https://bugs.exim.org/show_bug.cgi?id=1887
--- Comment #2 from Philip Hazel <ph10@???> ---
After several days playing around with different ideas in the code, I'm afraid
I have to admit failure for the moment. The code has been hacked around for
nearly 20 years, starting from a much simpler type of regex matching. It has
now so many special cases and optimization short cuts that it is very difficult
to understand and modify at this kind of level.
I have decided, therefore, that the long-term solution is to re-implement the
matching function, using a new algorithm that solves this bug and perhaps can
also implement non-atomic recursions (something else that has been asked for).
However: (a) This won't happen for a while, and (b) I will only be thinking of
this big project in the context of PCRE2 (the 10.xx series), not PCRE1. (I
recently completed a big re-factoring of pcre2_compile() for similar reasons.)
In the meantime, I'll put a note in the documentation.
--
You are receiving this mail because:
You are on the CC list for the bug.
