[PATCH] D53482: Add clang-format stability check with FormatTests

[MyDeveloperDay via Phabricator via cfe-commits]

MyDeveloperDay added a comment.

In D53482#1737203 <https://reviews.llvm.org/D53482#1737203>, @aardappel wrote:

> @MyDeveloperDay thanks for your thoughts.. while `-stable` would be helpful once you know you have the issue (or to permanently turn on in a git pre-submit), it does not help developers that don't even know this is a thing, and waste a bunch of time figuring out what is going on. In fact, it took me longer to figure out simply because I assumed it is unlikely that `clang-format` has such a bug :)
>
> It might almost make sense in reverse: `-stable` by default, and have a `-un-stable` for those users where `clang-format` becomes a performance bottleneck :) But once you admit that, might as well fix the core problem instead.


Sorry I didn't make myself clear,I wasn't thinking of a final customer facing solution,  I was only really thinking of adding the -stable purely so we could pass a flag down and try different levels of repeating so we could do some performance tests. I feel it might be related to these 3 below, but was trying to think about how we could easily experiment.. I think ultimately we'd want a stable solution.

  alignConsecutiveDeclarations();
  alignConsecutiveAssignments();
  alignTrailingComments();


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D53482/new/

https://reviews.llvm.org/D53482