CVS update: samba/source

[Andrew Bartlett]

John H Terpstra wrote:
> 
> On Sun, 14 Oct 2001, Jeremy Allison wrote:
> 
> > On Sun, Oct 14, 2001 at 06:27:39PM +1000, Andrew Bartlett wrote:
> > >
> > > I have a workaround for the package build (and I'm about to upload the
> > > RedHat 6.2 package it generated).  I'll commit that, but I'm not entirly
> > > sure exactly what the original commit intended so I'll leave that if you
> > > don't mind.
> >
> > I'm not sure thats a good idea. If the Makefile.in is incorrect I'd
> > rather you fix that than put a hack in the spec file.
> >
> > Also, I always build a binary package from the *released* tarball,
> > not out of any CVS tree.
> >
> > This is to ensure that anyone rebuilding the package on the platform
> > gets *exactly* the same as the code we released - that's *REALLY*
> > important.
> >
> > Please do this for the RH6.2 binary, not from a CVS tree.
> 
> I agree totally with Jeremy on this point. The key to our problem is
> deeper than just this one incident - we must unify our Linux build
> methods. Having so many separate SPEC files and patches to validate each
> time is insane. 

The RedHat SPEC file applies NO patches, btw.

> In short, I believe that the makefile should ALWAYS build
> a product that is useable. Codepages in the absence of binaries is broken.
> Think about it for a moment - how do you build the codepages from the
> source if the binaries are not installed?

But binaries without codepages is also broken, which one do you want? 
Make won't let you do both (circular dependency).  I think neither. 
'make install' should install a workable distribution, the others should
do *exactly* what they say, and nothing more.

Andrew Bartlett

-- 
Andrew Bartlett                                 abartlet at pcug.org.au
Samba Team member, Build Farm maintainer        abartlet at samba.org
Student Network Administrator, Hawker College   abartlet at hawkerc.net
http://samba.org     http://build.samba.org     http://hawkerc.net