17:30:01 #startmeeting FESCO (2011-03-23) 17:30:01 Meeting started Wed Mar 23 17:30:01 2011 UTC. The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:30:01 Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:30:01 #meetingname fesco 17:30:01 #chair mclasen notting nirik SMParrish kylem ajax cwickert mjg59 mmaslano 17:30:01 #topic init process 17:30:01 The meeting name has been set to 'fesco' 17:30:01 Current chairs: SMParrish ajax cwickert kylem mclasen mjg59 mmaslano nirik notting 17:30:06 ah, cool 17:30:16 i was worried i'd have to remember meetbot. 17:30:23 :) 17:30:40 I'm here and on for now, but have been having network issues today, so if I drop off someone might have to take over. 17:30:43 Your Internet working properly yet? 17:30:52 Oof 17:31:12 * cwickert is here 17:31:15 * notting is here 17:32:02 here but mobile 17:32:31 * mmaslano here 17:32:49 ok, I guess lets go ahead and dive in. 17:32:57 #topic #516 Updates policy adjustments/changes 17:32:57 .fesco 516 17:32:58 nirik: #516 (Updates policy adjustments/changes) - FESCo - Trac - https://fedorahosted.org/fesco/ticket/516 17:33:16 I didn't add any new items this week, because it's kinda getting to the bottom of the barrel on suggestions. ;) 17:33:29 https://fedoraproject.org/wiki/User:Kevin/Updates_Policy_Changes_Ideas_Container 17:33:58 many of the ones left are really "hey this would be nice if someone would implement it" 17:34:19 like more testers, or notifying testers more, or having test instances. 17:34:31 or implementing packagekit integration in testing more. 17:34:45 or improve fedora-easy-karma 17:35:08 These are more "great, please work on them" rather than policy we can add, IMHO. 17:35:22 So, does anyone see anything left in the ideas container we should look at adding to policy? 17:35:23 +1 17:35:30 not ATM 17:36:59 proposal: close out the ideas container, ask people to file new ideas as a new ticket to be addressed. 17:37:15 +1 17:37:59 +1 17:38:09 +1. 17:38:16 +1 from me on my own proposal. ;) 17:38:58 more votes? 17:39:37 +1 17:39:46 #agreed will close out the ideas container, ask people to file new ideas as a new ticket to be addressed. 17:39:53 #topic #515 Investigate a "features" repo for stable releases 17:39:53 .fesco 515 17:39:54 nirik: #515 (Investigate a "features" repo for stable releases) - FESCo - Trac - https://fedorahosted.org/fesco/ticket/515 17:39:58 cwickert: any news on this? 17:40:11 well, at least I have started 17:40:21 but it's not ready for discussion yet 17:40:23 sorry 17:40:52 ok, no problem at all. 17:41:03 #topic #517 Updates Metrics 17:41:03 .fesco 517 17:41:04 nirik: #517 (Updates Metrics) - FESCo - Trac - https://fedorahosted.org/fesco/ticket/517 17:41:15 kylem: were you going to look more into this? anything to report? 17:41:48 i think i'll have to give it back to the pool, i've still not had a chance for this 17:42:03 ok, no worries. ;) 17:42:11 #topic #563 suggested policy: all daemons must set RELRO and PIE flags 17:42:11 .fesco 563 17:42:13 nirik: #563 (suggested policy: all daemons must set RELRO and PIE flags) - FESCo - Trac - https://fedorahosted.org/fesco/ticket/563 17:42:14 how about this one? ;) 17:42:21 i just need to write this one up, i'll geto n that today 17:42:32 i keep putting things off thinkign i'll get them on monday or tuesday 17:42:40 and then it's suddenly wednesday again :) 17:42:41 what was the end result? we should just make it default? 17:42:48 yeah, time flies. 17:43:20 relro sure, pie has a few percent overhead on large stuff 17:43:27 Hey, sorry 17:43:31 Distracted by other things 17:43:58 ok, so relro=default, pie=targeted only some packages? 17:44:11 Sounds like some kind of progress 17:44:25 The alternative would be pie on everything but allow packages to disable it 17:44:27 yeah, it has to be very large (iirc OOorg) 17:44:34 before you even see it really 17:44:40 i'll make some graphs and mail them out 17:44:46 mjg59: ah, thats an option to think about too. 17:44:47 And, ironically, the largest ones are probably also the ones that deal with the most untrusted data 17:44:50 (i shoudl really game it.) 17:45:00 ie, we probably *want* OO.o to be pie 17:45:04 yeah 17:45:04 (mm, pie) 17:45:12 #action kylem to write up notes from benchmarking and update ticket. 17:45:21 thanks kylem 17:45:22 So perhaps default and allow per-package optouts, perhaps with fesco approval? 17:45:36 sounds pretty reasonable to me. 17:46:15 do we want to wait for the full info? or just vote on that now? 17:46:32 i'd say wait a week, btu :) 17:46:43 I don't think there's a great hurry... 17:46:51 * nirik will move on in a few then. 17:47:18 the thing we'll find, if we say it's the default, is how many build systems fail to propagate -fPIE 17:47:33 quite sure autotools only recently learned about that 17:47:43 Oh hey now we can't let reality stand in the way of our policies 17:48:01 I'm sure we can write an autoqa test 17:48:28 #topic #298 Revoke Paul Johnsons pacakger access and put him on probation. 17:48:28 .fesco 298 17:48:29 nirik: #298 (Revoke Paul Johnsons pacakger access and put him on probation.) - FESCo - Trac - https://fedorahosted.org/fesco/ticket/298 17:48:39 so, I exchanged a few emails with Paul... 17:49:16 I'm not sure we got too far. He agreed to try and learn workflow better and read docs, and was going to try and get on irc so he could meet with other mono sig folks and get help with workflow. 17:49:39 nirik: I think we've heard this story before 17:49:52 yeah. 17:50:27 why didn't he speak up in the ticket? 17:50:41 also, he is a sponsor still... 17:50:42 how did we get to this point? did he just get grandfathered along? 17:50:49 cwickert: not sure. 17:51:04 kylem: he's been involved for a long time... 17:51:21 proposal: remove (at least for now) his sponsor status. 17:51:24 +1 17:51:27 +1 17:51:31 +1 17:51:34 also proven packager please 17:51:41 +1... 17:51:45 cwickert, we did that last week iirc 17:51:49 +1 17:51:50 +1 to removing sponsor 17:51:52 kylem: ah, ok 17:51:52 cwickert: we agreed to remove that last week, and I did so. 17:52:05 ok, then we all agree here, nice 17:52:07 #agreed remove (at least for now) Pauls sponsor status. 17:52:30 ok, what do we wish to do about his packager status/workflow issues? 17:53:11 what can we do? 17:53:14 do we wish to remove packager? remove commits on packages so maintainers need to add him back in? or something else? 17:53:17 I'd prefer to see someone else uploading his packages 17:53:25 At least for a probationary period 17:53:31 he was up for provenpackager once and was rejected. Was up for provenpackager a second time and was granted. He stepped up when no one else was working on mono so became heavily involved with that stack. 17:53:37 i'm worried about the precedent we set if we drop that, since i'm sure there's a lot of other questionable people who could use a refresher on things 17:53:42 abadger1999: yeah. 17:54:05 But for anything where there are co-maintainers, then I guess we can leave that decision up to them 17:54:11 I'll note for the record he seems like a very nice guy... just doesn't get our workflow or how to work with others too well. 17:54:22 If we've dropped provenpackager then they can decide how much of a problem it is 17:54:41 I think removing packager is too much 17:54:59 i suggest we get his original sponsor towirk withhom 17:55:06 yepp 17:55:36 typing on a cellphone is murder 17:56:03 who is that? can we even determine it? 17:56:52 * nirik looks 17:57:00 if not we should find someone who was willing to sponsor him in monitor is work 17:58:16 FAS says it is Paul Howarth 17:58:18 * mclasen apologizes for missing the meeting so far 17:58:19 yeah. 17:58:25 just got there. 17:58:31 pghmcfc is his sponsor 17:58:45 Paul Howarth 18:00:17 proposal: ask pghmcfc if he could help pfj and mentor him in workflow and working with others 18:00:26 +1 18:00:27 (I have no idea how active pghmcfc is these days) 18:00:32 +1 18:00:46 +1 18:00:51 Paul is active 18:00:54 i'd agree... yanking his packager access seems overkill, and if co-maintainers have issues, they can bring them up 18:00:58 ok +1 18:01:27 ok. 18:01:35 +1. 18:01:39 #agreed ask pghmcfc if he could help pfj and mentor him in workflow and working with others 18:02:00 ok, any further action people would like to proposel with pfj? 18:02:14 None for now 18:02:48 ok, that goes on to the git binary issue related to this. ;) 18:03:25 One thing Oxf13 proposed was that when we rewrite history and remove blobs, we add a tag named with the old hash to koji 18:03:38 that way we know where to find the src.rpm's for those and we keep them. 18:03:50 or we can at least find them again to regenerate them. 18:04:28 so, that sounds good to me, but my outstanding questions are: 18:04:47 1. Is there a way we can identify problem commits to do this to them? 18:05:07 2. who is going to do the fixing/koji tag on the ones identified. ;) 18:06:01 I guess we can ask scmadmins to work on it... unless someone else would like to step up to do so? 18:06:04 is mono the only problematic package or? 18:06:12 no. 18:06:16 oh foo. 18:06:25 the ticket has some more. 18:06:30 yeah, there could be many more 18:06:39 i can work on getting the upload hook to refuse pushes of tarballs or stupidly large files 18:06:49 and I think clamav is affected. At least It has 35MB or so on a clone, which seems way too high. 18:06:51 this was something of a failure on my part not filtering those commits out when I did the conversion 18:07:02 I'd volunteer to do something here, but I've got a tight deadline on my internal project :( 18:07:06 yeah, thats 3. Decide policy for and implement hooks to prevent this. 18:07:26 ajax: I think tibbs|h might have started looking at it, but you guys could coordinate on it. 18:07:31 though it might be a week or two, i'm a little under the gun until beta freeze 18:07:43 * cwickert needs to leave. brb 18:08:06 proposal: file tickets on the 3 outstanding items and get folks working on them. 18:08:11 +1 18:08:16 +1 18:08:26 I guess they could be fesco tickets or infrastructure... 18:08:52 or even FES tickets 18:09:03 +1 18:09:11 yeah, true. 18:09:21 +1 18:09:31 +1 18:09:34 #agreed file tickets on the 3 outstanding items around binaries in git repos and get folks working on them. 18:09:43 #action nirik will file tickets somewhere. 18:09:59 ok, anything else on this? 18:10:03 * nirik will move on in a minute. 18:11:13 #topic #572 NetworkManager-0.9 18:11:13 .fesco 572 18:11:14 nirik: #572 (NetworkManager-0.9) - FESCo - Trac - https://fedorahosted.org/fesco/ticket/572 18:11:17 ok, fun topic time. ;) 18:11:34 * rbergeron tunes in to the channel 18:11:38 * mmaslano will leave in few minutes 18:11:51 this sounded like we got api compatibility for 0.8, yes? 18:12:04 nirik: I'd like to add -1 because last qa comment. They did a lot of work, but still it's late. 18:12:11 (still in a way that needs patching, but at least they're trivial patches) 18:12:13 ajax: yes, but unless we slip it won't be ready. 18:12:35 yes, but testing could take a long time 18:13:07 Every option we have here is bad 18:13:09 * adamw is here for a qa take. 18:13:31 * mclasen tries to lure dcbw over 18:13:32 who makes the 'slip' call? 18:13:32 yeah, a maze of doom. 18:13:39 ajax: Typically, the FPL 18:14:02 And for the record, I'd really prefer not to slip the schedule 18:14:13 i like how you answered in the third person ;) 18:14:19 The other side of this is obviously that going to nm0.9 means that we also have the new network panel for gnome3 18:14:19 jsmith: er, afaik, not really; it's usually made at the go/no-go meeting, with qa/releng present. before your term the fpl didn't always even come to that meeting, and didn't consider it their decision. 18:14:34 jsmith: my reading, anyway. 18:14:40 * bpepple takes a seat in the bleachers. 18:14:53 mclasen: How much are we compromising things by staying with nm0.8 and not shipping the new network panel? 18:15:05 adamw: Yeah, that's fair -- From a "who makes the decision" standpoint, it's the go/no-go meeting 18:15:07 ie, how much does that impact the overall vision 18:15:07 then we're not shipping GNOME 3 18:15:09 we are breaking the gnome-shell experience pretty badly 18:15:18 adamw: From the "who announces it, etc." side, it's the FPL 18:15:25 jsmith: sure. 18:15:27 adamw: I guess I read the other meaning of "call" 18:15:30 =) 18:15:30 and as walters said, basically aren't shipping gnome3 at that point 18:15:40 not sure what to call it then 18:15:58 Ok 18:16:01 so, to make it clear - qa isn't saying 'don't wait for 0.9 + fallback' 18:16:12 but we just want it to be understood that if we wait, the release will almost certainly slip 18:16:13 lets ask dcbw 18:16:17 So we have the choice of shipping with a broken flagship desktop, shipping with a broken KDE or slipping 18:16:27 dcbw: when do you think we can have nm09+compat packages that are testable ? 18:16:32 mjg59: empathy will need to be backported to the nm-0.8 api. 18:16:34 mjg59: I think that's a fair assumption 18:16:49 bpepple: It uses 0.9 now? 18:16:50 in fact, we'd like to make it explicit: if fesco decides to wait for the 0.9+fallback build, we'd like to make a firm decision to slip the release at the same time, and adjust the schedules 18:16:50 mclasen: I will commit to that by Friday morning 18:17:09 mclasen: we'll need to rebuild kdebase-workspace and kde-plasma-networkmanagement with the compat patches 18:17:32 dcbw: but you'll have those patches ready by Friday morning too ? 18:17:33 mjg59: yes. latest stable version that came out uses the 0.9 api. 18:17:37 bpepple: I think empathy will work with just a compile rebuild 18:17:44 We already missed the TC last week, haven't yet been able to do a TC this week, and as adamw stated, this seriously puts us in jeopardy of slipping the beta at least a week 18:17:51 dcbw: That shouldn't be a problem, assuming the patches work. Both packages should be in buildable state. 18:18:12 From the technical standpoint, I really don't care whether we ship 0.8 or 0.9 -- I'll let FESCo folks (who are obviously smarter than me) make that call 18:18:12 Ok, so sticking with 0.8 isn't really an option either 18:18:15 Kevin_Kofler: yeah, I've built them both locally for testing and they seem ok 18:18:16 dcbw: they merged your nm-0.9 patch with the latest stable version that came out yesterday. 18:18:19 Do other apps in general need changes or will they work with the compat layer too? 18:18:20 My concern is in the schedule 18:18:28 Which means the choice is pretty much "Break KDE" or "Slip" 18:18:32 We also need kdebase-runtime rebuilt with the small patch which is already in dist-git master, along with the other client apps. 18:18:38 (See dgilmore's comment about pidgin) 18:18:47 Kevin_Kofler: that talks to nm too? 18:18:51 Or is there any other subtlety I'm missing? 18:19:01 mjg59: yes, and I think "break kde" is unfair and bad. 18:19:04 dcbw: Yes, as a client, see solid-networkstatus. 18:19:07 mjg59: even with the 'break kde' option there's a chance of a slip; we're _already_ behind schedule. so it's more 'break kde and possibly slip' or 'do it all right and definitely slip'. just for info. 18:19:11 abadger1999: dcbw has already prepared 0.9 patchs for most things 18:19:19 nirik: Yeah, it's not KDE's fault that this change wasn't well communicated 18:19:21 But we already have the patch for that one to support 0.9 in upstream master and in dist-git master. 18:19:28 notting: k. And anything that was missed will require similar patches? 18:19:49 * nirik fears he is leaning toward 0.9 + compat patches and if that means slip, so be it. ;( 18:19:55 abadger1999: http://developer.pidgin.im/attachment/ticket/13505/nm09-pidgin.patch for example 18:19:59 Proposal: We ship 0.9 and fix up KDE with the 0.8 compatibility patches. Whether we slip or not is up to QA/releng. 18:20:01 mjg59: tbh, system settings have been around since 2008 too, so it's not like the kde stuff is up-to-date in any way... 18:20:19 http://pkgs.fedoraproject.org/gitweb/?p=kdebase-runtime.git;a=blob;f=kdebase-runtime-4.6.1-nm09.patch;h=29e1c5c6097eef54ed0041d933c33acc72b1f2ce;hb=HEAD is the kdebase-runtime patch, very similar to the Pidgin one. 18:20:24 dcbw: Well, that's a separate discussion 18:20:29 abadger1999: things that use nm-glib should just need rebuilt, i think. things that use dbus directly might need tweaks like the pidgin patch 18:20:39 mjg59: i'm +1 to that 18:21:03 dcbw: what are your weekend plans ? :-) 18:21:31 getting drunk at a firkin fest on Saturday, other than that, working on NM 0.9 compat + KDE? :) 18:21:32 mjg59: +1 18:21:33 To add to what notting said, things that use Solid should need no rebuild at all, only the kdebase-runtime patch (which is similar to the Pidgin patch). 18:21:38 notting: /me adds that info to ticket 18:21:46 if fesco decides to go with the compat patch plan, i will probably propose a special schedule meeting at which we can agree to a slip without waiting for go/no-go. 18:22:07 * mclasen is momentarily disoriented about current looming deadlines 18:22:14 adamw: how soon do you want to do that? 18:22:18 :) 18:22:25 Any other votes? 18:22:40 +1. 18:22:52 i'm in favor of mjg59's proposal, but (as with last week) abstaining due to possible conflict of interest unless we can't get a quorum 18:22:52 adamw: +1 to your schedule meeting proposal 18:23:01 +1 18:23:19 +1, too 18:24:21 so, that looks like it passed then... 18:24:26 mjg59: what exactly are we voting about` 18:24:27 ? 18:24:33 * cwickert just rushed in again 18:24:36 cwickert: "We ship 0.9 and fix up KDE with the 0.8 compatibility patches. Whether we slip or not is up to QA/releng." 18:24:41 18:19 <+mjg59> Proposal: We ship 0.9 and fix up KDE with the 0.8 compatibility patches. Whether we slip or not is up to QA/releng. 18:25:11 this means we make KDE running with 0.9? 18:25:29 cwickert: see the ticket for more details 18:25:33 dcbw has been working on adding nm 08 compat back in 18:25:42 so kde will only need fairly minimal patches 18:25:42 With a heavily-patched NM 0.9, with the NM patches already sorta working and expected to be complete by Friday morning. 18:25:43 nm-0.9 with a compatibility patch applied to our nm. 18:25:56 cwickert: No, it means adding 0.8 compatibility layer back in for KDE + other apps 18:26:20 ok, I missed that part, the last think I read was porting KDE to 0.9 18:26:35 so this is fine with me, as long as KDE is not breaking 18:26:36 jsmith: to be clear, it's only meant for KDE, and would be pulled out when KDE bits are updated to NM 0.9 upstream 18:26:43 it is not a "public" API so to speak 18:27:04 #agreed We ship 0.9 and fix up KDE with the 0.8 compatibility patches. Whether we slip or not is up to QA/releng. 18:27:09 dcbw: How many other apps use the 0.8 api (pidgin, etc.) that are likely to be affected? 18:27:12 ok, anything more on this topic? 18:27:21 +1 (for the record) 18:27:28 I'd like to thank everyone who worked to find a solution here. ;) 18:29:05 thanks for being agreeable here 18:29:12 So, whats the timeline again here? have working bits landed by friday? 18:29:17 * mclasen takes the blame for not having this on the radar for the gnome3 feature page 18:29:17 test over weekend? 18:29:28 yes, packages by friday 18:29:49 that should work for a one week slip 18:29:58 gives us until wednesday to smooth out any problems for a tc compose 18:30:03 er, tuesday 18:30:33 yeah, better communication betweeen NM and it's downstream consumers might be worth persuing down the road... 18:30:37 jsmith: I've got a list, hang on 18:31:05 jsmith: most are patched or have patches in bug trackers already 18:31:25 adamw: Assuming we don't have any other blockers or unforseen issues, of course :-/ 18:31:31 dcbw: Thanks :-) 18:31:35 nirik: agreed 18:31:46 ok, shall we move on then? 18:32:29 jsmith: we're on top of all other known issues, this is the last one blocking tc. 18:32:40 jsmith: anaconda, kde, balsa, empathy, geoclue, krb5-auth-dialog, libsocialweb, liferea, nntpgrab-core, evolution, firefox, pidgin, clawsmail 18:32:44 adamw: Yay! 18:32:53 of those nntpgrab-core does not have a patch 18:33:09 dcbw: And anaconda is using the new 0.9 bits, correct? 18:33:34 jsmith: right now, no, it backed out to 0.8. 18:33:44 jsmith: but only because 0.9 hasn't landed :) 18:33:53 they have a patch that was backed out until nm09 lands 18:33:53 the patch for 0.9 will be put back into anaconda when 0.9 lands. 18:33:58 anaconda was the first thing I sent a patch for 18:34:06 OK. 18:34:28 * jsmith doesn't want to hold up the meeting any longer -- I'll take any other questions offline 18:38:11 ok, anything further for now? Go forth and make it all work! :) 18:38:17 #topic #573 Request to approve deluge bug fix update 18:38:17 .fesco 573 18:38:18 nirik: #573 (Request to approve deluge bug fix update) - FESCo - Trac - https://fedorahosted.org/fesco/ticket/573 18:38:27 zodbot seems lagging 18:38:30 +1 from me for the fix 18:38:30 or me 18:38:31 or both of us. ;) 18:38:35 I'm +1 for this update. 18:38:48 * notting is fine with this; it's a leaf app. 18:38:59 +1 as well 18:38:59 +1 18:39:11 #agreed exception approved. 18:39:15 yeah, it's a packaging split, but *shrug* 18:39:32 RobbieAB: you around today? anything from FES? or should we move on... 18:39:55 I'm semi here... 18:40:10 #topic FES report 18:40:24 Not too much activity this last week... 18:40:38 But I have had other RL related matters on my mind for the past couple of weeks, as I am preparing to relocate to London. 18:41:11 yeah, understandable. 18:41:19 we'll get things moving at some point... 18:41:24 #topic Open Floor 18:41:31 anyone have anything for open floor? 18:41:35 I am really sorry to do this, but I don't see myself having mental energy to spare before the end of April. :( 18:41:39 Need to add Networkmanager here: https://fedoraproject.org/wiki/User:Kevin/DefaultServices 18:42:21 RobbieAB: thats ok. ;) 18:42:32 abadger1999: yeah. I had a note about that, but then forgot about it. 18:42:40 abadger1999: where should that page live long term? 18:43:34 I don't know "Default services policy" "Autostart services policy"? 18:43:51 FESCo policies aren't in their own namespace like Packaging Guidelines. 18:43:56 ok. yeah. ;( 18:44:01 So it just needs to have a proper category. 18:45:10 Services Starting Policy ? 18:45:22 * nirik is having trouble coming up with a good name here. 18:46:28 I'll come up with something... ideas welcome. 18:46:32 anything else for open floor? 18:47:20 nothing here 18:47:32 ok, will close out in a minute then... 18:48:18 thanks for coming everyone! 18:48:22 #endmeeting