[Gllug] Building a patch

[adrian at mcmen.demon.co.uk]

nix at esperi.org.uk wrote:
> On Sat, 03 Dec 2005, Adrian McMenamin prattled cheerily:
> > On Thu, 2005-12-01 at 11:03 +0000, Russell Howe wrote:
> >> On Thu, Dec 01, 2005 at 10:25:40AM -0000, Adrian McMenamin wrote:
> >> > I avoided using the alsa CVS for this because they had half implemented a
> >> > patch which didn't fully work and I wanted to avoid mixing up two
> >> > different ways of resolving this problem :(
> >> 
> >> Can you check out a revision of the code which doesn't have that
> >> half-patch?
> >> 
> >> With CVS you should be able to go 'back in time' and get an old copy of
> >> the tree.
> > 
> > Actually the CVS code just wouldn't work for me. I have submitted a hand
> > build patch - lots of patches joined together with cat - I hope the
> > maintainers accept it (and sooner or later every mainstream distro on
> > the planet will be built using my patch - whoo hoo)
> 
> Um, why doesn't `diff -urN' work for you?
> 
> It's just that, um, that or `cvs/svn diff' work for virtually everyone
> else: it's strange that *both* raw diff *and* your version-control
> system couldn't be made to behave.
> 
> (Personally diff has never misbehaved for me at all: the -x and -X
> arguments make it easy to avoid junk in the tree, but really there
> should be no junk there in any case: you should write the change in
> a pristine tree, and then test it in an emphemeral copy of that tree,
> so only that copy gets the build-time junk.)

The diff was probably my fault - trying to do this late at night whilst very tired. I needed to investigate further but was too tired to do so.

The CVS was a different problem: someone not using a cross compiling environment had in good faith tried to patch the sources but it just didn't work. Given I had a working set of patches I decided not to reinvent the wheel.

-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug