Re: Policy on pkgs with same filename and QA rant (was: Re: Mercury compiler (policy, anybody)?)
Chris Fearnley wrote:
> 'Christian Schwarz wrote:'
> >If you want to add something to our policy manual, someone has to try to
> >formulate the idea, so we can simple discuss it. We need something more
> >"general" for our manual, as for example:
> >``It is not allowed that two packages install programs with different
> >functionality but with the same filenames. (The case of two programs
> >having the same functionality but different implementations is handled via
> >`alternatives.') If this case happens, one of the programs has to be
> >renamed. The maintainers should report this to the developers' mailing
> >list where it will be dicussed. The VP of Engineering will make a final
> >decision which program has to be renamed.''
> Kill the last sentence replacing it with: "If consensus can not be
> reached, the two packages will have to conflict with each other."
> I'm appalled by the word "VP" occuring anywhere in the the policy
> manual. Someone else recently said that we should make our decisions
> based on technical reasons. I concur. I'm disappointed to see so many
> compromises on quality lately (right after our long thread on QA).
> Engineering excellence requires an uncompromising approach to
> problems: complete solutions must be sought (solutions that satisfy
> all parties). And consensus is the best way to accomplish said
> excellence (I cite the Internet STD documents and Linux kernel
> development as proof). Is my fear being realized: that the existence
> of a quality control manager will decrease quality because one can
> lobby the QA manager to get a hack "approved"?
Very unlikely. We're to big to get TOTAL concensus. Ask 5 Engineers
and you'll get at least 7 good solutions. Therefore, we need someone
responsible to call the shots once several solutions have been
discussed. Discussed does not mean beat it to death. Even without a
religous war, there are still disagreements. We need somone responsible
for making that call. Therefore, "(solutions that satisfy all parties)"
is not a viable alternative.
Finally, consensus is only useful for small group of experts. After
that, political desires tend to be big influences.
> We all need to be QA managers. We all need to be Distribution
Agreed. Each one should baby-sit themself. We need administration more
as a resolution tool. Total agreement of Board is much more likely.
Also, we can only manage our part. We need admin to coordinate. If we
waited for everything to be 'up-to-date' we'd never have a release.
> PS. Dale, can you make sure the Board emphasizes the importance of
> consensus in this constitution they are working on.
Yes, however, consensus != (solutions that satisfy all parties)