Subject: Re: krb5 integration proposal
To: Aidan Cully <aidan@kublai.com>
From: Simon Burge <simonb@netbsd.org>
List: tech-net
Date: 06/09/1999 23:55:54
Aidan Cully wrote:

> On Wed, Jun 09, 1999 at 01:41:29PM +1000, Simon Burge wrote:
> > Aidan Cully wrote:
> > 
> > > On Wed, Jun 09, 1999 at 12:07:36PM +1000, Simon Burge wrote:
> > > > Aidan Cully wrote:
> > > > > 4) Reorganize libtelnet, according to my local model.
> > > > > 	This means the following hierarchy:
> > > > > 	src/lib/libtelnet -- almost all libtelnet sources.
> > > > > 	src/lib/libtelnet/Nocrypt -- libtelnet source that might be overridden
> > > > > 	    by crypto-enabled files.
> > > > 
> > > > Can this be done with the right PATH magic so we don't need the Nocrypt
> > > > subdirectory?  It's a tad non-aesthetic, as well as making imports of
> > > > new versions of libtelnet a little more work (if that were to ever
> > > > happen again)...
> > > 
> > > I do it this way for .PATH reasons..  krb5 comes with a complete libtelnet
> > > distribution, and domestic/lib/libtelnet reaches over into it..  One of my
> > > goals was to use as much NetBSD source as possible, and only hook into
> > > MIT's for kerberos specific stuff, and I couldn't see a good way to let
> > > domestic/lib/libtelnet use MIT's encrypt.[ch], and NetBSD's ~everything
> > > else without creating the Nocrypt/ subdirectory.
> > > 
> > > What sort of thing would you suggest?
> > 
> > I'm not sure I see the problem.  As an example, I've got the
> > following directory structure:
> ...
> > Is this more-or-less what we're after, or am I misunderstanding the
> > problem?
> 
> I believe you're misunderstanding the problem..  Add a file
> ./domestic/dist/telnet/misc.c
> and get the domestic/telnet/Makefile to use the misc.c in ./telnet
> instead.

Eek, so you want the misc.c from the non-domestic telnet and the
encrypt.[ch] from the domestic one, where there's a misc.c in domestic
and a encrypt.[ch] in the non-domestic tree?

Is there a functional difference between the source that's common to
both the domestic and non-domestic distributions - might it be just as
easy to "cvs delete" the domestic one in this case if you don't want to
use it?

Simon.