Improvements to the dead code lint #17410
Conversation
|
The main change here is that rustc will now warn about variants that are never constructed. I concluded this would be a worthwhile lint after finding that is never constructed, yet there is plenty of code throughout rustc that deals with it. Which could be a bug but I haven't investigated yet.Funnily enough, with this change rustc still doesn't warn about that particular case due to the Clone trait being derived on the In addition, I changed the messages to be more descriptive with regards to the particular nodes that are proven unused. Last, but not least, I also investigated how much dead code there is in rustc after disabling re-exports. Turns out, quite a lot. I carefully removed what I thought was truly dead code but I still question if it's something we should do. |
| ExponentFormat, ExpNone, ExpDec, ExpBin, | ||
| SignFormat, SignNone, SignNeg, SignAll, | ||
| SignificantDigits, DigAll, DigMax, DigExact | ||
| }; |
There was a problem hiding this comment.
This module is still currently one I'd like to leave as a private implementation detail as part of libcore, was there a reason for exporting these variants?
I do feel bad for duplicating the parsing between std/core!
There was a problem hiding this comment.
Ah, I just noticed the duplication between libcore and libstd.
That is fair enough. Not all of the variants in libcore are actually being used so I can remove the unused ones or annotate them with #[allow(dead_code)].
There was a problem hiding this comment.
@alexcrichton I brought back the libcore versions of the types without the unused variants, if that's OK.
|
@alexcrichton Fixed the Windows issue. r? |
ghost
deleted the
This implements a wish suggested in rust-lang#17410, detecting enum variants that are never constructed, even in the presence of `#[derive(Clone)]`. The implementation is general and not specific to `#[derive(Clone)]`. r? @jakub-
No description provided.