<infinisil>
clever: Since you seem to know about this too, are you in favor of merging that PR?
<infinisil>
It's really hard to find people who know about this stuff
<clever>
infinisil: the mkif or the lazyattrs?
<infinisil>
clever: The lazyAttrsOf, but I wouldn't mind another pair of eyes on the other too
<clever>
infinisil: the comment reminds me of a common bug, imports = [ (pkgs.fetchFromGitHub {...}) ]; fails with infinite recursion
<clever>
infinisil: because you must first recurse over the entire imports tree, to know the value of _module.args.pkgs, to know how to fet the imports
<infinisil>
Yeah that's probably the most common cause of inf rec
<clever>
infinisil: if you have an Expr* that happens to contain a string, int, float, path (and some other exceptions), and you are turning it into a Value*, this code will skip creating a thunk, and just make a plain value of the right type
<clever>
infinisil: line 693, deals with things like `let foo=5; bar=foo; in` and will just reference the Value* from foo, so its not defering the .foo, and they can share things better
<infinisil>
Neat
<clever>
infinisil: many places in nixexpr, will then call maybeThunk, ExprAttrs for example, has a set of key=Expr's, and has to now eval them in a given context (foo: rec { bar = foo.bar; baz=42; another=baz; })
<clever>
infinisil: maybeThunk will defer the lookup of bar until later (a thunk that computes foo.ar)
<clever>
infinisil: but maybeThunk will just set baz to 42 right away, and copy baz->another
<clever>
but, that maybeThunk stuff happens in the first layer, and cant see keys from { "foo${"bar"}" = 42; }
<clever>
so if your referencing another key, whose value is dynamic, it becomes a thunk, and figures it out later
<clever>
but line 900 still forces those keys to become strict strings
<infinisil>
clever: I know you like talking about Nix source code, but I'm really just looking for some feedback for the PR xD
drakonis has joined #nixos-dev
<clever>
infinisil: approved the pr
<infinisil>
Cool thanks :D
<infinisil>
Now I wouldn't feel bad if I merged it myself (though I probably won't do that just yet, there's not too many reasons to want this type just yet)
Synthetica has quit [Quit: Connection closed for inactivity]
<andi->
oh, all are the same, they show up as 4 different builds in machine status…
Jackneill has quit [Ping timeout: 240 seconds]
<LnL>
domenkozar[m]: ping
Synthetica has joined #nixos-dev
<andi->
looks like those failure all come from failing DNS?
<domenkozar[m]>
LnL: pong
Jackneill has joined #nixos-dev
<LnL>
hi, I'm not familiar with github actions but I started looking into 10.15 for the installer yesterday and it seems like apfs.util bootstrapping has been fixed since I first tested it
<LnL>
so I think we can get the installer working without having to reboot for synthetic.conf first
orivej has joined #nixos-dev
asymmetric has joined #nixos-dev
<domenkozar[m]>
nice!
<domenkozar[m]>
LnL: do you have something I can play with?
asymmetric has quit [Ping timeout: 268 seconds]
<LnL>
no not yet, spent most of my time yesterday upgrading an old machine
<LnL>
I forget where it's located exactly, but find /System/Filesystems -name apfs.util, and -B is what you want
<gchristensen>
andi-: neat ... :|
asymmetric has joined #nixos-dev
<jtojnar>
worldofpeace do not forget about the references from passthru.tests
<worldofpeace>
Jan Tojnar: I was opening a PR for you to review just now
<jtojnar>
I tried libxmlb.tests and it worked, did not realize it was not ported
<worldofpeace>
Should we rename the passthru attr to `tests.installed-tests` and outputs to `installed-tests`?
<jtojnar>
worldofpeace bash does not support - in variable names
<jtojnar>
so it would not work for outputs in some cases
<andi->
gchristensen: no stress, we should distribute that burden! You shouldn't be the SPOF for that kind of stuff.
wild_willy has joined #nixos-dev
drakonis has joined #nixos-dev
<gchristensen>
andi-: killed
* andi-
bows to gchristensen
<gchristensen>
I think my prometheus monitor endpoint would have displayed this problem, worth checking ...
<gchristensen>
How long builds are taking per server yes
wild_willy has quit [Ping timeout: 260 seconds]
drakonis has quit [Ping timeout: 252 seconds]
drakonis has joined #nixos-dev
notme has joined #nixos-dev
<Ericson2314>
zimbatm: no worries! A few of us had accepted your invite but as there's no notification for anyone for that confusion was bound to happen. I will send an email confirming things in the future
<Ericson2314>
zimbatm: While there wasn't much disagreement, the RFC was in need of some editting so we'll just continue with that next time (though Michael already pushed some changes, thanks, everyone should weigh in and there's more to cover)
<Ericson2314>
zimbatm: Also, another layer of confusion, in my earlier email I was considering the following Thursday rather than today; I had missed that we had this 9am almost-everyone slot in the first when2meet, and so I was thinking just about the second when2meet. That alone deserved an email "oops, I had missed this earlier thursday looks good, let's do that too!"
<zimbatm>
I am quite confident that with a bit more practice it will become easier
<Ericson2314>
Thanks :)
janneke has quit [Ping timeout: 240 seconds]
FRidh has quit [Quit: Konversation terminated!]
ris has joined #nixos-dev
psyanticy has quit [Quit: Connection closed for inactivity]
<worldofpeace>
niksnut: could we talk about that again?
<worldofpeace>
It's pretty much the first step, moving iso generation out might be another. Though I'm not sure how this works, will the nixpkgs rev used in that jobset be revs that passed in the tested one?
asymmetric has joined #nixos-dev
asymmetric has quit [Quit: Peace.]
asymmetric has joined #nixos-dev
drakonis has quit [Quit: WeeChat 2.6]
drakonis has joined #nixos-dev
drakonis_ has joined #nixos-dev
drakonis has quit [Ping timeout: 268 seconds]
asymmetric has quit [Quit: Peace.]
asymmetric_ has joined #nixos-dev
asymmetric_ is now known as asymmetric
drakonis_ has quit [Read error: Connection reset by peer]
janneke has joined #nixos-dev
justanotheruser has quit [Ping timeout: 240 seconds]