Re: load balancing bazaar style
Ok, it seems like then that throwing more effective random and comprehensive testers at it is the
best.
How about for every bug that debian addresses, a unit test is developed (which has to be anyway to
fix the bug). And from then on, this test is always employed. This will at least be better than
nothing and assist in tracking regressions instaneously.
I think that a bigger regression test archive can also start developing.
Even after 10 years of gathering tests like this, I doubt that it would ever take more than a day
to run through this batch of data. And 10 years from now, we'll have machines that could turn this
day I speak of (in today's technology), into seconds.
So I see no reasons against this. But maybe u can find some to make the idea stronger
--- Daniel Jacobowitz <dan@debian.org> wrote:
> On Thu, Feb 20, 2003 at 12:52:18AM -0800, Hunter Peress wrote:
> > The fact that libc has been a bottlenecking-causing hotspot over the last year makes
> > me think that some better efforts be taken.
> >
> > I think that a simple protocol could be put in place whereby there is an effort to
> > make the weekly requirements for glibc is made easier for members of the public to
> > simply jump right in.
> >
> > I think there should be a stack type system where a bunch of tasks are popped on by
> > the glibc team, and they can be popped off by anyone.
> >
> > With some publicity, some of the less interesting, but necessary tasks can get done
> > at times by anyone that is capable of coding in c.
>
> Sorry, good thougt, but useless. Have you looked at the list of things
> that were holding glibc back? None of them could be addressed in this
> fashion. They were all bugs that people didn't know what to do about.
>
> And anyone who wants to help is more than welcome, and quite able.
>
> --
> Daniel Jacobowitz
> MontaVista Software Debian GNU/Linux Developer
__________________________________________________
Do you Yahoo!?
Yahoo! Tax Center - forms, calculators, tips, more
http://taxes.yahoo.com/
Reply to: