A useful VFS change ...

[Dustin Oprea]

On Fri, Mar 14, 2014 at 11:44 AM, Volker Lendecke <Volker.Lendecke at sernet.de
> wrote:

> On Fri, Mar 14, 2014 at 11:28:59AM -0400, Dustin Oprea wrote:
> > Think about FUSE... If a feature isn't supported, it gracefully handles
> it
> > without panicking. Consequently, you don't have to implement all 30 (or
> > 40?, 50?, ...) functions in order to get going. This obviously makes it
> > easier to iteratively develop and debug.
>
> Ok, understood. You want Samba to somehow limp along even
> with the opaque module returning ENOSYS everywhere. This is
> certainly a much larger task. But it's a very worthwhile
> one. So we need to just start running in this mode and
> eliminate panics one after the other...
>
> > I've attached the patch for the example.
>
> I'm with you on the naming thing, I'm not sure how we are
> supposed to handle the init_samba_module thingy. There's
> some macro magic for builtin vs shared modules around. Can
> you post them as two separate git format-patch patches?
>
> Thanks,
>
> Volker
>
>
It looks like they've already been changed:

9ca88eb1 (Stefan Metzmacher 2011-10-08 09:17:11 +0200 938) NTSTATUS
vfs_skel_opaque_init(void)
bee3d8f4 (Alexander Bokovoy 2003-06-04 13:13:41 +0000 939) {
a581b3d4 (Andrew Bartlett   2012-10-11 07:53:29 +1100 940)      return
smb_register_vfs(SMB_VFS_INTERFACE_VERSION, "skel_opaque",
a581b3d4 (Andrew Bartlett   2012-10-11 07:53:29 +1100 941)
            &skel_opaque_fns);
bee3d8f4 (Alexander Bokovoy 2003-06-04 13:13:41 +0000 942) }

I don't know if Stefan's function name works (I haven't tested it), but
Andrew made the change to the VFS name.


Dustin