search

idkwim / ouspg

Automatically exported from code.google.com/p/ouspg
3 stars 0 forks source link

radamsa mutation hang #86

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
Trunk version gets stuck using 100% CPU while producing no output using a 
particular sample set and seed. Fix before next release.

Info:
 node: Badger 0
 test run: radamsa-test-2, seeed 237683688921633105220569
 test time: ~749 hours

Original issue reported on code.google.com by aohelin on 31 Jul 2013 at 6:29

GoogleCodeExporter commented 9 years ago 
No time to fix yet, but it seems list fuser can go a bit too quadratic now that 
block size was increased (to allow learning mutations to get a better look at 
the data). In particular we are in a very slow run of try-choose at fuse.scm.

Original comment by aohelin on 2 Aug 2013 at 7:15

GoogleCodeExporter commented 9 years ago 
Fixed by bounding number of options after looking for a shared suffix. Almost 
always just one is used, but in case of equal data blocks where all suffixes 
survive the search and are equal, so the jump selection hits a O(n^3) case.

Original comment by aohelin on 6 Aug 2013 at 1:05