Re: Two Proposals (inter-related)

From: Ryan Pavlik <abiryan_at_ryand.net>
Date: Tue Jan 22 2008 - 13:59:46 CET

Luca Padovani wrote:
> On Jan 10, 2008 12:13 AM, Ryan Pavlik <abiryan@ryand.net> wrote:
>
>> 1. Math crashes anytime you try to use it.
>>
>
> Of course I can't be sure that this is related, but recently there
> have been reports of systematic crashes of gtkmathview even on Linux.
> After some investigation, it looks like there is a problem in recent
> versions of GCC (e.g. 4.2.1) that prevents some combination of
> templates+compiler optimizations to be compiled properly.
>
> Unfortunately the problem is very subtle to identify precisely. Gdb
> reports no apparent anomaly in the place of occurrence, valgrind is of
> no help because as soon as you compile gtkmathview in a large static
> library the bug goes away. ATM, the only workarounds I know of are
> 1) compiling gtkmathview statically
> 2) disabling optimizations -O0
> 3) disabling inlining -fno-inine
>
> If the problem you're reporting is indeed related to this, you might
> also consider one of these possibilities.
>
> --luca
>
>
Well, I now have GCC v.3.4.5 mingw version, which is what I used to
build GTKMathview, so I fear this route might not provide benefit.
Would there be any reason why disabling optimizations even with this GCC
version would help?

Ryan
Received on Tue Jan 22 14:04:08 2008

This archive was generated by hypermail 2.1.8 : Tue Jan 22 2008 - 14:04:08 CET