Re: [email@example.com: Re: Woody retrospective and Sarge introspective]
On Thu, 1 Aug 2002, Raphael Hertzog <firstname.lastname@example.org> wrote:
> Ok. What about offering that possibility to maintainers who are interested by
> it ?
> Instead of removing the link unstable->testing, we just add another one
> t-p-u->testing which is also governed by the testing scripts.
"Where do I upload?"
"It depends. On Mondays, Wednsdays and Fridays, you upload to unstable.
On Tuesdays, Thursdays and Saturdays you upload to t-p-u.
On Sunday, you rest."
More important, which distribution should users who want cutting edge software
use? Up to know, through all of Debian's life, the answer was "unstable".
Now, you propose it to be "unstable, but with t-p-u, except they
might not be installatable, so maybe testing with t-p-u except then you
miss out on some cutting edge stuff"? Try saying both answers five times
quickly to see which one is easier... ;)
> That way, maintainers who want to send updated packages in testing have
> a chance to do so even if their usual dependencies would block them in
Or, alternatively, software with many people who depend on it could upload
to experimental or to a personal staging area. If you say your goal is to
get experimental to autobuild instead, that indeed would be laudable. Of
course, James might not feel he wants to code that, but he probably feels
less inclined to code what you suggested, and certainly aj seems so. At
least a patch to get experimental to autobuild would not change anything
either maintainers or users do except give maintainers a choice: if your
software really might break other stuff, upload to experimental.
(That seems much cleaner than an upload to unstable + RC bug, anyway)