jakarta-oro-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Daniel F. Savarese" <...@savarese.org>
Subject Re: Jakarta-ORO experiments in PERL5 and REGEXs.
Date Sat, 22 Nov 2003 05:00:13 GMT

In message <93E75FCE8083D411A93500D0B78EC684622447@cissvmail1.sv.counterpane.co
m>, Nathan Fairchild writes:
>I realize that these are NOT well written regular expressions, but this
>whole experiment came about by having some hand me back my Java classes,
>saying that my code hangs.  I debugged down to the Perl5Matcher.match()
>call in my code, and then wrote my test program.  When stuff started
>blowing up, I started looking through the oro-dev archives, and I
>thought I would contribute to the "DON'T DO THIS" list.

This is great.  Is there any chance you can be persuaded into turning
your tests into a generic junit test that you can contribute for use
regression testing as we move forward to improve the handling of these
cases?  The reality is that even though one has to be careful
when writing regular expressions, we should try to handle the "bad
ones" more efficiently.

For what it's worth, a spamassassin pattern I wrote that worked great
for months recently met a pathological email that sent Perl 5.8 into
infinite backtracking.  So even Perl 5.8 is sensitive to the structure
of some patterns.


To unsubscribe, e-mail: oro-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: oro-dev-help@jakarta.apache.org

View raw message