https://bugs.exim.org/show_bug.cgi?id=1642
--- Comment #11 from Philip Hazel <ph10@???> ---
Sorry, people. I have twice replied to this thread by email, but the postings
don't seem to have made it into Bugzilla.
Compiling with -O0 on my 64-bit Arch Linux system gives no problem with the
default 8MB stack, even if I add -fsanitize=address (which, according to the
comment in RunTest, is why that script ups the stack to 16MB). Maybe that was
before the improvement we made to reduce stack usage.
Nobody else has complained about this issue, either on PCRE1 or PCRE2. Maybe
they always use -O3 or are using different compilers?
I'm afraid I cannot accept your patch because it contains GNU-isms, I think.
Also, I'm sure many OS (non-Unix-like) do not have a compatible ulimit command,
even if they have a Unix-like shell.
Petr's suggestion about checking memory at runtime also suffers from a lack of
a standardized way of finding out how much is available. There isn't a
standardized way of catching a "run out of stack" error either, as far as I
know.
--
You are receiving this mail because:
You are on the CC list for the bug.
