On 09.03.2017 18:11, ph10@??? wrote:
> Once this is released, the regular reports of "stack exceeded" bugs should go away. Yay!
I can confirm that starting with SVN 671, PCRE2 uses less stack.
Nevertheless, this test case from testinput6 still uses lots of stack (and exceeds it on a Windows machine):
#forbid_utf
#subject dfa
#newline_default lf anycrlf any
/(*LIMIT_DEPTH=100)^((.)(?1)|.)$/
\= Expect depth limit exceeded
a[00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]([00]
The stack overflow occurs during pcre2_dfa_match() using the 8-bit library.
Ralf
