Port #[used] to new attribute parsing infrastructure
#142818
Conversation
|
|
rustbot
added
A-attributes
|
Some changes occurred in compiler/rustc_attr_parsing Some changes occurred in compiler/rustc_codegen_ssa Some changes occurred in compiler/rustc_codegen_ssa/src/codegen_attrs.rs Some changes occurred in compiler/rustc_attr_data_structures Some changes occurred in compiler/rustc_passes/src/check_attr.rs |
| | | ||
| LL | #[used] | ||
| | ^^^^^^^ | ||
| = warning: this was previously accepted by the compiler but is being phased out; it will become a hard error in a future release! |
There was a problem hiding this comment.
This warning is new, but I think this is reasonable in this context.
Is adding this warning acceptable?
| } | ||
|
|
||
| if let Some(prev) = &group.result { | ||
| cx.emit_lint( |
There was a problem hiding this comment.
some infra that makes this nicer merges with #142493 so let 's sequence this after that and rebase onto
rustbot
added
S-waiting-on-author
JonathanBrouwer
force-pushed
the
used_new_parser
branch
from
44fa5f0 to
073e2c6
Compare
| @@ -131,6 +131,17 @@ impl Deprecation { | |||
| } | |||
| } | |||
|
|
|||
| /// There are three valid forms of the attribute: | |||
| /// `#[used]`, which is semantically equivalent to `#[used(compiler)]` except that the latter is currently unstable. | |||
There was a problem hiding this comment.
#[used] is only guaranteed to be an alias for #[used(compiler)], but in practice it is an alias for #[used(linker)] nowadays.
There was a problem hiding this comment.
Hmmm indeed this is kinda weird, I did not notice that, thanks for letting me know!
#[used] conflicts with #[used(linker)] but not with #[used(compiler)], even though it is treated as an alias for #[used(linker)] in codegen
Is this the desired behaviour or shall I change this? @bjorn3
There was a problem hiding this comment.
I've pushed a change that makes #[used] conflict with #[used(compiler)], and no longer conflict with #[used(linker)]. I have no expertise in exactly what is good here but this seems like the best option.
(Also added a test for this since apparently there were none)
There was a problem hiding this comment.
I think either #[used], #[used(compiler)] and #[used(linker)] should all conflict with each other, or it should be a ratchet where #[used(linker)] wins over #[used(compiler)].
There was a problem hiding this comment.
I'll make them all conflict then, I think that's the safe option here. Will implement this later this weekend
There was a problem hiding this comment.
After some discussion here I'll implement the ratcheting
https://rust-lang.zulipchat.com/#narrow/channel/131828-t-compiler/topic/attribute.20conflicts/with/525158490
Signed-off-by: Jonathan Brouwer <jonathantbrouwer@gmail.com>
JonathanBrouwer
force-pushed
the
used_new_parser
branch
from
073e2c6 to
468dd68
Compare
|
@rustbot author |
|
Reminder, once the PR becomes ready for a review, use |
|
☔ The latest upstream changes (presumably #142826) made this pull request unmergeable. Please resolve the merge conflicts. |
Ports
usedto the new attribute parsing infrastructure for #131229 (comment)r? @jdonszelmann