LnL has quit [Read error: Connection reset by peer]
LnL has joined #nixos-dev
vcunat has joined #nixos-dev
adisbladis has joined #nixos-dev
timokau has joined #nixos-dev
pie__ has joined #nixos-dev
pie_ has quit [Ping timeout: 240 seconds]
<Enzime>
gchristensen: graham-at-target? :p
<gchristensen>
Enzime: that is my github account for work I do on behalf of Target
<Enzime>
gchristensen: is it because your work owns the IP
<Enzime>
or you just don't want to log into your main account?
<gchristensen>
I think it is nice to make it clear that they're the reason I'm making it. Also: yes, they do own everything I make which isn't a public open source patch, and also yes, I don't want to log in to my main account on their hardware, because then they have my credentials.
adisbladis has quit [Quit: ERC (IRC client for Emacs 26.1)]
goibhniu has quit [Ping timeout: 268 seconds]
orivej has joined #nixos-dev
orivej has quit [Ping timeout: 260 seconds]
<infinisil>
What should be done for a new package regarding meta.platforms, when you have no idea what platforms it will run on?
<infinisil>
Can it just be not set in that case?
jtojnar has quit [Ping timeout: 240 seconds]
<Dezgeg>
surely you have tried it on at least something? put that
jtojnar has joined #nixos-dev
<vcunat>
Or you may put .all in there, make a PR, and use Borg to try some other platforms as well (and then correct the PR if it doesn't work and you don't want to spend time fixing them).
<{^_^}>
vcunat: 1 day, 19 hours ago <LnL> to ping me about staging
<vcunat>
LnL: ping :-)
<Dezgeg>
that works as well... but I personally won't be putting e.g. darwin to my packages at least because I have no way to test if it works
<vcunat>
I've been online for longer time, but apparently the bot only pings when one posts.
<gchristensen>
bots aren't aware of joins / parts
<vcunat>
Yes, building doesn't test everything.
<LnL>
infinisil: meta.platforms should be whatever upstream supports IMHO, while it's not what most people do I consider brokenness to be something different
<vcunat>
From Hydra perspective, repeatedly building something that's bound to fail is wasteful.
<LnL>
vcunat: there's some important darwin stuff in staging and I'm pretty sure some security updates where added
<vcunat>
I know about some security updates
<vcunat>
and I know about a regression
<gchristensen>
and from a user's perspective, it is frustrating to expect it to work and have it fail
<LnL>
I started an eval a few days ago but somebody canceled (possibly to give priority to staging-next)
<vcunat>
ah, right, there's a new one now
<Dezgeg>
how do you know a priori what upstream supports?
<LnL>
oh, so that stuff is probably already in next?
<vcunat>
LnL: I merged staging-next to master yesterday but haven't added anything into staging-next yet
<vcunat>
(though Matthew Bauer added some things in there)
<Dezgeg>
yeah I don't disagree that there could be some things to revisit for this meta.platforms stuff since now we have more things to think of, like the aarch64 builder and this musl stuff and this cross stuff
<Dezgeg>
but it would really need some actual design and discussion - for instance there is apparently this meta.badPlatforms in some packages nowadays which isn't documented anywhere