IRC logs for #baserock for Monday, 2016-01-18

*** gtristan has quit IRC03:47
*** gtristan has joined #baserock04:16
*** toscalix has joined #baserock08:16
*** rdale has joined #baserock08:45
*** edcragg has joined #baserock09:03
*** bruce_ has joined #baserock09:05
*** bashrc_ has joined #baserock09:07
*** jonathanmaw has joined #baserock09:36
*** CTtpollard has joined #baserock09:42
radiofreehow attached are we to the qt documentation? rdale?10:12
radiofreeit's not a major issue, but means you can just do "build-system: qmake" with no chunk if you don't build it10:13
rdalewrt to the patches in baserock for installing it? i originally wanted to have a complete qt dev environment with qt creator and enlightenment desktop, but that doesn't seem relevant any more10:14
paulsherwoodrdale: why not?10:14
rdaleperhaps it is then10:14
paulsherwoodand also, any news on your 'frenzy'?10:14
paulsherwoodrdale, gtristan - would it make sense to add qt to the gnome system?10:15
rdaleno frenzy, but i've worked out how i'm going to do it10:15
* radiofree will leave the docs then10:16
paulsherwoodrdale: but have you worked out *when*? :-)10:16
radiofreefor qtbase it adds about 5 minutes to the build, other things it's pretty tiny10:16
rdaleif you are going to use a baserock system as a qt dev environment, then i think you want to docs installed10:16
rdalepaulsherwood: today or tomorrow with luck10:17
paulsherwoodcool!10:17
rdaleradiofree: i don't think 5 mins in the context of a qt build is much, as long as the doc patches aren't hard to maintain10:18
radiofreei don't think they'll require any maintenance (they certainly haven't in the 2... 3 times i've upgraded qt)10:19
rdaleok, then i would vote for leaving them then10:19
rdaleleaving them in, i mean10:20
radiofreemotion carried10:20
paulsherwood:)10:21
*** gtristan has quit IRC10:21
*** Lachlan1975 has joined #baserock10:40
*** gtristan has joined #baserock10:51
*** gtristan has quit IRC10:52
*** gtristan has joined #baserock10:53
gtristanpaulsherwood, I dont think it's relevant to the GNOME system really, there is nothing "gnome" which wants qt10:54
paulsherwoodgtristan: yes fair enough. but the gnome system is our closest to a general user/dev machine, and some user/devs need qt...10:57
paulsherwoodwhich means revisiting the concept of how to add 'apps' to systems i think10:58
*** edcragg has quit IRC11:35
*** edcragg has joined #baserock11:37
radiofreethe instances thing for ybd, is this just something i put in a config file?13:52
radiofreeinstances = 2 or something will automatically spawn two builds?13:52
pedroalvarezyes13:53
pedroalvarezIs something that you put in a config file13:53
paulsherwoodradiofree: echo "instances: 2" > /src/ybd/ybd.conf (for example)13:54
radiofreeok thanks13:55
radiofreethough probably be best to use ">>" since i don't want to wipe my current config13:55
paulsherwoodybd will check /src/ybd/ybd.conf after /src/ybd/ybd.conf (which has the defaults) and definitions/ybd.conf13:55
paulsherwoodradiofree: ack13:55
radiofreeoh i didn't know that13:55
radiofreewhen does ybd/config/ybd.conf (in the source tree) get checked?13:56
paulsherwoodsorry i should have said... "ybd will check /src/ybd/ybd.conf after /src/ybd/config/ybd.conf"13:56
paulsherwoodso you can keep your own config separate, to avoid any git merge clashes on /src/ybd/config/ybd.conf13:57
paulsherwoodhttps://github.com/devcurmudgeon/ybd/blob/master/ybd/app.py#L10913:58
radiofreeah13:58
radiofreeok i'll do that13:59
radiofreeprinting the time the build started might be useful14:02
paulsherwoodit does...?14:04
radiofreei only get the time since the build started14:05
radiofree"16-01-18 00:01:17 [1/20/310] [qtwebkit] Running command:"14:05
paulsherwood15-11-07 00:00:03 [1/800/800] [stage1-binutils] Starting build of stage1-binutils.dac85c48b260aa710b7910ab871bf5b5cce1e27910bf090b4b60bf2dc35fbbbb15-11-07 00:00:03 [1/800/800] [stage1-binutils] Starting build of stage1-binutils.dac85c48b260aa710b7910ab871bf5b5cce1e27910bf090b4b60bf2dc35fbbbb15-11-07 00:00:03 [1/800/800] [stage1-binutils] Starting build of stage1-binutils.dac85c48b260aa710b7910ab871b14:05
paulsherwoodf5b5cce1e27910bf090b4b60bf2dc35fbbbb[A14:06
paulsherwoodoops sorry14:06
paulsherwoodradiofree: do you mean actual wallclock time??14:07
radiofreeit's useful to figure out how long things have taken, but kinda useless to gauge how long something has been building14:07
radiofreeunless i make a manual note of what time something started building14:08
paulsherwoodradiofree: you could set log-elapsed: False14:09
radiofreewhat does that do? prints the actual time?14:09
radiofreeah, looks like i want that14:10
paulsherwoodyup14:12
*** gtristan has quit IRC14:59
*** bruce_ has quit IRC15:09
*** gtristan has joined #baserock15:18
paulsherwoodso... i'm wondering about richard_maw's shallow conclusions15:28
paulsherwooder, deep conclusions about git-shallow15:28
paulsherwoodmaybe we should just bite the bullet and find a way to get the whole of gcc history onto gbo?15:28
jmacsWhat's stopping you doing that, ooi?15:30
richard_mawthe problems are speed and space, they are related but not the same15:30
paulsherwoodiirc lorry timed out when we tried initially15:30
richard_mawaye, we could potentially introduce additional rules letting an initial import have a wider period15:31
* richard_maw recalls taking >8h when he fetched it at home, but that was on some shaky wifi15:31
paulsherwoodiiuc downstream users don't *have* to mirror all of gbo15:31
paulsherwoodbut i wonder if we did this would we end up accidentally adding n-gigabytes downloads for all users?15:32
richard_mawmost of gcc's history should compress well, but there's a couple of git behaviours that make it painful15:33
richard_maw1.  The server usually generates packs on-demand, rather than sending the ones the content is stored in.15:34
richard_maw2.  GNU style changelog files are the worst-case input for git's packing algorithms15:34
richard_mawso it ends up taking a long time to generate a very large pack to send15:35
jmacsIt doesn't take that long for me to clone gcc from github. Maybe 30 minutes.15:35
richard_mawgithub's non-standard git implementation is a lot better than whatever's on savannah15:35
jmacsDo you lorry from another git repository, or straight from gnu's svn?15:35
* richard_maw can't remember the details15:36
richard_mawI think back then we used the git mirror on savannah15:36
richard_mawand github is very good at its packing15:37
paulsherwoodmaybe the answer is just that, then?15:38
richard_mawit's worth a try, but I no longer have time to try it15:40
paulsherwoodack15:42
paulsherwoodis the process just to submit a lorry file to gerrit?15:42
paulsherwoodand if it were accepted, are we confident that this on its own does not add ngigabytes of download for users and downstream troves?15:43
richard_mawfor imports we have reason to suppose would work, normally yes, I'd rather time a local import first though, since if it then fails we would have downstream troves creating an empty repository that we don't fill15:44
richard_mawunfortunately I think it will still cause big downstream updates even if it now imports successfully, since there's a couple of extra hops between upstream and downstream that could require a repack15:45
paulsherwoodeven if our default systems don't use it?15:46
richard_mawyes15:46
richard_mawthe lorry configuration still expects that it should mirror everything15:46
richard_mawdownstreams by default get everything in delta/15:46
richard_mawgitano lets you hide repositories by setting project.archived=true, which would stop downstreams from getting them, but they wouldn't show up on the cgit UI or gitano ls by default15:47
radiofreeok this is a new one15:48
radiofreeOSError: [Errno 22] Invalid argument: 'install-files/essential-files/.//etc/os-release'15:48
radiofreewhen trying to upgrade a system15:48
* radiofree just gets rid of "install-essential-files" for now15:51
radiofreethe essential-files don't seem that essential15:52
*** franred has quit IRC15:58
*** bruce_ has joined #baserock16:02
radiofreeok, another new one16:03
radiofreeduring the call to system-verison-manager16:03
radiofreeUpdating system configuration16:03
radiofreecmp: /tmp/tmp.B74HqrMEjZ/systems/weston-devel/run/etc/xdg/systemd/user: Is a directory16:03
radiofreeERROR: xdg/systemd/user: replacing a file with a symlink is only supported if there are no changes to the content.16:03
radiofreei'm trying to upgrade a weston-system16:03
radiofreedoing this upgrade from a mustang works ._o16:10
paulsherwoododd16:10
paulsherwoodrichard_maw: i'd accept the hidden from downstreams by default, as a first step?16:12
*** franred has joined #baserock16:13
*** franred has quit IRC16:38
*** edcragg has quit IRC16:48
*** franred has joined #baserock16:49
*** rdale has quit IRC16:52
*** jonathanmaw has quit IRC17:03
*** bruce_ has quit IRC17:10
*** rdale has joined #baserock17:14
*** rdale has quit IRC17:19
*** rdale has joined #baserock17:24
*** rdale has quit IRC17:29
*** rdale has joined #baserock17:32
*** bashrc_ has quit IRC18:03
* paulsherwood wonders what others think of that18:07
*** rdale has quit IRC19:30
*** Lachlan1975 has quit IRC19:35
*** rdale has joined #baserock21:42
*** rdale has quit IRC21:56
*** toscalix has quit IRC22:54

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