Re: Fix for grub_assert_fail undefined on NetBSD and other platforms

[Felix Zielcke]

Am Donnerstag, den 24.12.2009, 22:55 +0100 schrieb Robert Millan:
> On Tue, Dec 22, 2009 at 09:39:07PM +0530, BVK Chaitanya wrote:
> > Hi
> > 
> > 
> > Attached is the patch, which removes use of undefined
> grub_assert_fail
> > function for catching bad-type-cast errors, with a better version
> > __attribute__((error("msg"))) gcc extension.  With this extension,
> gcc
> > can give the exact location of the bad type cast at compile time.
> 
> Is this really a kind of error we'd like to report at run time?  Sorry
> if
> I'm missing something, but if we need additional code to handle it,
> and it
> was known at compile time, why do we do this?

__attribute__ ((error)) still reports it at compile time just like the
old grub_assert_fail method.
But the advantage is that you can specify the error message instead of
just getting a `ld: unknown symbol grub_assert_fail' error during
linking. And as BVK said above also the exact location where this
happened.

-- 
Felix Zielcke
Proud Debian Maintainer and GNU GRUB developer