| === Malsasa_ is now known as Malsasa | ||
| === Malsasa_ is now known as Malsasa | ||
| === Malsasa is now known as Guest94873 | ||
| === Malsasa_ is now known as Malsasa | ||
| === duflu_ is now known as duflu | ||
| === marcusto_ is now known as marcustomlinson | ||
| === Malsasa_ is now known as Malsasa | ||
| === greyback_ is now known as greyback | ||
| sil2100 | Saviq: hey! Did you guys see the unity8 AP test failures on the vivid dashboard? | 10:07 |
|---|---|---|
| Saviq | sil2100, yeah, we're having trouble reproducing locally unfortunately :/ | 10:08 |
| Saviq | every time I release, I run the whole suite and get 100% pass (with maybe one flaky) | 10:08 |
| sil2100 | Saviq: ouch... from the recent discussions I had with CI, smoketesting shouldn't be doing anything specific during image testing | 10:08 |
| sil2100 | hm | 10:08 |
| sil2100 | I think the only difference is in the unlock script | 10:09 |
| sil2100 | Maybe that's the cause? | 10:09 |
| Saviq | unlikely | 10:09 |
| Saviq | we'd need a video of the phone as it's running the tests | 10:09 |
| sil2100 | I wonder if something like that can be arranged | 10:11 |
| === MacSlow is now known as MacSlow|lunch | ||
| === alan_g is now known as alan_g|lunch | ||
| === Facu is now known as Hacu | ||
| dandrader | Saviq, I had to manually add Jenkins to https://code.launchpad.net/~unity-team/unity8/shellRotation/+merge/248400. Will that work? | 13:33 |
| dandrader | I wonder why Jenkins wasn't there in the first place | 13:33 |
| Saviq | dandrader, huh, good question, I'm not totally sure how the MP scrubber works, but I'd say it probably won't change anything, let me see the job list | 13:34 |
| === Hacu is now known as Facu | ||
| === dandrader is now known as dandrader|afk | ||
| === MacSlow|lunch is now known as MacSlow | ||
| === alan_g|lunch is now known as alan_g | ||
| === dandrader|afk is now known as dandrader | ||
| === sonne_ is now known as sonne | ||
| === dandrader is now known as dandrader|lunch | ||
| === dpm_ is now known as dpm | ||
| === dandrader|lunch is now known as dandrader | ||
| === alan_g is now known as alan_g|EOD | ||
| === dandrader is now known as dandrader|afk | ||
| kgunn_ | Saviq: so i was playing around with cross building, and wanted to add a ppa to a chroot | 18:22 |
| kgunn_ | was reading https://wiki.ubuntu.com/SimpleSbuild#Temporarily_adding_PPAs | 18:22 |
| kgunn_ | why does it say "add the ppa but don't update" ? | 18:23 |
| kgunn_ | seems odd | 18:23 |
| kgunn_ | ....i've been tinkering and have noticed, apt-get update and upgrade don't seem to behave the same way | 18:23 |
| kgunn_ | as my regular machien | 18:23 |
| === dandrader|afk is now known as dandrader | ||
| anpok | i guess because it updates on every build run | 18:31 |
| anpok | and hm that way you reduce the impact on the source chroot | 18:33 |
| kgunn_ | anpok: so i have to re-run mk-sbuild ?...if so, what's the point of sbuild-update and such ? | 18:43 |
| anpok | kgunn_: hm you dont have to rerun mk-sbuild .. i mean the instruction tells to add the ppa to the source chroot.. which should usually only contains the build essentials.. and you want to stay clean | 18:53 |
| anpok | +it | 18:53 |
| anpok | as far as i understood the build happens in an overlay .. that usually runs apt update | 18:53 |
| anpok | so there the ppa will affect the build process | 18:54 |
| Saviq | kgunn_, what anpok said, unless you want the PPA to be added permanently, don't update the *source* chroot - it will still update every time you build | 20:32 |
| Saviq | kgunn_, this way you can just remove the .list file and have the chroot pristine | 20:32 |
| kgunn_ | Saviq: so say i don't mind, and i went into the chroot and did update/upgrade...i would expect to see it pull packages...but it didn't | 20:33 |
| kgunn_ | i'm just wondering...maybe i don't understand | 20:33 |
| Saviq | kgunn_, no, that should work just fine | 20:33 |
| Saviq | kgunn_, apt-cache policy $a_package | 20:33 |
| kgunn_ | Saviq: right...i do that, and it shows the ppa i added as a candidate...but it doesn't come up as installed | 20:34 |
| kgunn_ | do i have to install manually ? | 20:34 |
| Saviq | kgunn_, is any other installed? | 20:34 |
| * kgunn_ checks | 20:34 | |
| Saviq | kgunn_, build chroots are generally very small | 20:34 |
| Saviq | kgunn_, meaning you only have the base system, and any dependencies are installed on demand | 20:34 |
| kgunn_ | Saviq: right...so apt maybe doesn't behave the same way | 20:35 |
| kgunn_ | i was gonna build macslows branch | 20:35 |
| Saviq | kgunn_, why do you say it doesn't behave the same way? | 20:35 |
| kgunn_ | just installed silo 0 ppa first... | 20:35 |
| Saviq | kgunn_, "installing" a ppa | 20:35 |
| kgunn_ | Saviq: meaning, i thot after upgrade apt-cache policy would show those packages from that ppa as installed | 20:36 |
| Saviq | kgunn_, is nothing more than upgrading your packages from a ppa, not installing all the ones available in the ppa | 20:36 |
| kgunn_ | got it | 20:36 |
| kgunn_ | so if it wasn't installed prior | 20:36 |
| kgunn_ | then it still won't be | 20:36 |
| Saviq | yup, nothing will change | 20:36 |
| Saviq | but | 20:36 |
| kgunn_ | i get it now | 20:36 |
| Saviq | when you build, that's when it's gonna get pulled in | 20:36 |
| kgunn_ | got it | 20:36 |
| Saviq | as a build dependency | 20:36 |
| kgunn_ | like magic ? | 20:36 |
| kgunn_ | :) | 20:36 |
| Saviq | kgunn_, well, debian/control does that for you | 20:37 |
| kgunn_ | i got it...right | 20:37 |
| kgunn_ | this is the cmake part where it goes and looks for pkgs | 20:37 |
| kgunn_ | and then installs | 20:37 |
| kgunn_ | starting to make sense | 20:37 |
| Saviq | not cmake | 20:37 |
| Saviq | kgunn_, sbuild just picks debian/control and installs everything declared in build-depends there | 20:37 |
| Saviq | kgunn_, cmake can only verify that something is available, can't install anything | 20:37 |
| kgunn_ | got it | 20:38 |
| kgunn_ | so they work in coordination | 20:38 |
| Saviq | which is why we get this unfortunate duplication of deps between cmake and debian/control | 20:38 |
| kgunn_ | ok...so i'm just gonna try some building now and watch the magic | 20:38 |
| Saviq | but that's because other people don't necessarily use dpkg | 20:39 |
| Saviq | so they just want the source and will install the dependencies by whatever means they want to | 20:39 |
| kgunn_ | right, but we should all be uniform in this respect | 20:39 |
| kgunn_ | w/in canonical projs | 20:39 |
| kgunn_ | (in theory i suppose :) | 20:39 |
| Saviq | oh yeah, by "other people" I mean people that don't use a debian-derived distro | 20:40 |
| kgunn_ | Saviq: so do you mount your home dir to work out of ? like build existing bzr branches you want to cross build ? | 20:40 |
| kgunn_ | i did that...just curious is that's common | 20:40 |
| Saviq | kgunn_, I have my home mounted in the chroot, yes, that's the easiest, really, as you don't need a separate space to share between chroot and host | 20:41 |
| Saviq | kgunn_, but, if you use sbuild, it doesn't matter, as sbuild does all the things it needs to put the source and output in the right places | 20:41 |
| kgunn_ | mmm | 20:42 |
| Saviq | kgunn_, it only really matters if you open the chroot manually (via schroot) and muck around in there | 20:42 |
| kgunn_ | is it just personal pref ? | 20:42 |
| kgunn_ | for some reason working in chroot makes me feel better | 20:42 |
| Saviq | kgunn_, it does mean you have to do manual steps | 20:42 |
| Saviq | kgunn_, when with sbuild you just call it in a source tree and go have a tea | 20:43 |
| Saviq | kgunn_, like just going at it manually you have to install deps manually (sure, via apt build-dep, but still it's you that has to call it) | 20:43 |
| Saviq | kgunn_, caveat: sbuild doesn't deal well with the :native bit in our debian/control, need to get rid of that if you want to use it | 20:44 |
| * Saviq files a bug | 20:44 | |
| kgunn_ | i'm reading... | 20:50 |
| kgunn_ | Saviq: so i just sbuild -s <dir of source> | 20:50 |
| Saviq | kgunn_, not -s | 20:51 |
| Saviq | kgunn_, -s would build you just a source package (so tarball + some dpkg bits) | 20:52 |
| Saviq | kgunn_, to cross-build, you go `sbuild --host=armhf -c name-of-chroot` | 20:52 |
| kgunn_ | Saviq: sorry, i'm dense....how does it know what you want built? if you have a local source, e.g. from some branch-yet-to-be-proposed | 20:54 |
| kgunn_ | like where do you tell it "i want you to use this source" | 20:54 |
| kgunn_ | ah...nvmd...i found it i think | 20:59 |
| kgunn_ | i skipped over that part on the wiki | 20:59 |
| Saviq | kgunn_, you can pass the source folder, or a .dsc file (that you can build with `bzr bd -S` from the branch) | 21:00 |
| Saviq | kgunn_, or you can even go `bzr bd -S --builder='sbuild -c chroot-name --host=armhf'` | 21:00 |
| kgunn_ | yep i see it now under the heading "using the schroot" | 21:00 |
| Saviq | kgunn_, the last one has the disadvantage of locking the bzr branch for the duration of the build | 21:01 |
| kgunn_ | Saviq: do i have to dch -i & update-maintainer to make it happy? | 21:02 |
| kgunn_ | would think i could skip...but i'm usually wrong :) | 21:02 |
| kgunn_ | :) ....going for tea | 21:05 |
| Saviq | kgunn_, you can skip, you'll have to install the .debs explicitly anyway | 21:09 |
| Saviq | kgunn_, -i only helps when you put them in an archive somewhere | 21:10 |
| Saviq | kgunn_, update-maintainer, never had to do that myself | 21:10 |
| kgunn_ | right...it was just instructions, figured it was to prevent direct uploads w/o traceability | 21:10 |
| kgunn_ | but then again deb is too smart sometimes...didn't know if it would gripe | 21:10 |
| Saviq | kgunn_, if you want to upload to the archive, then yeah, that's when stuff are more tricky, you need to sign the source package with gpg key that's accepted by the archive, other than that you can do what you want | 21:42 |
| kgunn_ | Saviq: just interesting...left for a tea (aka exercise) too quick, failed with dep isssue | 22:35 |
| kgunn_ | sbuild-build-depends-core-dummy : Depends: crossbuild-essential-armhf but it is not going to be installed | 22:35 |
| kgunn_ | i'll install manually...but wouldn't have expected this to fail | 22:35 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!