[ On Tuesday, March 2, 2004 at 23:21:34 (+0900), Bang Jun-Young wrote: ]
> Subject: Re: GCC 3.3.3 upgrade appears done
>
> BTW, 'cvs diff -rFSF' against gcc doesn't give expected results. I think 
> vendor branch is a nice feature of cvs, so wouldn't there be an easy 
> solution without importing the whole 3.3.3 again?

By design CVS cannot really ever give logical or expected results in the
scenario where the magic vendor branch feature (i.e. "cvs import") is
used along with normal CVS branches, particularly in the way the main
NetBSD source tree uses normal CVS branches.

This is caused by the "lazy" nature of normal CVS branches conflicting
with the way the default branch is left on the magic vendor branch for
modules which were created with "cvs import".

The only time this will ever work exactly as expected, especially when
diffing any normal branch against the vendor branch, is when all files
have been locally modified, but of course making unnecessary local
modifications creates its own headaches.

-- 
						Greg A. Woods

+1 416 218-0098                  VE3TCP            RoboHack <woods@robohack.ca>
Planix, Inc. <woods@planix.com>          Secrets of the Weird <woods@weird.com>