Re: removing net/samba?

[David Brownlee <abs%absd.org@localhost>]

On Thu, 5 Oct 2023 at 09:40, Stephen Borrill <netbsd%precedence.co.uk@localhost> wrote:
>
> On Thu, 20 Jul 2023, Martin Husemann wrote:
> > On Thu, Jul 20, 2023 at 08:11:42AM -0400, Greg Troxel wrote:
> >> I suspect we'll end up with "no users in pkgsrc", and 8 years of no
> >> upstream maintenance certainly seems over the bar for no expectation of
> >> use.  But the Windows world is funny so we'll see if anybody speaks up.
> >> (I certainly don't use it.)
> >
> > Not sure this is worth much, besides anectodal value:
> >
> > I used to have it running on a tiny machine (single core armv5, 512 MB
> > RAM) acting as an (internal only) file server (for niche purposes,
> > access speed did not matter). The daemon processes in samba4 are a lot
> > more heavyweight (memory + cpu wise) and this tiny box had issues
> > running samba 4 (which I use on all "real" file servers when they
> > are not NFS-only).
> >
> > But that tiny hardware (unfortunately) has died and the replacement
> > is a full grown 8GB aarch64 system and of course has no problem running
> > samba 4.
>
> I never got Samba 4 to work as a domain member with mapping domain users
> to local users, i.e. If logged onto Windows as DOMAIN\myuser, you can
> access \\netbsdbox\myuser and the user id is as defined in /etc/passwd not
> determined through winbindd in a different range from local users. The
> Samba stance was that on Windows DOMAIN\myuser is not the same as
> LOCALPC\myuser (which is correct), so this was expected behaviour. The
> functionality was removed somewhere along the 4.x release branch.

I'd suggest re-importing net/samba as net/samba3 as there may be
"protected" networks and situations where it could still be useful,
but we should avoid people accidentally installing it when the look
for "samba"

Ah, and agree on updating all dependencies to samba4 where possible :-p

David