[02:50] PR snapd#10677 closed: snapstate: fix misleading `assumes` error message [05:34] morning [05:38] mborzecki: hi! [05:39] mardy: heya [06:28] morning [06:33] pstolowski: hi! [06:37] pstolowski: hey [07:30] PR snapd#10682 opened: many: shellcheck fixes <âš  Critical> [07:31] mvo: hi, quick question there ^ [07:32] mvo: btw can you land https://github.com/snapcore/snapd/pull/10617 ? [07:32] PR #10617: cmd/snap: add Size column to refresh --list [09:51] PR snapd#10683 opened: interfaces/dsp: add /dev/ambad into dsp interface [09:53] mvo: when you're around, can you take a look at https://github.com/snapcore/snapd/pull/10158 ? [09:53] PR #10158: packaging: build without dwarf debugging data [09:53] i think it's ok to land it, there's an observable gain when building the package manually, however we need to investigate why this doesn't translate to size change in the snapd snap artifact produced by github workflow [10:15] golang question: I want to mock this function: https://github.com/mardy/snapd/blob/mount-control-1/wrappers/services.go#L1661 [10:16] but I cannot, because the "interacter" type is private: "cannot use f (type func(snap.PlaceInfo, string, Interacter) error) as type func(snap.PlaceInfo, string, wrappers.interacter) error in assignment" [10:25] mardy: yeah you cannot reference a private symbol from another package. try to define a type alias in export_test.go, i.e. type Interacter = interacter [10:28] mardy: that should work as long as it's the test for the given package (but not from other packages). in the test you will need to reference the exported type with wrappers.Interacter [10:45] pstolowski: it's a test in another package :-( [10:47] mardy: ok, and you're defining a MockRemoveUnitFiles function in services.go for other packages to use in their tests? [10:50] pstolowski: yes, this is what I'm trying to do: https://github.com/mardy/snapd/commit/0370da64344e69d645b8c7b0114b7a13e9ffd9cc [10:52] good morning [10:52] * zyga-mbp is pretty tired/sleepy today [10:55] mardy: ah so it's a bit different than what i imagined. but nvm.. i'm afraid you will need to make Interacter exported in wrappers [10:56] hey zyga-mbp [10:57] hey [10:57] today I slept for 10 hours :P [10:57] pstolowski: yep, I thought of doing the same, but wanted to ask if there was a better way. Anyway, thanks! [10:57] playing with rpi and my scope to do some experiments [10:58] zyga-mbp: hi! [10:58] but man I feel tired somehow [10:58] probably the wether [10:58] hey mardy :) [10:59] mardy: alternatively, inside snapstate/backend wrap the actual call to wrappers inside a small function that takes/passes progress bar to wrappers, and mock that function instead in your tests, but i'm not sure it's the best [11:02] mborzecki: will look now [11:06] PR snapd#10158 closed: packaging: build without dwarf debugging data [11:08] woot [11:08] thank you for merging mvo [11:08] mvo: I _really_ wonder if upx would make more difference on top of this [11:11] zyga-mbp: on the snap not much but probably in a package [11:13] mvo given upx makes page sharing harder, I would not upx snapd but I would compress snap, snap-seccomp and other short-lived tools [11:15] zyga-mbp: intressting idea [11:15] mvo try upx --brute on snap [11:17] zyga-mbp: hm not entirely convinced, this will impact ram use, so negative impact on systems where there already is some pressure and then lower end cpus will lead to longer start times because the snap command is in the critical chain :/ [11:18] mborzecki ram usage of _snap_ is minimal [11:18] it will take exactly the amount used by snap today [11:18] zyga-mbp: can upx selectively unpack relevant pages only? [11:18] IIRC no, it unpacks to tmp or to temporary memory [11:19] I think your earlier idea to combine binaries was better [11:19] but it's harder with more and more things added to snapd [11:19] mborzecki snap-seccomp is another easy target though [11:19] it's not a critical chain app [11:19] all the initrd programs are in the same pool [11:20] zyga-mbp: we had a system where there was little memory and things were being pushed in and out of the cache, even running `snap` binary was a challenge :P [11:20] mborzecki I recall but TBH that's a lost cause IMO [11:20] go 1.17 is nice, it makes a small difference in the right direction [11:21] zyga-mbp: yeah, maybe, me and ondra ended up with a shell script to replace /usr/bin/snap [11:32] pstolowski, hi [11:32] I left you a comment https://github.com/snapcore/snapd/pull/10443#discussion_r696542490 [11:32] PR #10443: tests: check files and dirs are cleaned for each test [11:33] pstolowski, does it make sense? [11:46] cachio_: replied [11:53] the test situation today is really grim :( [12:13] pstolowski, finally what is working well is [ -z "$(cat /tmp/fs-diff)" ] [12:13] other solutions have provlems [12:13] cachio_: find -empty is wrong as well i guess? [12:14] I think sol [12:14] let me recheck [12:15] i played a bit locally and it seems to be the same as using -s [12:17] yes [12:17] both work bad [12:17] for this scenario [12:17] the only way to make it work well was doing [12:17] if [ ! -z "$(cat /tmp/fs-diff)" ]; then [12:18] pstolowski, thanks for the revie [12:18] w [12:18] I'll push the fix now [12:22] cachio_: nice [12:22] pstolowski, about the tests, I see more frequent errors related to uc20 failed to reboot [12:23] kill-timeouts [12:25] cachio_: one more comment to the PR [12:42] pstolowski, thanks [12:42] PR core20#105 closed: hooks: enable systemd-time-wait-sync.service [12:42] PR core20#106 closed: Add motd and issue to writable /etc files [12:45] cachio_: +1. i think this should be squash-merged [12:55] PR core18#174 closed: Resurrect bash-completion [12:56] pstolowski, thanks [13:11] PR snapd#10617 closed: cmd/snap: add Size column to refresh --list [13:36] PR snapd#10682 closed: many: shellcheck fixes <âš  Critical> [13:42] PR snapd#10684 opened: many: shellcheck fixes === not_phunyguy is now known as phunyguy [15:22] PR snapcraft#3573 opened: snaps: eliminate extra call to query snap info [15:37] PR snapcraft#3574 opened: lint: disable new shellcheck warnings [16:12] PR snapd#10666 closed: snapstate: abort kernel refresh if no gadget update can be found <âš  Critical> [16:21] * cachio_ luncj [16:21] * cachio_ lunch [17:27] PR snapd#10685 opened: many: fix run-checks gofmt check [17:32] PR snapd#10686 opened: many: remove unused/dead code [18:07] PR snapcraft#3574 closed: lint: disable new shellcheck warnings (CRAFT-482) [19:13] PR snapd#10687 opened: tests: use "shellcheck" snap in `go` unit test === mup_ is now known as mup [21:07] PR snapcraft#3573 closed: snaps: eliminate extra call to query snap info (CRAFT-479)