From bug 43058 comment 4, in regard to the random delay chosen before crashing when WEBKIT2_CRASH_WEB_PROCESS_RANDOMLY is set: > It seems to me that a reproducible pseudo-random sequence is better than a higher quality random number for this purpose. This bug represents implementing this.
<rdar://problem/8407376>
I guess pseudo-random is only part of the story. We’d also somehow want to synchronize with the amount of CPU time used instead of using wall clock time or something like that. That may be impractical.