Should we keep smbwrapper?
derrell at samba.org
Fri Feb 3 07:07:18 MST 2012
On Thu, Feb 2, 2012 at 23:25, Andrew Bartlett <abartlet at samba.org> wrote:
> What is the status of smbwrapper? I notice that you did a major rework
> on it in 2002. I've been unable to get it to work in 3.5 or even
> compile in master.
> I would also rather like to rid Samba of all the configure checks that
> this code entails, and an any case, it seems there are much better ways
> to create a virtual file system these days, with FUSE, gvfs or just
> using a cifs kernel mount.
> I'm also worried that it cannot be complete - it does not include the
> new dup3 for example, and probably many other calls. We have no
> testsuite for it either.
> Finally, I would like to see it removed so as to avoid giving others any
> strange ideas, as tridge's blog at http://blog.tridgell.net/?p=141 shows
> how, when this idea is taken to the extreme, this can go horribly,
> horribly wrong.
I was using smbwrapper in my company's product from 2002 until 2.5 years
ago. It worked flawlessly. At the time that I rewrote it, either FUSE
didn't exist yet, or I hadn't heard of it. FUSE is the more appropriate way
to do it. smbwrapper was "cool" in being able to do something that
shouldn't have been possible, but it always seemed hackish and brittle.
Although I'm much less active with this code than I used to be, I still
watch the samba-bugs mailing list and would have attempted to fix bugs in
smbwrapper if they were reported, but I haven't seen any come along. (I
haven't checked in bugzilla recently, though.) I will be very interested to
see if anyone else comments in this thread that they're using smbwrapper.
It's not out of the question that our code was the only user of it, and our
company is now defunct. Since there is now a good alternative to it in
FUSE, I'd have no objection to deprecating smbwrapper.
More information about the samba-technical