Tuesday, 2018-11-20

*** BitEvil is now known as SpeedEvil01:21
*** SpeedEvil is now known as Guest6915710:41
*** frinring_ is now known as frinring12:35
spiiroinreview appreciated: menu/home/back button backlight tweaking12:49
spiiroinhttps://git.merproject.org/mer-core/mce-dev/merge_requests/912:49
spiiroinhttps://git.merproject.org/mer-core/mce/merge_requests/9812:49
Son_Gokuanyone who looks are package manager things, would appreciate this review: https://git.merproject.org/mer-core/libsolv/merge_requests/513:27
r0kk3rzabranson: ^^13:28
r0kk3rzSon_Goku: first things, make a mer bug about it, and reference it in the commit13:29
Son_Gokua bug is required for this?13:30
r0kk3rza bug is required for basically everything13:30
r0kk3rzif you look at the commit history you will see the commit format used13:31
Son_Gokuugh13:32
Son_Gokuso I have to have one-line commit messages13:32
abransonr0kk3rz: we can do that in the merge commit too, but it's nice to have in the actual commit message. good to know what problems it solves too.13:32
abransonSon_Goku: no, it only dictates what the first line says. you can have more lines.13:32
abransonthough the second line should always be blank in git commits in general i think13:33
Son_GokuI'm going to guess JB# is jolla bugzilla?13:33
Son_Gokuso what's the indicator we use for mer bugs13:33
abransonyeah. MER# can be used to refer to a mer bug13:33
abransonall this stuff makes sure the changelogs get generated properly13:33
r0kk3rzhonestly the jollanauts should be better at using mer bugzilla for mer components :P13:34
abransonthis is true ;)13:34
Son_GokuI was trying to figure out how the changelogs were generated actually13:35
Son_GokuI wasn't sure if it was verbatim commit messages like Mageia does, or if it was specifically marked lines in messages (as Fedora is considering), or something else13:35
Son_Gokuabranson: how's this? https://bugs.merproject.org/show_bug.cgi?id=199013:37
merbotMer bug 1990 in libsolv "libsolv should use updated configuration from Fedora" [Task,New]13:37
abransonnot sure of the details, but I think just those first lines are extracted from git. the CI also records it against the referenced bug, and closes it if you say 'Fixes'13:38
Son_Gokuabranson, like so? https://git.merproject.org/mer-core/libsolv/merge_requests/513:39
abransonSon_Goku: like this sorry: [topic] description. Fixes MER#199013:41
Son_Gokupushed that into the commit message13:43
Son_Gokuhttps://git.merproject.org/mer-core/libsolv/merge_requests/5/commits13:43
abransonSon_Goku: sorry could you drop the brackets too. It might work like that, but I've never tried :)13:44
Son_Gokumy inner writer is screaming now :P13:44
Son_Gokuabranson, done13:47
abransonSon_Goku: thanks. that's great. good writeup in the bug too13:53
Son_GokuMer being half openSUSE and half Fedora is weird though :P13:53
Son_GokuI still remember when Moblin was forked from Fedora all those years ago13:54
r0kk3rzseems odd you need so many flags for stuff like that tbh13:55
Son_Gokumls doesn't like changing the defaults, so...13:56
Son_Gokur0kk3rz, I'd much rather most of these options be default, but I'm not upstream and mls doesn't want to, so *shrugs*13:59
r0kk3rzyeah not your fault, just a passing comment13:59
abransonSon_Goku: yeah it is strange. at least there's not so much Debian left over...14:00
Son_Gokuthe Fedora spec actually has a ton of bconds because Fedora and RHEL have two different sets of settings :(14:00
abransonI guess using obs internally brings a lot of suseness14:00
lbtyep14:00
Son_Gokuactually, you can drop almost all of that if you don't use zypper14:00
abransonwell here's the guy who'd know for sure :D14:00
lbt:)14:01
Son_Gokuheh14:01
Son_GokuI'm slowly adapting ol's work against OBS git master14:01
abransonzypper's available but not installed by default. it's nice to be able to use it though.14:01
Son_Gokuso that I can package the next version of OBS for Fedora14:01
lbtSon_Goku: oh14:01
lbtah I see - excellent14:01
lbtwell we're working on that right now14:01
Son_Gokuabranson, you use PK by default, right?14:01
Son_Gokuoh awesome14:01
Son_Gokuthen we should team up :)14:02
lbtI've just got 2.9.5 running and building sb214:02
lbtyes please!!!14:02
Son_GokuI'm also trying to unfuck the horrible packaging of obs at the same time14:02
Son_Gokuit's way messier than it needs to be14:02
Son_Gokuthe only reason it's so messy is because it has to support sle14:02
Son_Gokuwhich is a minefield all by itself14:03
Son_Gokulbt, I'm the maintainer of all the OBS components currently packaged in Fedora, except obs-signd14:03
lbtExcellent14:03
Son_Gokuobs-signd is maintained well by the Fedora COPR team, so I've left that alone14:03
Son_GokuI may request co-maint for that anyway once I get open-build-service done14:03
lbtso do you use sb214:04
Son_Gokuno, mainly because I don't know what it is14:04
Son_Gokuand it looks like no one ever contributed it upstream14:04
lbtexactly14:05
Son_GokuI'm currently planning on writing a dist-git source service and combining it with the mock build engine14:05
lbtso I'm working on pushing it upstream14:05
Son_Gokuawesome14:05
lbtand I thought you may be using it and that would add weight14:05
Son_Gokusb2 stuff is the bulk of ol's fork14:05
lbtyes14:05
Son_GokuI just never knew what exactly it did14:05
lbtit's a cross-compile accelerator14:05
lbthttps://www.merproject.org/~lbt/SB2_internals_1st_ed_20120425.pdf14:06
Son_Goku404 link? :(14:06
lbtwe build the entire SFOS for all devices using it so it's proven pretty capable14:06
lbthttp://www.merproject.org/~lbt/SB2_internals_1st_ed_20120425.pdf14:07
lbthabit to prepend https nowadays14:07
Son_Gokuwell, if someone packages sb2 tooling for Fedora, I have no problem supporting it for the Fedora package14:07
Son_Gokuof open-build-service14:07
lbtmord: FYI ... ^^14:08
lbtmord is also working on getting these patches ported14:08
Son_Gokusweet14:08
lbtI've recently cleaned up the patches a lot14:08
Son_Gokuthat's good14:09
lbtso squashed the set against 2.5 and then refactored them to funcitonal patches14:09
Son_GokuI've been rewriting a lot of ol's patches for the code to run on Fedora systems14:09
lbtstill against 2.5 - but with zero git diff14:09
lbtOK14:09
lbtthen I've begun ported those cleaner patches up to 2.9.514:09
Son_Gokumainly all the changes he made in 2.8 (which was the last version with a working patchset) to work with Fedora's apache user, etc14:10
Son_GokuI've also been talking to the OBS team about finally switching over to the systemd units14:10
Son_Gokubecause screw the suse-bsdrc-compat stuff14:10
lbtI think they're blocked on their enterprise for that14:10
* mord reads back -- excellent14:10
lbtOBS still needs 42.3 and won't run on 15 iirc14:11
Son_Gokuthey now have two SLE releases with systemd, so they're finally willing to do it14:11
Son_Gokuonly SLE 11 didn't have systemd14:11
Son_Gokuso SLE 12 / Leap 42 and SLE / Leap 15 have systemd14:11
Son_Gokuso now I can port forward the work done by Josef Stribny (formerly of RH) to use systemd units instead of sysvinit scripts14:12
lbtwell we've been systemd from the start so that's good for us14:12
Son_Gokuwell, the sysvinit scripts don't work outside of suse distros anyway14:15
Son_Gokunot without the compat module ol wrote14:16
abransonSon_Goku: test build on the libsolv worked fine. we're really close to branching though apparently so I'll merge it after that. a week or so I think. thanks!14:50
Son_Gokubranching?14:50
Son_Gokuabranson, I'm confused, you have a branching model for mer?15:00
abransonfor sailfish15:01
Son_Gokuoh :(15:03
Son_Gokuwell, I had only just noticed you guys managed to get to rpm 4.14 last week15:03
abransonSon_Goku: yes, all that was a bit tricky.15:04
Son_Gokuyes, I tried to do it myself and kept stumbling15:04
Son_Gokuit didn't help that a good deal of rpm was monkeypatched15:04
abransonheh15:04
Son_Gokuabranson, I wanted to ask there was interest in switching from the zypp stack to the dnf one15:04
Son_Gokuas yocto had done, among a few others15:05
abransonthe trickiest part was sb2. it 'accelerates' certain binaries by using the host's native version if the effect should be the same. we had to add some version checking to it to avoid some really impressive crashes.15:06
Son_GokuI think it was a year ago that I asked about this, and there was some interest in order to drop things like boost in the dep change15:06
Son_Gokus/change/chain/15:06
abransonSon_Goku: yeah I remember. there were some questions about delta rpm support but I think that's ok now. the biggest hurdle is that our usage of zypp is quite heavily customised, so we couldn't just switch over.15:07
Son_Gokuwell, that's something I wanted to ask about15:08
Son_Gokuwhat is customized about your usage of zypp?15:08
abransonthat starts to get a bit outside of my knowledge, but I think it mostly concerns managing interactions with the online store? we have a component called ssu which manages version repos and auth.15:09
abransonlbt probably knows more about that part15:09
Son_Gokulbt, what exactly is custom about your zypp usage?15:13
lbtI think the ssu plugins we have go into libzypp15:14
Son_GokuI couldn't find any sources of plugins using libzypp15:15
lbtyeah - ssu uses an authentication system there15:15
lbthttps://git.merproject.org/mer-core/ssu/15:16
Son_Gokuso libdnf supports plugins as well15:17
Son_Gokuand there's a couple of plugins in the main source tree already15:17
Son_Gokuhttps://github.com/rpm-software-management/libdnf/tree/master/plugins15:17
Son_Gokuthis one is probably relevant in your case: https://github.com/rpm-software-management/libdnf/blob/master/plugins/rhsm.cpp15:17
lbtthere's a cost/benefit to think of too15:18
Son_Gokulbt, an upcoming feature in dnf is delta repodata15:19
Son_Gokuthrough zchunk15:19
Son_GokuI figure that would actually be legitimately useful for mobile devices15:19
Son_Gokulbt, it's something we've been developing in Fedora for a while now15:22
Son_Gokuit would make refreshing metadata so much cheaper15:22
lbtwe are at a point where we can review how we're handing packages and the surroundings15:23
Son_Goku... ?15:26
Son_GokuI'm not sure what that means15:26
lbtinternally I mean15:26
lbtin terms of looking at what we can and should do in taking the core forwards15:27
Son_Gokulbt, well, is there anything in particular you're specifically examining?15:31
lbtright now. No :)15:31
lbtcurrent focus is on OBS and getting that upstreamed15:31
lbtthen we have a lot of automation stuff which needs bringing up to date15:31
Son_Gokuokay15:31
lbtthis doesn't stop experimentation from the community15:32
lbtand that kickstart often helps make a decision15:32
Son_Gokuokay15:32
lbtI think the last toolchain upgrade started that way15:32
Son_Gokuinteresting15:32
Son_Gokuanyway, the libsolv thing will probably be desirable anyway15:33
Son_Gokusince among other things, it will make solver access to rpmdb more reliable15:33
lbtIt is, yes15:33
Son_Gokuand it'll make debugsolver data actually work :)15:33
lbt:)15:33
Son_Goku(that was a bit frustrating to find out :P)15:33
lbtI think plugging ssu into a dnf system may be of interest but I'm not familiar enough to know how15:34
Son_GokuI could probably help on the dnf side15:34
Son_GokuI just don't know how ssu works15:34
lbtdoes dnf do the downloading?15:34
Son_Gokulibdnf does through librepo, yes15:35
Son_Gokulibrepo is a specialized wrapper around libcurl15:35
lbthawkey specifically doesn't15:35
Son_Gokuhawkey is dead15:35
Son_Gokulibdnf replaced it15:35
lbtthe docs crosstalk15:35
lbtI saw that hawkey was embedded in dnf15:35
Son_Gokuthe docs really need updating :/15:35
lbtabsorbed may be a better term then?15:36
Son_Gokubut yes, there's a repo class now: https://github.com/rpm-software-management/libdnf/tree/master/libdnf/repo15:36
Son_Gokulbt, yep15:36
r0kk3rzwhat would be the benefit of using libdnf?15:38
lbtr0kk3rz: that's kinda my point15:38
*** BitEvil is now known as SpeedEvil15:48
ol<freenode_Son "I'm slowly adapting ol's work ag"> Can I take a look please?17:09
ol<freenode_Son "mainly all the changes he made i"> I have working 2.9.5 already: https://obs.infoserver.lv/project/monitor/obs:server:fedora:2.917:16
*** frinring_ is now known as frinring21:02
*** jrayhawk_ is now known as jrayhawk21:40

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!