2009/7/18 Piotr Piastucki <leech miranda gmail com>: > I did a quick test (run meld filediff.py meldapp.py, add a single character) > and measured the time taken by _update_highlighting(): > meld 1.3 - 1160ms > my patch - 12 ms > Kai's patch - 52ms I'd be interested to know how you did these tests. I've had real problems getting reliable results when testing performance improvements. > No matter which patch is applied the performance gain is *huge*, however, > Kai's code looks much better :) Cheers. I've attached a second-pass patch with very simple cache eviction, and the results seem to be pretty promising. I'm sure it could be tweaked, but there's no doubt that it's a huge improvement. Kai
Attachment:
0001-Cache-inline-diff-results-to-avoid-rediffing-blocks.patch
Description: Binary data