On 6/2/07, Peter Valchev <[EMAIL PROTECTED]> wrote:
> > Here is an update to my first attempt to create a port for last
> > boehm-gc. This one works well with inkscape and w3m on i386.
>
> boehm-gc is difficult, it needs someone with access to multiple
> machine architectures and a fair understanding of OS internals to get
> it configured properly.
>
> it should probably be ONLY_FOR_ARCHS which have been specifically
> tested and known to work. I'm not sure the in-tree port works quite
> right on many !i386 but it gets further than this update, which
> fails to build on amd64:
>
> os_dep.c: In function `GC_get_stack_base':
> os_dep.c:1095: error: `result' undeclared (first use in this function)
> os_dep.c:1095: error: (Each undeclared identifier is reported only once
> os_dep.c:1095: error: for each function it appears in.)
> os_dep.c: In function `GC_register_data_segments':
> os_dep.c:1483: error: `DATASTART' undeclared (first use in this function)
>
> (this update is also not made on -current, e.g. distinfo is still
> old-style) - I didn't try any slow arch yet.
A good way to start would be to look at every one of the patches in the
existing boehm-gc port, and applying them to the new version by hand.
The current port works on amd64, and you have broken that.. as well as
many other things, I'm sure, so follow the old patches.
This exactly what I did. Of course it would be too easy if it worked
like this. So i take old patch then change what need to be changed and
so on. At least there is for now a current working version for i386.
Currently working on amd64.
--
bchesneau.info | neurofriends.net | osbud.net
