Closed GaetanLepage closed 1 month ago
Changing report.json will need changes in nipkgs-update as well.
From the code it looks like you handle local evaluation as well, no?
From the code it looks like you handle local evaluation as well, no?
No, the build_commit
method of Review
is broken and has not been adapted.
I have a sequential version working for --eval local
.
Basically, it calls nix-env
for each system, one after the other.
We could work on calling them all in parallel, but we would need to provide an option as having 4 of them simultaneously requires a lot of RAM (sometimes even 64GB is not enough).
nixpkgs-review
resultGenerated using nixpkgs-review
.
Command: nixpkgs-review pr 342734
x86_64-linux
aarch64-linux
x86_64-darwin
aarch64-darwin
This PR addresses #400 by allowing to evaluate a PR/revision for multiple systems at once. You can now provide the
--systems
option with a list of systems to evaluate:--systems current
(default): legacy behavior--systems "x86_64-linux aarch64-darwin"
: evaluate for both intel linux and ARM darwin--systems all
: evaluate on all four main platformsProgress
report.json
,report.md
, stdout...)--eval ofborg
--eval local
(sequential)Support for-> Let's do this in a following PR. This one is already big enough.--eval local
(optionally parallel)I am eager to receive some feedback !
Technical details
dict
mapping eachsystem
to this list.--system
option has been softly deprecated (warning + forwarding the value to--systems
) to avoid breaking the workflow of existing users.report.json
.cc @Mic92 @natsukium