Should UTMP be compiled by default now?

David Lee t.d.lee at
Thu May 29 16:30:00 GMT 2003

On Sat, 24 May 2003 jra at wrote:

> On Sat, May 24, 2003 at 08:30:53PM +1000, Andrew Bartlett wrote:
> > I was wondering - should we enable the use of utmp as a 'stable' feature
> > for Samba 3.0, given it has been in Samba 2.2 with very few complaints
> > for quite some time now.
> >
> > When compiled in, the utmp code must still be activated with a global
> > 'utmp = yes' parameter.
> >
> > What do people think?
> Yes I don't see any harm in this. We should monitor the compile
> problems people have during the betas though.

For info to list folk: Andrew Bartlett has just (offlist) asked me (as
original author and coordinator of the utmp code) to look at this.

I have just sketched a possible patch to implement this, and am about to
email it to Andrew to discuss (there are a couple of interactions with his
"session" code).

But I would re-iterate Jeremy's point about monitoring compile problems.
In its early days the utmp code exposed an interesting array of
dependencies on utmp-offering OSes.  So while the utmp code seems to be
well tested on most of the major OSes represented here, it is possible
that there are a few OSes (that offer utmp) where buglets will emerge.

I'll try to keep an eye on the change once it enters SAMBA_3_0, and, it it
would help, am happy to try to coordinate any required subsequent fixes.

P.S. Could "--with-pam" also now be considered for "compile by default"?


:  David Lee                                I.T. Service          :
:  Systems Programmer                       Computer Centre       :
:                                           University of Durham  :
:            South Road            :
:                                           Durham                :
:  Phone: +44 191 334 2752                  U.K.                  :

More information about the samba-technical mailing list