01:44 <abadger1999> geppetto: Do you want in on that?  sgallaghand I are going to discuss it on #fedora-devel
19:02:25 <abadger1999> sgallagh: 1PM EST, correct?
19:02:39 <sgallagh> That's what I have in my calendar, yes
19:02:42 <abadger1999> Cool.
19:02:56 * abadger1999 will close the meeting in 60s
19:03:06 <geppetto> sitll in the other meeting
19:03:17 <nonamedotc> thanks FPC for input on my ticket!
19:03:33 <abadger1999> geppetto: Cool.  Will be tomorrow, 1PM EST on #fedora-devel if you're interested.
19:03:49 <geppetto> I'm not sure I can add much … usual stuff of "shipping config. in rpms has always sucked. And it still does. Abusing conflicts isn't the end of the world, I guess"
19:03:53 * geppetto nods
19:04:25 <abadger1999> #endmeeting