[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: splitting experimental by arch?

>>"joost" == joost witteveen <joost@rulcmc.leidenuniv.nl> writes:

>> The opposing argument is that this is a new package, and if people
>> load it, it is caveat emptor.

joost> Well, the opposing argument was that unstable is just the place
joost> for your package.

	Sorry. There are a lot of people that live on unstable. And
 the contention is that even packages in unstable are trying to be of
 a quality that does not leave the machine unusable, yes, there are
 bugs, but the packages are not known (or expected) to have major

	We ask the developers to run unstable; and some packages just
 do not belong on the general machine. 

	Packages in experimantal are meant for the subset of the
 developers actively working in the are, and who are expected to keep
 their eyes very wide open.

>> I disagree. Packages in stable should be tested and be close to
>> production quality. If the package is experimental in the authors
>> eyes, it should not be handed to users even if we know of no bugs
>> (yet).

joost> Yes, of cource, your package (the one you described abouve)
joost> should not be in "Stable", it should be in "unstable". On the
joost> (rather rare) chance that there is a changeover from unstable->
joost> stable, then probably you should (just before the freeze or
joost> whatever) have the package removed.

	Easier to have and use experimental, just in case.

	My thesis is that we do need three levels of instability:
 stable (released)      : for the general public
 unstable: (beta)       : for developers, beta testers, and adventurous users
 experimental: (alpha)  : for people who are hacking the package

	I think a three level system like this has worked quite well
 for us in the past. Technically, an alpha testbed shall help us
 experiment with packages while they are in a state more unstable than
 the beta level unstable justifies.

joost> And if it will take a long time for the package to become
joost> usable, then maybe Debian doesn't really want to distribute it?

	Huh? I definitely do not agree. If I upload a new package next
 month (I am likely to upload a version ov cvs-buildpackage with major
 changes), and then hamm freezes: then answer is we do not want
 cvs-buildpackage? Just so there is no experimental category? Why on
 earth for? what is your technical justification for not having
 experimental just like we do now?

>> This is a quality of distribution issue.

joost> Of unstable?

	Pewcisely. Unstable becomes stable automatically. If unstable
 is shoddy quality, stable is likely to be of a lower quality. Improve
 the quality of unstable, you'll have more people willing to test
 unstable, and become developers, etc. 

	Why is quality of unstable such a novel concept?


 By one count there are some 700 scientists with respectable academic
 credentials (out of a total of 480,000 U.S. earth and life
 scientists) who give credence to creation-science, the general theory
 that complex life forms did not evolve but appeared "abruptly."
 Newsweek, June 29, 1987, pg. 23
Manoj Srivastava  <srivasta@acm.org> <http://www.datasync.com/%7Esrivasta/>
Key C7261095 fingerprint = CB D9 F4 12 68 07 E4 05  CC 2D 27 12 1D F5 E8 6E

TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org . 
Trouble?  e-mail to templin@bucknell.edu .

Reply to: