[clang-tidy] matchers::matchesAnyListedName() confusingly uses regular expression searches in bugprone-use-after-move

[higher-performance]

::free should not match the name of ::freedom::Id* S::blah(); (!)

https://godbolt.org/z/vEcY74h8M

namespace freedom {
class Id;
struct S { Id* blah(); };
}
void foo(freedom::S* s) {
  s->blah();
  s->blah();
}

$ clang-tidy --checks='bugprone-use-after-move' --config="{CheckOptions: {bugprone-use-after-move.InvalidationFunctions: '::free'}}"

<source>:7:3: warning: 's' used after it was invalidated [bugprone-use-after-move]
    7 |   s->blah();
      |   ^
<source>:6:6: note: invalidation occurred here
    6 |   s->blah();
      |      ^

Despite its name, apparently matchesAnyListedName treats its arguments as regexes rather than names, causing this issue.

[llvmbot]

@llvm/issue-subscribers-clang-tidy

Author: None (higher-performance)

Details

`::free` should ***not*** match the name of `::freedom::Id* S::blah();` (!)

https://godbolt.org/z/vEcY74h8M

namespace freedom {
class Id;
struct S { Id* blah(); };
}
void foo(freedom::S* s) {
  s->blah();
  s->blah();
}

$ clang-tidy --checks='bugprone-use-after-move' --config="{CheckOptions: {bugprone-use-after-move.InvalidationFunctions: '::free'}}"

<source>:7:3: warning: 's' used after it was invalidated [bugprone-use-after-move]
    7 |   s->blah();
      |   ^
<source>:6:6: note: invalidation occurred here
    6 |   s->blah();
      |      ^

[vbvictor]

Agree that we should rename it to matchesAnyListedRegexName.
Deleting regex at all seems not right now because it will break a lot of current configs.

[higher-performance]

I agree renaming it would be better than keeping it as-is, but I'm wondering how many users of the function realize it uses regexes at all. Even just looking at bugprone-use-after-move, I'm struggling to tell if it should be using regex matching vs. hasAnyName. Any thoughts?

[vbvictor]

I agree renaming it would be better than keeping it as-is, but I'm wondering how many users of the function realize it uses regexes at all. Even just looking at bugprone-use-after-move, I'm struggling to tell if it should be using regex matching vs. hasAnyName. Any thoughts?

There is definitely some confusion aboit regex vs plain name, and there was recent effor to correct docs to cover it, #161142, #160991 (comment).

Thank you for code cleanup; a good next step would be to adjust user-docs (if any left).