Don't allow duplicate attribute parsers

[JonathanBrouwer]

r? @jdonszelmann

Do you anticipate this being needed?

[rustbot]

jdonszelmann is currently at their maximum review capacity.
They may take a while to respond.

[rustbot]

Some changes occurred in compiler/rustc_attr_parsing

cc @jdonszelmann

[jdonszelmann]

yea, we're using this right now, at least also in my open PRs for doc and lint attrs. this is the more efficient way to let one attr parser influence other attribute parser. Much better than any strategy in finalize() that looks through the entire list again. In theory this pattern would be valuable to have more of, especially once we start validating.

[JonathanBrouwer]

@jdonszelmann Ah ok. Does that mean you're effectively parsing the same attribute twice? It would be nicer if one attribute can see the parsed form of another, or sth like that

[JonathanBrouwer]

But since you're using this, I'll close. Just thought I'd do this if there were no plans to use this