OK, I'm putting together a test case now. It's turning out more complicated than I thought, but I'll post it here when I get it working.
I profiled this with NYTProf and it is indeed the pos()= pos() statement that is taking all the extra time. However, the exact conditions that trigger it are currently eluding me; I'll continue to isolate it.
In reply to Re^2: Is there a way to allow consecutive zero-length matches without using pos()?
by jsm
in thread Is there a way to allow consecutive zero-length matches without using pos()?
by jsm
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |