[Python-Dev] New SRE core dump (was: SRE 0.9.8 benchmarks) (original) (raw)
Guido van Rossum guido@beopen.com
Thu, 03 Aug 2000 08:31:16 -0500
- Previous message: [Python-Dev] SRE 0.9.8 benchmarks
- Next message: [Python-Dev] New SRE core dump (was: SRE 0.9.8 benchmarks)
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
andrew wrote:
> Hmm... here's an old problem that's returned (recursion on repeated > group matches, I expect): > > >>> p=re.compile('(x)*') > >>> p > <SREPattern object at 0x8127048> > >>> p.match(500000*'x') > Segmentation fault (core dumped)
Effbot:
fwiw, that pattern isn't portable:
Who cares -- it shouldn't dump core!
...but sure, I will fix that in 0.9.9 (SRE, not Python -- Christian has already taken care of the other one ;-). but 0.9.9 won't be out before the 1.6b1 release...
I assume you are planning to put the backtracking stack back in, as you mentioned in the checkin message?
(and to avoid scaring the hell out of the beta testers, it's probably better to leave the test out of the regression suite until the bug is fixed...)
Even better, is it possible to put a limit on the recursion level before 1.6b1 is released (tomorrow if we get final agreement on the license) so at least it won't dump core? Otherwise you'll get reports of this from people who write this by accident...
--Guido van Rossum (home page: http://www.pythonlabs.com/~guido/)
- Previous message: [Python-Dev] SRE 0.9.8 benchmarks
- Next message: [Python-Dev] New SRE core dump (was: SRE 0.9.8 benchmarks)
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]