[PATCH] D112105: [ADT] Simplifying hex string parsing so it runs faster in debug modes.

[Stella Laurenzo via Phabricator via llvm-commits]

stellaraccident accepted this revision.
stellaraccident added a comment.
This revision is now accepted and ready to land.

In D112105#3076917 <https://reviews.llvm.org/D112105#3076917>, @dblaikie wrote:

> We don't usually optimize for debug performance - is there a particular case where this comes up that's worth prioritizing/changing code for (& is there a good heuristic we can use about which debug performance is going to matter and which isn't?)

Interested who the "we" is. Debug builds should be generally usable and this case was extreme, pushing things into unusable for very common programs.

I'm +1 on accepting patches that improve such gradients -- if the patch was important enough for a busy engineer to take their time to fix in order to improve qol for all of us, I'll accept that - and the reported gradient is extreme. The code as written is not surprising to me for a foundational, hot item like this.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D112105/new/

https://reviews.llvm.org/D112105