[PATCH] Final removal of lp_posix_pathnames() from the smbd server main code paths.

[Jeremy Allison]

On Wed, Mar 23, 2016 at 08:09:05PM +0100, Ralph Boehme wrote:
> On Wed, Mar 23, 2016 at 12:04:47PM -0700, Jeremy Allison wrote:
> > On Wed, Mar 23, 2016 at 07:56:20PM +0100, Ralph Boehme wrote:
> > > Hi,
> > > 
> > > On Wed, Mar 23, 2016 at 09:13:27AM -0700, Jeremy Allison wrote:
> > > > I know people are really busy with other More.Important.Things. :-(
> > > > right now.
> > > 
> > > I still have this on my todo list as well.
> > > 
> > > > But for those who aren't, it'd be really great to get this in
> > > > as then I can start prototyping the SMB2 unix extensions !
> > > 
> > > fwiw, in case those slipped your attention, there's some prior art
> > > from Volker and Richard:
> > > 
> > > <https://git.samba.org/?p=slow/samba.git;a=commitdiff;h=d786c5b6bdd7a44ca7a439c089862c158a0a5d70>
> > > <https://git.samba.org/?p=slow/samba.git;a=commitdiff;h=efbd4981533ff3352e14fedc9c57e79fc977fdd5>
> > > 
> > > I wanted to start work on this in the next weeks as well. :)
> > 
> > No, I hadn't forgotten. The whole point of this epic
> > patchset was to be able to use that work without
> > running into lp_posix_pathnames() all over the damn
> > place :-).
> > 
> > Personally, I think less-is-more in terms of option
> > negotiations. If you SMB2_CREATE with the POSIX
> > create context and you get a reply, the server is
> > telling you all the POSIX features it has helpfully
> > turned on for your benefit :-).
> > 
> > I don't think the client should get to choose.
> > 
> > That way lies combinatorial madness in testing :-).
> 
> agreed -- if that includes posix rename behaviour :-)))

Oh sure - that means everything POSIX, including changing
the meaning of doing SMB2_LOCK/UNLOCK to match posix etc.
etc.