17:00:38 #startmeeting RELENG (2018-07-26) 17:00:38 Meeting started Thu Jul 26 17:00:38 2018 UTC. 17:00:38 This meeting is logged and archived in a public location. 17:00:38 The chair is mboddu. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:38 Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:00:38 The meeting name has been set to 'releng_(2018-07-26)' 17:00:38 #meetingname releng 17:00:38 The meeting name has been set to 'releng' 17:00:38 #chair nirik tyll sharkcz masta pbrobinson pingou puiterwijk maxamillion mboddu Kellin dustymabe 17:00:38 #topic init process 17:00:38 Current chairs: Kellin dustymabe masta maxamillion mboddu nirik pbrobinson pingou puiterwijk sharkcz tyll 17:00:52 good morning everyone. 17:01:06 hello 17:01:15 nirik: Morning, I guess you have some important stuff to discuss 17:01:24 * mboddu might have some questions to nirik 17:01:30 * nirik has 2 items for whenever: the rawhide numbering ticket/thing and fedora-release/fedora-repos moving to all src.fp.o. 17:01:42 Yup 17:01:49 puiterwijk: Hello 17:02:02 Welcome back!!! :) 17:02:12 Hah. I've never l;eft 17:02:56 puiterwijk: Hehe :) 17:03:04 Okay, lets get started 17:03:17 * masta is here 17:03:18 hello 17:03:26 Hi masta 17:03:30 #topic #7445 [RFE][PATCH] make $releasever return "rawhide" on Rawhide 17:03:36 #link https://pagure.io/releng/issue/7445 17:04:04 nirik: So, you wanted to go with option 2? 17:04:04 That would mean packages like kernel-4.10.0-1.frawhide? 17:04:17 puiterwijk: nope. 17:04:19 mboddu: nope. 17:04:27 kparal talked me into option 1 17:04:35 okay. 17:04:51 it doesn't change dist, except for dnf (and before we do it, packagekit, etc) 17:04:52 nirik: Okay, thats good 17:05:04 and we can drop fedora-repos-rawhide 17:05:18 ah, okay. 17:05:26 but it does mean people can screw up things easier if they 'dnf --releasever=rawhide install whatever 17:06:28 hum.. 17:06:41 is everyone ok advocating for that plan? or any concerns? 17:07:30 nirik: Instead of "dnf --releasever=rawhide install whatever" they have to use the actual release number like "dnf --releasever=30 install whatever" is that all, right? 17:07:43 right now they would have to: 17:07:50 i like the fedora-repos-rawhide, and I do 'dnf install --disablerepos=* --enablerepos=rawhide foo-package'... so that would change to a --releasever=rawhide, or is that not right? 17:08:02 dnf install fedora-repos-rawhide; dnf install --releasever 29 whatever' 17:08:33 masta: yes, you could just specify rawhide and wouldn't have to disable anything. 17:09:15 also would fix a long standing issue around moving to rawhide from a stable relase and gpg key... 17:09:35 because the gpg key file is called 'whatever-rawhide' not 'whatever-29' 17:09:38 ok so we are removing the label of rawhide, and proposing to use the number? (just trying to be clear) 17:11:05 no, we are using 'rawhide' the label instead of the number. 17:11:30 right now I have to know that '29' is rawhide. After this I don't need to care... 'rawhide' will always point to rawhide. 17:11:51 that sounds great! 17:11:55 * kparal is here, joins late 17:12:10 this was orig a request from qa, but I think it makes everyone's life easier. 17:12:21 nirik: you can do --releasever=rawhide on stable release right now and it works as expected 17:12:36 ok so in a way this disambiguates rawhide from branched, if I parse this correctly. 17:12:37 kparal: if you have the default/metalink yeah 17:13:42 nirik: regarding gpg key name, I haven't particular checked this case. it's possible, but not tested 17:14:17 all that changes is how dnf sends network requests when $releasever is in URL 17:14:29 currently, $releasever gets replaced by 29 on rawhide 17:14:40 it would be replaced by 'rawhide' with the change 17:14:47 nothing else changes 17:15:03 This would also help with stats actually... how many people really are on rawhide 17:15:05 that means that rawhide repos don't need to have 'rawhide' hardcoded in the url and can use $releasever 17:15:15 that applies to fedora repos but more importantly to 3rd party repos 17:15:54 e.g. copr now serves also two different repo files, based on whether it's rawhide repo or not 17:16:01 you could also serve just one 17:16:03 this is option #1, right, just to be clear? 17:16:07 masta: yes 17:16:12 * masta is sold 17:16:22 +1 option #1 17:16:45 the downside is that PackageKit needs to get fixed. currently it doesn't use the same approach for resolving $releasever as dnf 17:16:54 there might be other tools that need a change as well 17:17:02 I haven't talked to PK devs yet 17:17:06 yeah, possibly dnfdragora, ostree... 17:17:14 wanted to first see the opinion of releng 17:17:25 I'm sure a handfull of things would struggle short term, but we can agree to move forward with the "intention to do this" 17:17:27 kparal: I'd suggest we make a tracking bug for this, then file bugs against the package managers blocking that. 17:17:41 ok 17:18:06 I am +1 to the change 17:18:31 I hope I haven't overlooked something important and you don't come all yelling later :) 17:18:43 but shouldn't be difficult to revert, in the worst case 17:18:54 right, it's reversible 17:19:08 and I'll need to do more research with more tools first anyway 17:20:11 kparal: Yeah, if all the package management tools are updated that would solve 90% of the use cases 17:20:44 I still think it's a good idea to support both releasever=rawhide and releasever=NN in mirror manager 17:21:13 sure. 17:21:48 nirik: That is just a redirect on MM, but we have to remember to change it everytime when we branch, am I right? 17:22:11 yep 17:22:45 nirik: Okay 17:24:42 So, the course of action, email on devel list, talk to PK and other package manager maintainers and then ask them to do the necessary changes and then we change it on fedora-release 17:24:47 Am I missing anything? 17:25:06 create the tracking bug 17:25:15 might be some other tools, but hopefully they chime up 17:25:26 kparal: Yup, thanks :) 17:25:31 * cverna is around 17:26:14 I'll do the tracking bug and talk to PK devs. do you want to post the proposed change to devel list on your behalf, or should I? 17:26:40 either way. ;) Might take me a bit to get to it... 17:26:52 so if you could that would be great. 17:26:58 ok, I'll handle it 17:28:03 #info kparal is going to create the tracking bug and then email to devel list about the proposal (option #1) on https://pagure.io/releng/issue/7445 adn then talk to PK devs. Once all the necessary changes are in place, we will update fedora-release package. 17:28:08 And hopefully nothing breaks 17:28:11 thanks kparal! 17:28:16 kparal++ 17:28:18 thank you 17:28:22 thanks kparal 17:28:44 The cookie jar is currently broken, should be fixed later today. 17:29:46 nirik: Oh, no cookies? :( 17:29:53 Anyway, moving on 17:30:00 #topic #7612 moving fedora-release entirely to src.fp.o 17:30:07 #link https://pagure.io/releng/issue/7612 17:30:43 pbrobinson offered to help on this and last time we talked, he is expecting to get the PR ready pretty soon 17:30:54 ah right, I was going to ask where this was. 17:30:54 nirik: Any updates you want to add? 17:31:02 also, is fedora-repos going to be moved too? 17:31:06 nirik: Yup 17:31:07 it's in the same boat pretty much 17:31:07 yes, I'm also doone with it, hoping to do a PR tonight or tomorrow 17:31:21 yes 17:31:24 we need to also re-enable PR's on src.fedoraproject.org 17:31:28 pbrobinson: Thanks again :) 17:31:37 and... do we want to keep the 'must use signed off' thing? 17:31:39 I plan to PRs for both release/repos 17:31:50 I think it makes sense 17:31:58 nirik: +1 to 'must use signed off' thing 17:32:10 I find it pointless and kinda a barrier, but ok. 17:32:27 although that's presuming the functionality isn't turned off on src.fp.o (I've not looked) 17:32:50 its not. they are currently set that way for commits. 17:33:05 which breaks a lot of the automated fixing people have been doing of late. 17:33:41 well TBH I'm not sure we want automated fixes in either of those packages 17:34:08 why not? 17:34:22 one of them was removing the "%defattr(-,root,root,-)" 17:34:46 because most of the automated fixes would probably break a bunch of the special petal bits that those packages explicitly provide 17:35:18 I'd say go with sign-off in the commit 17:35:21 I can't see how, but ok. If you all want to keep that we can. I just will continue to think it's silly. ;) 17:35:22 the defattr is likely different but I've already spent a bunch of time fixing FTBFS from half done auto fixes 17:36:36 anyhow, let me know if I can help any moving things... and thanks for doing that. 17:36:38 I'm kind of the opinion if they can't work out how to do signed off bys they're probably going to get the enabling of defaults wrong and they'd be better off filing a bug requesting someone else do the change 17:37:20 that is my sentiment too 17:38:52 pbrobinson, nirik : Anything else? 17:39:42 nope. I'd be happy to discuss why its dumb more over a beer, but I am happy to move on now. 17:40:09 not from me 17:40:15 nirik: Hehe, we got FLOCK for that :P 17:40:21 pbrobinson: Thanks for working on this 17:40:32 Cookie jar is broken, but still pbrobinson++ 17:41:00 \o/ I'll drink to that :-P 17:41:59 #info pbrobinson is working on this and going to make PR's for both fedora-release and fedora-repos packages. 17:42:26 Okay, next topic 17:42:40 #topic Alternate Architectures Update 17:42:49 sharkcz: Are you around and any updates? 17:43:05 pbrobinson: since you are around, do you want to chime in ;) 17:44:19 The only outstanding thing I can think of is that armv7 f28 thing... which is a real weird one... 17:44:56 otherwise I think everying is going along fine 17:45:13 nothing really here 17:45:26 nirik: what's the f28 v7 thing? 17:46:03 also I noticed the appliance-creator build failed, looked like the package wasn't in the build root but I never got any further to see due to meetings 17:46:10 pbrobinson: https://bugzilla.redhat.com/show_bug.cgi?id=1576593 17:46:32 basically f28 armv7 vm's 'pause' or bus error. So ours are back on f27 for now. 17:46:47 I think the real issue is in binutils or something. 17:47:14 appliance-creator failed because a python2 package it needs got dropped. adamw was sorting it out. 17:47:29 well, dcantrel is trying to rebuild it now 17:47:30 but it is failing. 17:47:31 nirik: ah, yes I know that one :-/ 17:47:47 so far, we suspect the new binutils that showed up between the build that disabled py2 (yesterday) and now 17:47:51 we need a fix to build the fix for the fix 17:48:40 i need a drink 17:48:56 .whereisitfiveoclock 17:49:01 adamw: Can I join? 17:49:02 what, that's not implemented? shocking. 17:49:20 * mboddu also needs a drink 17:49:23 adamw: I have Gin and martini glasses in the freezer, dirty martini in about 45 mins when I'm back from the supermarket ;-) 17:49:33 at this point i'd accept a clean one 17:49:49 so, yeah, anyhow, we're working on that. then maybe we'll get a compose? perhaps? who knows? 17:49:54 you can have what ever one you want... 17:50:58 adamw: I dont know whether you get "la fin du monde" in your area, but I tried it in Quebec city and it was amazing 17:51:05 Its a beer 17:51:10 mboddu: i've got a bottle of it in the fridge right now. 17:51:27 adamw: Okay, you are all set then 17:51:37 * mboddu got 30 bottles of those when coming back :D 17:51:49 .whereisitfiveoclock 17:51:49 nirik: [It's always 5 o clock _somewhere_. CHEERS! 🍻🍻] 17:52:08 nirik: Haha :D 17:53:00 hehe 17:53:00 #info Other than https://bugzilla.redhat.com/show_bug.cgi?id=1576593 there is nothing much to report on alt arches 17:53:13 nirik: =) 17:53:22 #info And everyone wants to have a drink because of all the issues lately 17:53:29 nirik: (though i was figuring it'd be literally the city where it turned 5 o' clock most recently) 17:53:49 oh thats harder. 17:54:08 * adamw cracks whip 17:54:21 London or somewhere in the Atlantic atm ;-) 17:56:54 #topic Open Floor 17:56:58 Anything else guys? 17:57:18 I will be on PTO from next Wed till FLOCK 17:57:51 sightseeing? 17:58:10 nirik: Yup 17:58:17 cool 17:58:17 Berlin and then Wroclaw 17:58:43 And will meet cverna Tue before FLOCK and will come to Dresden together 17:58:52 (hopefully) 17:59:00 nice 17:59:07 mboddu: a news about your train tickets :P ? 17:59:11 * mboddu haven't received my physical train ticket in mail yet :( 17:59:12 any* 17:59:18 cverna: Nope :'( 17:59:32 that doesn't sound good 18:00:02 * cverna is not sure where he actually put his physical ticket :S 18:00:16 cverna: I emailed them, no response yet, probably should email them in German 18:00:29 * mboddu dont know how good is google translate from English to German 18:00:50 mboddu: German are usually good in English not like French :P 18:01:25 Where is pingou? 18:01:32 :D 18:01:43 Anyway, its past the scheduled time 18:01:48 Thank you all for joining 18:01:51 #endmeeting