Tweak short_ty_string to reduce number of files#118389
Conversation
|
r? @b-naber (rustbot has picked a reviewer for you, use r? to override) |
| "the trait `{}` is not implemented for `{}`", | ||
| old_pred.print_modifiers_and_trait_path(), | ||
| old_pred.self_ty().skip_binder(), | ||
| self.tcx.short_ty_string(old_pred.self_ty().skip_binder(), &mut file), |
There was a problem hiding this comment.
Why introduce short_ty_string here? I'm not sure that the lifetime bounds we get with this change are really preferable.
There was a problem hiding this comment.
Although I agree that the prior output is likely "better", if we let the full type leak anywhere we have the real risk of ending up with a "wall of text" in some case. We could as a follow up change it so short_ty_string has the same output we had prior instead.
There was a problem hiding this comment.
But did you really handle all occurrences where types are printed in diagnostics in this PR? I would have expected many more changes in the diffs in the tests if that was the case. And if you haven't, I find this change kind of arbitrary.
There was a problem hiding this comment.
@b-naber I addressed all that I could find searching the codebase. There are many that we had already handled and this was an effort to find any stragglers.
2a5c5eb to
5fb1782
Compare
|
Thanks for clarifying. This looks good to me. r=me |
When shortening types and writing them to disk, make `short_ty_string` capable of reusing the same file, instead of writing a file per shortened type.
5fb1782 to
9d846fc
Compare
|
@bors r=b-naber |
|
☀️ Test successful - checks-actions |
|
Finished benchmarking commit (5f73b00): comparison URL. Overall result: no relevant changes - no action needed@rustbot label: -perf-regression Instruction countThis benchmark run did not return any relevant results for this metric. Max RSS (memory usage)ResultsThis is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.
CyclesResultsThis is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.
Binary sizeThis benchmark run did not return any relevant results for this metric. Bootstrap: 671.988s -> 673.133s (0.17%) |
When shortening types and writing them to disk, make
short_ty_stringcapable of reusing the same file, instead of writing a file per shortened type.