[Rt-devel] Re: Contributing to RT.

[Todd Chapman]

On Tue, May 31, 2005 at 10:23:13PM +0200, Florian Weimer wrote:
> * Todd Chapman:
> 
> > I understand. If only we had some way to use community resources 
> > to improve the situation. 
> 
> How easy is it to set up the test suite?  Unfortunately, I haven't
> tried to run it yet.
> 
> If it's reasonably straightforward, it might make sense to implement
> something like GCC's development process:
> 
>   <http://gcc.gnu.org/contribute.html>
> 
> To summarize: After running the test suite, developers submit patches
> for review.  The relevant maintainers comment.  Once the patches are
> approved, the original authors commit them themselves (and it's their
> job to resolve all conflicts).  Most patches should come with test
> cases.
> 
> The underlying idea is that because of the test suite, you can
> concentrate on what the patch does, and not so much how it interacts
> with the rest of the system.  There's no designated integrator, so
> this chokepoint is eliminated.  Of course, by implementing such a
> model, you relinquish quite a bit of control.


How many light years are you out? This is quite an old thread. I
can feel Jesse cringing at your suggestions above and I don't
blame him.

What I would love to see though, is for BP to take the time to
publish specifications for wish list items they would like to
do but don't have time for. I think they would get a lot more
community contribution that way, and would greatly increase the
chance that patches are accepted, which would probably create
some sort of really awesome feedback loop, perhaps ending with
the Singularity. :)

-Todd