C++: Performance fix for existsCompleteWithName#1831
Conversation
36cee9f to
3700a63
Compare
|
Code change LGTM. |
|
@pavgust I've tried to investigate the cause of this performance regression in code that hasn't changed for a long time. I think the slowdown is a side effect of an optimiser improvement, possibly the fix for QL-878. Whatever it was, it caused a slight simplification of how the In conclusion, I think the optimiser is to blame for this fix being necessary, but I think the optimiser is doing a better job in 1.22 than in 1.21 even though random chance meant this PR went from being nice-to-have to being critical. For the record, this is the RA from 1.21: This is the RA from 1.22: |
|
(RA edited) |
No description provided.