Arch Linux

assigning to some more developers. maybe we should revert that patch?

see http://repos.archlinux.org/viewvc.cgi/core/base/grub/PKGBUILD?root=core&r1=1.39&r2=1.40 and http://bugs.archlinux.org/task/9864

I could confirm this.
As i'm was the "guy with the gpt patch" i would meantime prefer to remove this patch. If one needs GPT support in grub it's easier to build a own grub (or we put a package in AUR). Also my approach to Arch and patching have changed, maybe this report and the discussion would be a good argument against patching in core packages.

One thing of interesst: Could someone with a gpt-patched grub on another distro (debian,gentoo) also confirm this on their patched grub? I have no access to such a machine currently, so test could be: set partition type of your swap partition
to swap (which was no change, but test the command):
parttype (hd0,0) 0x82
were hd0,0 must reflect the swap partition of YOUR installation.

I wan't to push this report again. On archlinux.de we have another problem resulting from the gpt-patch where hide/unhide of windows partitions get failed with this error. The OP says, that neither Ubuntu or gentoo have this problem (and these distris have a gpt patched grub).

So we should look what else must be patched to get back the grub base functions (and gpt). Or we revert the patch, have a full functionally grub again, and if someone needs gpt or other patches in Grub he should do this himselve (or via a AUR grub paket).
I would assume to remove the gpt patch ASAP, it breaks the bootloader.

Ok, i've found the reason what's wrong with the gpt patch file. The author has made syntax errors: he patched the code for some sources so that a if-construct follows directly after a while, before the opening { on this while. So the while-block was never checked.

There was a comment from Robin Johnson in this thread http://bugs.gentoo.org/show_bug.cgi?id=211584 which takes me on the path. Im not a c developer but after some hours staring on the patch code an some cups of strong coffee i saw the mistake ;-)

I attach a patch which fixes these. It's a patch against the patched code when grub-0.97-gpt.patch has beed applied. So my fix should run after this patch. Cause this is patching a patch i think it's not the best way. But i'm not very firm with patches, so if one of you could perhaps correct the grub-0.97-gpt.patch-File direkt...
(Oh my english... May anyone understand what i mean with that.. ;-)

Thomas pointet me to the combinediff tool, so here is a better gpt patch, it should replace the original grub-0.97-gpt.patch.

Andy, the decision to either revert, or apply the fixed patch is left to you. I'd probably prefer removing the patch altogether, but i'm not strongly opinionated either way