https://github.com/timon-ul requested changes to this pull request.
Without really looking at the code and just trying the feature out locally I
have some thoughts. First off I wonder if we shouldn't mirror the behaviour we
have for constructors today, this means that when I request on `make_unique` I
still end up at the definition of `make_unique`, but if I send the request on
top of the opening brace I end up on the actual constructor. If you are
unfamiliar with what I am talking about, here an example (using our testing
syntax):
```
struct [[Abc]] {
Abc() {};
}
auto a = A^bc();
```
The code above leads me to the type itself, meanwhile the code below leads me
to the actual constructor.
```
struct Abc {
[[Abc]]() {};
}
auto a = Abc^();
```
I know this is a bit unintuitive and I am not so sure how many people generally
even know this exists. But as long as constructors behave that way I think
staying consistent with this behaviour is what I would prefer. Mainly it means
we do not trip up users who like todays behaviour, we do not trip up users by
inconsistent behaviour and also (at least when using vscode) we do not get this
multiple response view that at least takes me out of the navigation flow, but
instead we have clear behaviour when we end up where.
https://github.com/llvm/llvm-project/pull/199480
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
