/srv/irclogs.ubuntu.com/2019/06/27/#snappy.txt

=== pleia2_ is now known as pleia2
mborzeckimorning05:34
mborzeckihttps://github.com/snapcore/snapd/pull/6992 needs a 2nd review and can land06:13
=== morphis16 is now known as morphis
zygagood morning06:54
zygacoffee and work06:54
mborzeckizyga: https://github.com/snapcore/snapd/pull/6992 just needs a 2nd review and we can land it07:01
mborzeckishould be easy, just some test cleanup07:01
=== pstolowski|afk is now known as pstolowski
pstolowskimorning07:04
mborzeckipstolowski: hey07:10
pstolowski+107:10
mborzeckiyay07:15
zygamborzecki: looking07:27
zygahey pawel :)07:27
zyga9:3007:27
zygabetter than yesterday07:27
zyga26C and looking great07:27
mborzeckizyga: nice07:29
mborzeckizyga: sorry to keep poking you :) https://github.com/snapcore/snapd/pull/692207:30
zygadoing now07:30
mborzeckianyone managed to order pi4 4GB variant?07:32
zygamborzecki: I was considering07:38
zygabut07:38
zygamborzecki: here I can get the es keyboard layout (I wanted the desktop kit)07:38
zygamborzecki: in poland I can get the .pl keyboard layout07:39
zygamborzecki: in both cases I'd have to wait up to a month07:39
zygamborzecki: apart from micro-hdmi (WHY!?!) it is ok07:39
zygamborzecki: I'd prefer to have 3 usb-c instead07:39
zyganobody loves micro hdmi cables07:39
zygamborzecki: reviewing, should be done soon07:40
mborzeckizyga: i checked a couple of places but sold out07:40
zygamborzecki: yeah07:40
zygathey flew from the shelves07:40
zygajust wait, GBP/PLN is favourable :)07:40
mborzeckizyga: maybe availability was like amd's vega cards ;)07:43
zygamborzecki: at this price point they will fly off any store07:49
zygamborzecki: I read a funny tweet yesterday, cannot remember it exactly but it roughly read "how many people will buy it just to put it on the shelf next to older pi's, also unused'07:50
pstolowskiheh, was considering it a few days ago (botland.com.pl). out of stock already07:51
zygapstolowski: preorder, they are the only official reseller07:53
zygapstolowski: others buy from them07:53
=== abeato_ is now known as abeato
zygamborzecki: https://github.com/snapcore/snapd/pull/6922#pullrequestreview-25505027208:40
zygamborzecki: I'll task switch to work on the feedback from both you and jstrand08:41
zygaI think this will land today08:41
mborzeckizyga: ok, great08:42
mborzeckizyga: i think i'll leave a todo about preserving file permissions, osutil.CopyFile can do that but in a funny way by calling cp, and while it's ok for a scenario when you build an image, it feels fragile when you update :/08:44
zygamborzecki: I think it's not needed now but it'd be worth making this explicit (as in, this is not supported and is documented as such)08:44
mborzeckizyga: mhm, sounds fair08:45
zygamborzecki, pstolowski: desktop set or just bare pi4? (hard questions)08:47
mborzeckizyga: i'll try getting the regular one, the keyboard/mouse look like toys tbh08:49
zygayeah but OMG shiny :)08:50
pstolowskizyga: bare08:50
mborzeckizyga: and the desktop set is what $100+? :P08:50
zygait's not much, case + psu08:51
zyga+ cables you'd otherwise how to source separately (2$ a piece)08:51
zyga+ keyboard / mouse (with only one usb port)08:51
Chipacaxnox: 00-ꜱɴᴀᴘᴅ-ᴄᴏɴꜰɪɢ.ʏᴀᴍʟ09:05
valentindzyga, It is not straight forward to bind mount /usr from the snap application. Snapd bind mounts several things within /usr, but it does it before the layout binding. So those paths disappear. I would need to make a script to generate all the paths to mount in order not to mount over those paths.09:11
zygavalentind: yeah, mounting all of usr is tricky :/09:11
zygavalentind: can you help me understand why you'd need that?09:11
zygavalentind: I assumed that the freedesktop runtime would be much like flatpak runtime model09:12
zygavalentind: with / being a full read only image (mostly)09:12
zygavalentind: and /app being the application image09:12
zygavalentind: with extra places for writable data09:12
valentindYes.09:12
zygavalentind: in this model, can you not keep the runtime as the base snap, with /usr and all the other suspects as-is09:12
valentindWell, if we cannot run commands from /usr/bin it is annoying.09:12
zygavalentind: perhaps that's what we should fix, we could change snapd to generate different apparmor template for if the base snap is the freedesktop runtime09:13
valentindI could also just mount /usr/bin from the app with the files copied from the runtime.09:13
valentindzyga, Yes, this is what I would like.09:13
valentindBut I do not know how much time it would take to get there.09:13
zygavalentind: perhaps not that much, can you tell me if what you are missing is simply09:14
zyga(in apparmor terms)09:14
valentindI do not mind providing a apparmor template for freedesktop sdk, this is fine.09:14
zyga /usr/bin/** ixr,09:14
zygaperhaps coupled with something similar for /usr/lib09:14
valentindThis would be enough I think.09:15
zygaI think we can do that then, is there a thread I can reference in the pull request?09:15
valentindI think we need run only for ld.so, and it is already there.09:15
zygavalentind: can you make that modification locally?09:15
valentindLibraries do not need execution otherwise. Just that the dynamic loader can be executed.09:15
valentindSure, I can do that.09:15
zygavalentind: you can install a snap, edit /var/lib/snapd/apparmor/profiles/snap.$SNAP_NAME.$APP_NAME, reload that using apparmor_parser -r /path/to/profile and see if that fixes it09:16
zygavalentind: once you have the set of changes I can make a quick PR09:16
zygathank you!09:16
valentindAh libexec as well09:18
zygaI'm sure there's a small list you can find by iterating09:18
xnoxChipaca:  ship it!09:18
xnoxChipaca:  i should totally do that somewhere sometime, as the defualt filename.09:18
valentindzyga, yes that works. I will check a bit more to see what makes sense to add.09:20
zygathanks!09:20
Chipacaxnox: that's the output of 'unifonter -k k 00-snapd-config.yaml' fwiw09:20
zygabrb, short walk to the library09:25
zyga(free wifi, ac and super quiet)09:25
zygare09:42
zygaI forgot how nice this place was09:42
valentindzyga, I have found few other places than /usr/bin/ and /usr/libexec/ where there are binaries. But I think I will change freedesktop sdk to move them to libexec. Not sure why they are there.09:43
zygavalentind: what kind of executables were those?09:52
valentindIt is in the library directory.09:52
valentindThey look like libexec things.09:52
zygavalentind: in /usr/lib?09:57
zygavalentind: I think it's okay to add to the template instead of changing the base snap too much09:57
valentindOK, then.09:57
valentindSo yes, {,/usr}/bin/* {,/usr}/libexec/** and {,/usr}/lib/**. Should I make a mr? Or will you take care of that?09:59
zygavalentind: I can do that10:04
zygavalentind: I have some ideas where that should be10:04
valentindOK Great.10:04
Chipacacould I have a second review of a very very simple but important pr, #7040 ?10:31
Chipacahttps://github.com/snapcore/snapd/pull/7040/files10:31
mborzeckiChipaca: do you know if we're doing another . release?10:39
zygamborzecki: can you do another pass over infofile subset of https://github.com/snapcore/snapd/pull/7026/files10:45
zygamborzecki: I will address the remaining comments now10:45
mborzeckizyga: sure10:47
zygaI need to take a break, the library is closing soon10:50
zygasee you soon10:50
wieczorek1990Can I get any love in the reviews? https://snapcraft.io/gitl11:06
mborzeckizyga: added some comments about new info scanner12:00
zygaThanks12:06
zygaResuming work now12:06
wieczorek1990How often are review-tools updated?12:09
zygawieczorek1990: on demand, I think12:21
mborzeckiwieczorek1990: jdstrand likely knows best12:34
zygabrb for lunch12:37
zygawieczorek1990: or roadmr also knows AFAIK12:38
wieczorek1990mborzecki: jdstrand told me to wait :)12:49
jdstrandwieczorek1990: the review-tools are updated. roadmr is getting them to production as we speak. it should be soon12:54
wieczorek1990cool12:54
=== ricab is now known as ricab|lunch
zygamborzecki: trying to join standup13:01
jdstrandwieczorek1990: it seems likes you may have auto-uploads setup for gitl? if so, you may want to pause them for the next day or so until they start passing automated review13:23
wieczorek1990just let me find that option :)13:24
wieczorek1990jdstrand: removed from the build list13:25
jdstrandwieczorek1990: I'll ping you when the tools are in proc then13:25
wieczorek1990jdstrand: thanks13:26
dot-tobiasPing kyrofa / sergiusens: https://forum.snapcraft.io/t/cmake-part-with-build-snaps-gets-mangled-include-parameters/ → any quick hint what's going wrong here?13:51
ogradot-tobias, what did you break again ?13:52
ogra:)13:52
dot-tobiasogra: Just one of my user-errors again, I hope 😄13:55
dot-tobiasHolding It Wrong™13:55
ogramust be the heat13:56
dot-tobiasmost definitely13:56
=== ricab|lunch is now known as ricab
jdstrandwieczorek1990: ah, the update is in prod now and gitl passes automated reviews :) (see r141). all the other revisions are now being scanned and passing (up to r147 now)14:37
zygajdstrand: hello, I sent several updates to https://github.com/snapcore/snapd/pull/7026 -- could you do another pass please15:01
zygajdstrand: I'm still doing a few changes to the comments as you suggested but the "essence" of the changes are good now IMO15:01
jdstrandzyga: ok15:19
* zyga moved closer to home 15:34
jdstrandSaviq: is is possible to adjust the mirror when using snapcraft with multipass? (maybe that is a question for sergiusens... hi to both of you!)15:38
jdstrandI'd like to use something other than archive.ubuntu.com15:38
Saviqjdstrand: not easily with snapcraft, no - it doesn't give you a way to modify the cloud-init15:39
Saviqtoday, only option would be to `snapcraft pull` a part, then get into the instance and switch the mirror15:39
Saviqlong-term, we're working on configurable cloud-init defaults in Multipass, which could help with this (although snapcraft provides a cloud-init, I wonder if we should mess with it…)15:40
Saviqjdstrand: one way or another, yes I'd say it's a sergiusens question :)15:40
Saviqb/c I'd rather see snapcraft allowing you to specify a mirror rather than you hacking the instances under its feet :)15:41
Saviqthere's also an option of a snapcraft plugin…15:41
Saviqbut you'd need to modify each snapcraft.yaml to make it use the plugni15:42
Saviqso yeah, I really think this kind of thing would be a snapcraft one :)15:42
ograjdstrand, you can easily do that by adding an extra part to your snapcraft.yaml that calls add-apt-repository ... i.e. pilibs here adds a PPA (but any proper deb archive should actually work here), pistream (the part afterwards) then uses a package from there (libraspberrypi-bin) https://github.com/ogra1/picamera-streaming-demo/blob/master/snap/snapcraft.yaml#L3715:45
ograthat way should work in any build env (multipass, launchpad buildd etc)15:46
ackkhi, is it possible for a snap to know if it's installed with --devmode?15:48
Chipacaackk: what for?15:51
ackkChipaca, basically conditional logic in hooks as you can do stuff like chown if in devmode15:51
ogragiven that devmode isnt something you should/would use permanently you should really not do any permanent hacks in your hooks for it15:52
Chipacaackk: if you really really need to do this, try snap list $SNAP_INSTANCE_NAME15:54
Chipacaackk: if you can do that, you're in devmode*15:54
Chipaca* or you have snapd-control15:54
ackkChipaca, ah, interesting trick :)15:55
ackkChipaca, thanks15:55
Chipacaactually it works also if you have snapd-control :)15:56
Chipacabecause you can't run snap :)15:56
ackkheh, cool15:56
=== pstolowski is now known as pstolowski|afk
* zyga -> nap16:42
zygazzzz16:42
wieczorek1990https://www.irccloud.com/pastebin/3kyusPpM/17:04
wieczorek1990So, I've been trying to enable aliases in gitl run using snap, and I'm kind of blocked. Getting this error:17:04
wieczorek1990error: could not lock file /home/luke/.gitconfig: Permission denied17:04
wieczorek1990Thought there won't be errors after adding the personal-files interface17:04
wieczorek1990Here's my snapcraft.yaml: https://github.com/wieczorek1990/gitl/blob/master/snap/snapcraft.yaml17:04
wieczorek1990Here's the function in git causing problems: https://github.com/git/git/blob/ab15ad1a3b4b04a29415aef8c9afa2f64fc194a2/lockfile.c#L7317:04
wieczorek1990Any hints?17:04
wieczorek1990Other thing is that normal aliases don't work, they exit with a message like this:17:07
wieczorek1990expansion of alias 's' failed: 'status' is not a git command17:07
wieczorek1990but they work if written as "s = !git status", strange17:07
ogradid you actually connect the interface ? it surely doesnt automatically connect17:08
wieczorek1990it's there when I run `snap connections gitl`17:09
ograand connected ?17:10
ogra(did you explicitly run snap connect for it after install)17:10
wieczorek1990No, I ran install from the store17:10
ograyeah, that wont connec it17:11
ograyou need to explicitly run snap connect <snapname>:<plug>17:11
ogra(i never used personal-files so i dont know where to you need to connect)17:11
wieczorek1990connecting did not change a thing17:14
wieczorek1990Did not expect so much work for deploying app with snap :)17:24
ograwell, the good thing about snap packaging is that once you figured it out you usually do not need to touch it again17:29
zygastore is somewhat red17:53
zygahttps://www.irccloud.com/pastebin/YI8ilsRW/17:54
zyganoise][: ^ https://status.snapcraft.io/ is green, should I see anything ?17:54
noise][nothing alerting but let me take a look17:56
zygait's not all the stuff I do but I've restarted tests over small one-off failures a few times17:59
Paddy_NIWill the "Gnome Software" ever be gaining support for displaying markdown read from a snap apps description properly?18:01
noise][zyga: we are seeing some elevated err rates, coinciding with a deploy, investigating and will update status page in a minute18:01
Paddy_NIAscii cast and video support would also be awesome18:02
zygaas always, thank you for keeping the store running :)18:02
noise][zyga: all fixed now, was 1 bad unit in a pool, monitoring caught it but as a low-prio so nobody had looked at it right away18:13
noise][we will fix that as well :)18:13
zyganoise][: super, thank you19:04
=== harrisj_ is now known as harrisj
jdstrandSaviq: ack, thanks20:13
jdstrandogra: thanks. the thing is I want to change the mirror (not a ppa) only for local builds, not LP builds20:14
zygajdstrand: https://github.com/snapcore/snapd/pull/7026 is green20:17
zygaI have some more tests piled up but I'll wait for your review first20:17
zygajdstrand: the only significant change I'd like to introduce is to allow to define a memory cap so that unbound input cannot happen20:17
jdstrandzyga: is that in a subsequent PR or this one?20:25
zygajdstrand: perhaps subsequent since this is green now21:41
zygajdstrand: or either if you'd like to see that earlier21:42
zygajdstrand: I'm mainly trying to be practical and land it soon21:42
jdstrandzyga: I'd prefer to review 7026 if you're done with it. I think that improvement could come in a subsequent PR (we do control the file after all, so I consider this landable without it (sans my 2nd review))21:43
zygajdstrand: +121:43
zygajdstrand: I'm happy with that21:43
jdstrandzyga: put another way, I didn't want to review it and then have you add more :)21:44
zygasure21:44
jdstrandif we could avoid it (can't always)21:44
jdstrandok, so I'll take a look at it after a couple things. might be tomorrow morning21:44
zygathank you, I'm heading to bed anywa21:45
zygaI'll work with maciej to get it re-reviewed21:45
zygahopefully it could land tomorrow21:45
jdstrandzyga: good night! :)21:45
zygagood night :)21:45

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!