02:13
hmpffff_ has joined #nixos-borg
02:16
hmpffff has quit [Ping timeout: 252 seconds]
07:04
JJJollyjim has joined #nixos-borg
07:19
cole-h has quit [Quit: Goodbye]
07:57
orivej has joined #nixos-borg
08:10
orivej has quit [Ping timeout: 246 seconds]
09:57
orivej has joined #nixos-borg
12:48
orivej has quit [Ping timeout: 260 seconds]
14:10
orivej has joined #nixos-borg
14:52
cole-h has joined #nixos-borg
16:45
hmpffff_ has quit [Quit: nchrrrr…]
16:46
hmpffff has joined #nixos-borg
16:46
hmpffff has quit [Client Quit]
16:46
hmpffff has joined #nixos-borg
16:47
hmpffff has quit [Client Quit]
17:32
orivej has quit [Ping timeout: 260 seconds]
18:11
hmpffff has joined #nixos-borg
18:37
hmpffff has quit [Quit: nchrrrr…]
18:55
hmpffff has joined #nixos-borg
19:22
<
LnL >
Any thoughts on #3584?
19:23
* cole-h
hopes it gets merged ASAP.
19:23
<
cole-h >
Would this be another eval step, or would it just be tacked on to a single/all other steps?
19:23
<
LnL >
I could probably make a one line patch for that which can be applied to nix in the deployment
19:24
<
LnL >
one of the current checks is nix-env -qaP --no-name --out-path
19:24
<
LnL >
my thought was to just add it there
19:25
<
cole-h >
SGTM. I didn't know if it would need to be added to all the steps in order to see the benefit or not.
19:25
* cole-h
hasn't consciously looked at all the eval steps and what they execute
19:28
<
LnL >
it's the reverse of what I was thinking
19:29
<
cole-h >
LnL: How would you translate that PR into a one-line patch in deployment?
19:30
<
LnL >
changing --out-path to use queryOutputs(true)
19:31
<
LnL >
a slight behaviour change but not really a big deal for ofborg I think
19:31
<
cole-h >
Man, overloading in C++ still gets me sometimes
19:31
<
cole-h >
"Wait, how can the same function take both 0 and 1 arguments?"
19:34
<
LnL >
it looks like that's only used by the rebuild checker however
19:38
<
cole-h >
LnL: So, how does changing `--out-path` to use `queryOutputs(true)` change the behavior?
19:40
<
LnL >
that true is onlyOutputsToInstall
19:41
<
LnL >
^ note how --out-path includes all outputs while --outputs only has bin;man
19:41
<
LnL >
because of meta.outputsToInstall
19:42
<
cole-h >
OK, so `bin=...;dev=...` is from `--out-path`, and `--outputs` is `bin;man`?
19:42
<
cole-h >
Got it. So if you just opted to set that to true in your PR instead of adding a flag, it would break people's scripts if they somehow rely on that format being stable.
19:43
<
LnL >
yeah, then there's no way to query all the outputs anymore
19:44
<
LnL >
but for the rebuild checker, if anything in a package changed all outputs will be different
19:45
<
cole-h >
Could I ask you to to run that `--outputs` stuff on a nixpkgs that was affected by that PHP change? I'm interested in what the output would be (and am currently bouncing back and forth between IRC and writing an essay lol)
19:45
<
LnL >
only difference that a meta change would also show up
19:48
<
LnL >
specifically for outputsToInstall that is, everything else still won't be counted
20:03
<
cole-h >
Perfect. Hope somebody in the know gives you some feedback soon, so it can be merged shortly :D
20:36
<
cole-h >
btw LnL I just noticed the "checks completed" and "builds received" stuff :D Dashboard very pretty now
20:37
<
LnL >
not great but at least it gives a better indication of steps happening
20:39
<
LnL >
wish there was a way to make an event show up as a single dot without changing the resolution
22:04
tilpner has quit [Remote host closed the connection]
23:04
orivej has joined #nixos-borg