Re: Mad idea: grub-fuse
On Mon, Sep 07, 2009 at 06:39:13PM +0800, Bean wrote:
> On Mon, Sep 7, 2009 at 6:18 PM, Colin Watson<firstname.lastname@example.org> wrote:
> > On Mon, Sep 07, 2009 at 06:07:48PM +0800, Bean wrote:
> >> Actually I've considered this before. My goal is to add a mini FUSE fs
> >> interface so that it can have read/write support in grub2 !
> > I'm actively uninterested in write support; in this case it would do
> > much more harm than good.
> It has some usage, actually both OpenFirmware and EFI have write
> driver, it can be used to edit boot config files. And we don't need to
> support all fs, for example, write support for fat is quite useful
> while don't require much effort to implement.
In the case of os-prober, write support is unnecessary and I do not want
it enabled. I don't mind what's done elsewhere, but for the use case of
os-prober it absolutely must be possible to turn off write support.
> > The level of functionality provided by the existing filesystem modules
> > would be quite sufficient.
> Oh I see, you mean the other way around, using the grub fs code to
> implement a FUSE driver. But I don't see why not just modify os-prober
> to work with grub-fstest, this is much simpler than writing a FUSE
The changes to do that in os-prober would be sufficiently intrusive that
they simply aren't going to happen (remember that os-prober needs to
work without GRUB too). A FUSE wrapper would make it possible to just
replace the mount step, which is much easier from the client side. (Yes,
I know it's some work in GRUB, which is why this thread is labelled "Mad
Colin Watson [email@example.com]