mrustc: init at 0.9 #113817
mrustc: init at 0.9 #113817
Conversation
ofborg
bot
added
8.has: package (new)
|
Please squash all package related commits together. |
| cp run_rustc/output/prefix/bin/cargo $out/bin/cargo | ||
| cp run_rustc/output/prefix/bin/rustc_binary $out/bin/rustc | ||
|
|
||
| cp -r run_rustc/output/prefix/lib/* $out/lib/ |
There was a problem hiding this comment.
Maybe we should move this to a separate output.
There was a problem hiding this comment.
It's definitely something we can look into, but I think for now I would prefer to just keep the bootstrap installation as simple as possible. I expect the final user-facing rustc derivations will disallow references to these bootstrapping stages anyway, so it shouldn't make any difference in terms of closure size whether this uses split outputs.
@SuperSandro2000 Is this all right with you? As a reviewer I sometimes like to see separation of concerns in commits, but I can just squash if it's not actually helpful. |
I bet 10 bucks on that I will probably forget it.
If you don't rebase with master on force pushes I can see from the force push diff what you did. I am also using the mark-files-as-read-feature. Either works for me but keeping review comments around takes another force push before it can be merged. |
Haha, fair enough :) |
|
@ofborg build mrustc-minicargo |
|
I started updating the nix expression to work with the current mrustc master, which supports rust 1.39: progval@07b59ad But I still can't use it to build even build rust 1.29 ( |
|
Does the build use many CPU cores if they're available, or is it a memory-bound operation? |
|
@grahamc It's partially parallel, but its bottleneck is by far the compilation of |
|
This is a semi-automatic executed nixpkgs-review with nixpkgs-review-checks extension. It is checked by a human on a best effort basis and does not build all packages (e.g. lumo, tensorflow or pytorch). Result of 3 packages built:
|
mrustc is mostly patched to use shared LLVM sources but still uses in-tree source for compiler-rt from LLVM 7. This needs to be patched to compile under glibc 2.31 or later. It's easy enough to reapply all our compiler-rt patches here.
|
@alyssais Thanks for the suggestions, updated. |
|
@ofborg build mrustc-minicargo
|
|
Now let's get an up to date one!
Feel free to request a review from me in any followups.
|
|
@r-burns jsyk, in future when you take a commit like that, you can mark
yourself as a co-author on that commit using Co-authored-by, so that
GitHub and other tools will recognise your work on it.
https://docs.github.com/en/github/committing-changes-to-your-project/creating-a-commit-with-multiple-authors
|
|
Yay, it built! https://hydra.nixos.org/build/141770406#tabs-summary |
This is rebased from the excellent work done by @progval in #85542, with the hopes that we can merge the underlying mrustc expressions and iterate on them while we sort out the bootstrapping logic.
Functional changes:
Cleanups:
The mrustc-bootstrap builds rustc and cargo using mrustc. This is the main
motivator for mrustc and is very finicky so it's important that we
build this regularly and keep it green.
@progval says:
Can I tell hydra that this build will need a lot of RAM?
Compilation might take several hours on less beefy machines;
I hope Hydra doesn't timeout on this.
Motivation for this change
Things done
sandboxinnix.confon non-NixOS linux)nix-shell -p nixpkgs-review --run "nixpkgs-review wip"./result/bin/)nix path-info -Sbefore and after)