Brian Mancuso writes: > > Below is a patch that decouples the match variables from the > op tree and places them in the thread structure. This problem is not thread-specific. I do not see how moving things into the thread-local storage will help. > - (s = rx->startp[paren]) != -1 && > - (t = rx->endp[paren]) != -1) > + (s = get_match_node(rx)->startp[paren]) != -1 && > + (t = get_match_node(rx)->endp[paren]) != -1) Nope. You do not want to make function calls in this tight code. Did not look any further. IlyaThread Previous | Thread Next