<ekleog>
If people have opinions, I'm wondering whether it would be a good idea to request a link to the changelog in the commit message of commits bumping package versions https://github.com/NixOS/nixpkgs/issues/50483
<{^_^}>
#50483 (by Ekleog, 46 seconds ago, open): Package upgrades: Request link to changelog in commit message?
<drakonis>
how would anyone deal with packages that are repo snapshots?
<ekleog>
adding `Changelog: none`
<ekleog>
if there's no changelog provided by upstream :)
<drakonis>
seems counterproductive to be fair, but alright
<ekleog>
yeah, it's kind of an issue, but I don't really see how we could fix it from a distro standpoint :/
<ekleog>
TBH I'm not really sure this is a good idea myself either
<drakonis>
there's a lot of packages that actually don't provide logs either
<drakonis>
they usually assume the user can gleam the changes from the repo most of the time
<drakonis>
bar the big projects, these do provide logs
<ekleog>
yeah, well, that's a wrong assumption because a git log and a changelog are totally different things :(
<ekleog>
like, a refactoring shouldn't make it to the changelog, but the bugfix that was done at the same time than the refactoring should
<ekleog>
but yeah, projects do that often :/
init_6 has joined #nixos-dev
<samueldr>
ekleog: maybe part of the contribution template, highly suggested, maybe even requested when missing?
<samueldr>
(I mean, if I were to see an upgrade without either "no changelog" or "changelog here" I would ask first)
lassulus_ has joined #nixos-dev
lassulus has quit [Ping timeout: 252 seconds]
lassulus_ is now known as lassulus
<thoughtpolice>
Couple more failures for release-18.09 should land soon, hopefully!
<thoughtpolice>
Err, they should go away, I mean. That.
<colemickens>
looks like rt.perl.org is blocking the installer tests for nixos-unstable
<timokau[m]>
The 18.09 channel and the unstable channel are both 9 days old, which is why I thought there may be some common issue
<timokau[m]>
I should learn how to interpret hydras test matrix to find out what is blocking
<ekleog>
most recent evaluation is left (for like 6 months I thought it was at the right ._.), and once the whole column is green the channel is supposed to move forward :)
<ekleog>
and I'm pretty sure the two failures are unrelated, I'm the first who broke nixos-unstable (but it's been fixed for a few days) and AFAIK the issue hasn't been backported :p
<timokau[m]>
Hm first of all how do I find the damn thing again?
<timokau[m]>
The test matrix I mean
<ekleog>
timokau[m]: howoldis, the rightmost column
<ekleog>
about howoldis, anyone remembers who handles it? I'd like to suggest removing no-longer-supported channels (ie. <=17.09), so that there's less red for bystanders who don't know they are no longer supported :)
<timokau[m]>
Oh right, was trying to find it by clicking through hydra directly
<timokau[m]>
But then why does the test matrix still display the evaluation where one test fails while you linked to a newer one where all tests pass?
<ekleog>
yeah, hydra interface is really hard to navigate, sometimes I wonder if it's even a connected graph
<ekleog>
timokau[m]: the trick is most recent evaluation is the leftmost one
<ekleog>
leftmost one is all-green for nixos-unstable
<timokau[m]>
Ooh I even read that comment but somehow parsed it as "rightmost" and thought "duh"
<ekleog>
heh :p
<timokau[m]>
Just one more unintuitive detail I guess
<ekleog>
I told you it took me 6 months to get it :p
<ekleog>
for release-18.09 it looks like some builds have been restarted (“?” are not-yet-finished runs), so I guess there has been some kind of transient failure
<timokau[m]>
So then that is nixpkgs-unstable? Because nixos-unstable still hasn't advanced
<timokau[m]>
According to howoldis it should be nixos-unstable, so whats stopping it?
<ekleog>
yeah, I don't get why nixos-unstable hasn't advanced, looks like it finished correctly 1h ago
<timokau[m]>
Maybe "all tests pass" doesn't mean "all builds finished"?
<ekleog>
I would guess either hydra or howoldis or both have a cron that runs every 1hr or something like it, but the “still waiting for builds” does make sense too
<timokau[m]>
Well at least now I know more. Thanks for the crash course :)
<timokau[m]>
How are you not a nixos member yet? Didn't that come up here just recently?
<ekleog>
that's not something for me to decide :p
<ekleog>
(iow, no idea, I don't know if a request has been formally made ^^')
<timokau[m]>
Well my voice doesn't have much weight here either but I think you should be :)
<timokau[m]>
domenkozar ?
phreedom has quit [Remote host closed the connection]
<ekleog>
thank you :)
phreedom has joined #nixos-dev
drakonis_ has joined #nixos-dev
drakonis has quit [Ping timeout: 272 seconds]
orivej has quit [Ping timeout: 268 seconds]
init_6 has quit [Ping timeout: 244 seconds]
drakonis_ has quit [Remote host closed the connection]
<ekleog>
any ofborg trusted user (who can build for mac) available to trigger a build on https://github.com/NixOS/nixpkgs/pull/50505 ; if possible with @GrahamCOfBorg build php71 php72 and @GrahamCOfBorg test owncloud elk ?
<ekleog>
(there's a supposed building bug on mac, to check it still happens with latest version)
<timokau[m]>
Sure, done (if ofBorg knows how to take two commands in a single comment)
<ekleog>
thanks! :) (and yes it does :))
drakonis has joined #nixos-dev
<ekleog>
looks like it still fails :(
<andi->
Didn't we have a 18.09-small jobset on hydra? I am unable to find that.
<andi->
oh, it just isn't visible in the overview but still exists o.O
<ekleog>
… wow, github appears to now give a link to when people force-push, with the differences between before-force-push and after-force-push :D
<andi->
nice
<andi->
someone should probably have a look at the build queues again.. There is a job that has been running for 6d: https://hydra.nixos.org/build/83923477 ( LnL can you abort that?)
orivej has joined #nixos-dev
<LnL>
hmm, that doesn't seem to actually stop them
* LnL
aborts all the jobs
<LnL>
I can't find those last 2
joachifm has joined #nixos-dev
joachifm has left #nixos-dev [#nixos-dev]
<gchristensen>
I'm not supposed to type yet, but I hopped on early to kill those jobs. signed, one-handed-graham
<gchristensen>
andi- ^
drakonis has quit [Quit: WeeChat 2.3]
<cransom>
i'm telling on you.
<LnL>
I think I already killed everything that was blocking the release, the rest probably wasn't very urgent