[PATCH] Updated spec file for RPM

[Andrew Bartlett]

Gerald Carter wrote:
> 
> On Fri, 9 Feb 2001, Andrew Bartlett wrote:
> 
> > I have made a few changes to the RedHat RPM spec file for Samba 2.2,
> > allowing a functional install of samba via RPM.  It rebuilds on RH6.2
> > and RH7+rawhide except in that the makerpms script doesn't work on
> > RPMv4.
> >
> > I've had this sitting around waiting to be dealt with for a while, so
> > something recent might have broken it, but in any case its better than
> > what's currently in CVS.
> >
> > Quirks:  I have removed the second copy of the samba book, pointing swat
> > (which I don't use and haven't tested) to the RH7 documentation
> > directory.  I have also moved code page generation into the RPM, thus
> > getting rid of the code-page source from the final distribution (idea
> > from what RH does on their own spec file).
> 
> I don't know if this ever got applied.  I'll check on Monday and
> if not will deal with it then.  Thanks for the help Andrew.  Did you ever
> get any feedback on your PAM patch?  If not I'll look into that as well.
> 
> CHeers, jerry

No, it didn't get applied.  I have since made some further
modifications, attempting to use the --with-fhs option.  A decision
needs to be made as to if we are going to use the full FHS, or to keep
things like smb.conf/smbpasswd in their present location (/etc).  (I'm
not even sure if we can keep them in /etc, the --with-fhs option seems
to override --sysconfdir and --with-private-dir.)

I'll give these a test, and submit a patch.

(The other thing to look at it if to use the new RPMv4 macros,
particularly as RPMv4 has been released for all platforms.  )

As regards the PAM patch, I talked to Tridge at a recent CLUG, and while
he agreed with the account checking (for encrypted passwords, giving us
account expiration), I am yet to convince him about session support.  He
did think that maybe it could be moved into the authentication context,
where we are actually allowed to a 'sorry, permission denied'.  (Or at
least that's my recollection, but it was getting pretty late...).

My PAM patch also adjusts the utmp not to record entries for IPC$
connections, but all connections (including IPC$ sessions) are put
through PAM.  This allows pam_limits to do its magic, as well as
allowing interesting things like PAM based utmp-like accounting. 

The PAM patch works, as tested by the staff and students at Hawker
College :-).

Tridge also commented about the code, apparently its pretty obvious
exactly how long I have been coding C... :-). 

Thanks,
Andrew Bartlett
-- 
Andrew Bartlett
abartlet at pcug.org.au