IRC Logs for #crux-devel Saturday, 2019-05-04

*** stenur has quit IRC00:34
ryuofun morning.03:45
ryuohttps://bugzilla.mozilla.org/show_bug.cgi?id=154897303:45
Romsteroops05:55
frinnstlol08:28
*** onodera has joined #crux-devel10:47
frinnstshould we drop chromium completely? The version in our repos is very out of date13:19
frinnstand nobody seems to want to maintain it13:19
juefrinnst: yep13:41
pedjasomeone mentioned on the ml that consolekit can't be installed on 3.5, so kde and mate are broken13:51
frinnstI use my old 0.4.6 port of consolekit. works fine13:51
TimB_pedja: builds fine for me13:51
frinnstConsoleKit is currently not actively maintained. The focus has shifted to the built-in seat/user/session management of Software/systemd called systemd-logind!13:52
frinnstjoy13:52
pedjaidk, the mail isn't really clear13:52
pedjaconsolekit2 is a bit more actively maintained :)13:53
pedjaby xfce devs, iirc13:53
pedjathere is also elogind, but that's a gnome dep, afaik13:55
juethat was about polkit/js -> https://lists.crux.nu/pipermail/crux/2019-April/006194.html14:00
juenot consolekit14:00
pedjaah. sorry for the noise, then14:12
juebut we should either fix opt/polkit and opt/spidermonkey or remove them, both are very outdated14:29
jueI don't have a use for polkit, but I've added a polkit port to my private repo for testing14:30
juealso an mozjs52 port14:30
TimB_pedja, jue: there should be an updated version in jaegers mate repo, which work fine over here :)14:31
juethe mozjs52 port uses the firefox esr sources, seems to work fine14:31
juewould be great if someone could test the ports ;)14:32
pedjaI used 52.2.1 for spidermonkey/js14:33
jue52.9.0 is the latest esr version14:34
pedjagot random segfaults with some other versions14:34
pedjahave to check what lfs is using, I used that the last time I fiddled with it14:34
jueTimB_: I saw jaeger's polkit, but don't want to install icu on my box14:35
TimB_jue: just out of curiosity, why not install icu?14:35
juepedja: could you try the version in my repo, please?14:35
pedjawhat's the link again?14:36
pedjayou'll have to forgive me, I am old and forget stuff :)14:36
jueicu is _very_ intrusive, a lot of ports will link against it14:36
juepedja: https://crux.nu/portdb/index.php?a=repo&q=jue14:37
TimB_jue: I see. Never saw that14:37
TimB_I am trying mozjs60 and polkit from your repo jue14:38
TimB_breaking polkit isn14:38
TimB_'t as bad as I remember it, I'm sure :P14:38
TimB_jue: sidequestion about ck4up, have you ever used it with gitlab hosted projects?14:39
pedjajue, I use the same consolekit version, just with polkit enabled :)14:39
jueTimB_: polkit version is still 0.115 because 0.116 is broken for me14:40
TimB_0.116 is not a release last time I checked14:40
TimB_just a beta or whatever14:40
TimB_I think they prepare some bigger changes iirc?14:40
juedunno, but i get an build error with polkit 0.116 and mozjs6014:42
pedjajue, something like https://gitlab.freedesktop.org/polkit/polkit/issues/7014:44
jueno, it's different14:46
TimB_0.115 doesn't build with mozjs-60 as well..14:47
jueTimB_: as I said, that won't work ;)14:48
jueuse mozjs5214:49
TimB_jue: o:) I always want to take a peek myself. currently compiling your mozjs5214:49
pedjaheh. that ff bug just hit14:51
pedjaif only I hadn't shutdown ff so cuda build won't go out of memory...14:53
juepedja: that's the linker error I get with polkit 0.116 ->   CXXLD    polkitd14:53
jue./.libs/libpolkit-backend-1.a(libpolkit_backend_1_la-polkitbackendjsauthority.o): In function `mozilla::Vector<JS::Value, 8ul, js::TempAllocPolicy>::growStorageBy(unsigned long)':14:53
juepolkitbackendjsauthority.cpp:(.text._ZN7mozilla6VectorIN2JS5ValueELm8EN2js15TempAllocPolicyEE13growStorageByEm[_ZN7mozilla6VectorIN2JS5ValueELm8EN2js15TempAllocPolicyEE13growStorageByEm]+0x8a): undefined reference to `moz_arena_malloc'14:53
juepolkitbackendjsauthority.cpp:(.text._ZN7mozilla6VectorIN2JS5ValueELm8EN2js15TempAllocPolicyEE13growStorageByEm[_ZN7mozilla6VectorIN2JS5ValueELm8EN2js15TempAllocPolicyEE13growStorageByEm]+0x111): undefined reference to `moz_arena_malloc'14:53
juef***, sorry14:53
juehttps://750f71d5ddfc0961.paste.se/14:53
pedjaI guess they didn't update it for js60 far enough :)14:54
jueyeah, looks like14:54
pedjaor they test against some magical version of it, with no mention of the commit14:55
pedjamozjs-60 is in the release notes14:56
pedjaand no actual release tarballs at gitlab. well, great14:57
TimB_jue: mozjs52 and polkit work fine here on 3.5, but I changed sysconfdir=/etc, authfw=pam and defined the polkit user/group15:02
TimB_http://sprunge.us/zm9zEC pam.d/polkit-1 and the modified Pkgfile: http://sprunge.us/WAKPbp15:03
pedjasimilar to mine, I just disable elogind too (can't remember why, tbh)15:07
juewhat do you mean with 'defined polkit user/group' ?15:07
TimB_not sure why but I have a user for polkit.. polkitd, it's --with-polkit-{user,group}=..15:08
jueah, I see15:09
jueok, changed the user to polkitd but keep sysconfdir and authfw for 3.415:14
TimB_ok :)15:15
TimB_jue: sorry, I have to take that back. polkit doesn't work for me17:11
TimB_(polkit-mate-authentication-agent-1:582): polkit-mate-1-WARNING **: 16:59:31.504: Error getting authority: Error initializing authority: Error calling StartServiceByName for org.freedesktop.PolicyKit1: GDBus.Error:org.freedesktop.DBus.Error.Spawn.ChildSignaled: Process17:11
TimB_.. org.freedesktop.PolicyKit1 received signal 11 - sorry17:12
jueoops ...17:16
jueTimB_: thanks, so we have to say that building from the firefox sources doesn't work17:18
TimB_well, I still could test it with --authfw=shadow, but I have my version with pam working so I don't suspect that this will make a difference17:19
ryuofrinnst, jaeger, jue: currently pkgadd requires a flag to perform an upgrade operation vs an install with its argument. strictly speaking this is not necessary and seems redundant even. i've been considering eliminating this flag entirely or making it an ignored flag and just have pkgadd determine whether to do an install or upgrade operation based on whether the package is already installed or not. do17:26
ryuoyou know if this would be an issue?17:26
pedjahow do you force reinstall then?17:32
pedjapkgadd should do what its told, not try to determine anything, imho :)17:34
ryuopedja: well... it'd probably just overwrite the package.17:41
ryuopedja: i was considering changes to make 'batch' mode more practical.17:42
*** xor29ah has quit IRC19:23
*** xor29ah has joined #crux-devel19:24
pedjaryuo, batch mode?19:29
pedjajue, maybe use this for js60? https://ftp.acc.umu.se/pub/gnome/teams/releng/tarballs-needing-help/mozjs/mozjs-60.1.0.tar.bz219:34
nomiusSo pam will be included in the next release?20:27
frinnstyes20:27
nomiusAnd also polkit I see20:28
*** onodera has quit IRC22:06
ryuopedja: install or upgrade multiple packages in one run.22:49
ryuopedja: an optimization so you don't need to run it separately for each one.22:49
ryuoif you have multiple to do.22:49
ryuofork+exec isn't cheap.22:49
pedjadoesn't prt-get take care of that?22:51
pedjaah. that optimisation will still apply, since prt-get uses pkgadd and friends behind the scenes, afaik22:53
frinnstryuo: "upgrading" a package that's not installed would be annoying23:18
ryuoyes.23:37
ryuofrinnst: in which case it would fallback to a regular install.23:37
ryuofrinnst: that can be determined by whether it already exists in the database file.23:37
ryuoi just observed that the flag appears to serve no functional purpose. it's something that can be determined by the program.23:38
ryuobut if it must be kept, i'll just abandon my batch mode idea since it's pretty much a requirement for it to work.23:38
ryuothe fork+exec overhead is real, but not the biggest bottleneck in pkgadd runs.23:39

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