Every time I hack in rustc I think I must be doing something wrong to make my feedback loops so slow.
To verify I tweaked some stable attributes right in https://github.com/rust-lang/rust/pull/56998 I ran
./x.py test --doc src/liballoc which completed in 0:21:43. Does that sound reasonable?
that tracks with what i've experienced - you may want to try adding
--stage 1 to your call to cut out one layer of compiler builds
@Dale Wijnand I will also put a
-i in there so that it builds incrementally.
--keep-stage 1 also helps if you've only made changes to tests for example
@Dale Wijnand As a rule of thumb, after making some changes, I would just run
./x.py check, so that every thing type-checks without wasting much time, sort of early feedback. Then I would build the compiler upto stage 1 and build libstd with this new compiler.
./x.py build -i --stage 1 src/libstd
But as @centril points out, if I am just modifying the tests, I will keep stage 1 compiler.
Thank you, all.
I'll make to use
-i, I was operating under the assumption that it had stabilised and was on by default (and the docs were just stale).
I'll keep learning how the various stages work so I know what to run.
Does x.py check vs build/test suffer from https://github.com/rust-lang/cargo/issues/3501?
there's an entry in config.toml to turn on incremental all the time without the