[nem-en] Random test.exe stack overflow

[Kamil Skalski]

positiva/matching.n crashes with stack overflow on MS.NET from time to
time and this is happens since a long time ago...
The workaround is to either:
- modify test.n to use the same stack workaround as ncc.exe
- find out which place in compiler's source code it crashes on and try
to change the too long recursion into loop
- modify the matching.n testcase to do not trigger the stack overflow

On 8/1/06, Michal Moskal <michal.moskal at gmail.com> wrote:
> On 7/31/06, Snaury <snaury at gmail.com> wrote:
> > Hi everyone,
> >
> > Yesterday it was second time that I got test.exe crash with
> > StackOverflow, both times while doing make check after doing make
> > clean, configure and make. Unfortunately both times it was with my
> > try/catch patches active, so I thought maybe I made a typo somewhere,
> > but so far I could find none. Even more unfortunately, I run make
> > check again and again and it doesn't crash! Both times (it seems) it
> > was while compiling positive/matching.n. I tried to reproduce this at
> > work, tried now at home, but just doesn't happen anymore.
> >
> > Any ideas why this could happen? And had anyone encountered same thing
> > on trunk sources?
>
> If you cannot reproduce this, then I guess we cannot do anything about it.
>
> The workaround is only in ncc.exe, therefore it is possible that
> test.n wouldn't have enough stack, especially on large example (with
> long functions) like the matching one. The MS.NET has ridiculously
> small limits (like one megabyte). If it doesn't happen anymore, you
> shouldn't be worried.
>
> >
> > P.S.
> > Just now a thought crossed my mind and I remembered a stack workaround
> > I've seen for ncc.exe, is it possible that test.exe suffers the same
> > problem under some rare conditions?
>
> --
>    Michał
>
> _______________________________________________
> https://nemerle.org/mailman/listinfo/devel-en
>
>
>


-- 
Kamil Skalski
http://nazgul.omega.pl