Hi Jim, I have one patch. It fixes the bug which waste a lot of memory in dfamust() for a very large pattern. Now, I will send it.
> There's also Bug#16586 which is marked as important but which I can't > reproduce. It reproduces as following. /usr/bin/printf "\xe9\x65\n\xab\n" | env LC_ALL=en_US.utf8 grep -P '.e|.?z' | head However, I seem that it's a bug of PCRE library. PCRE library may assume that a text doesn't have invalid sequence in UTF-8. Norihiro
