Re: potato late, goals for woody (IMHO)
> > * better and newer boot-floppies in shorter time (language support in four months and
> > not in 16 months)
>
> This is a function of the number of people working on it, and the amount
> of testing done, not the release cycle.
Correct, but if we have it ready after one month of work, we must wait
14 months to release it. Ok, this is not 100% fully true, but i hope you
get the idea.
> > * new stable architectures (as powerpc and arm currently; mips and
> > hurd for the future)
>
> Again, they'll be ready when they're ready, according to the amount of
> work being done. Doing more releases just means that there will be more
> "Nope, not ready for stable release yet" decisions.
Again, if the architecture is ready now, we release it now and not next year.
> > can fix faster bad setups (boot-floppies, x11, usb, firewire,
> > keyboards, languages, ...)
>
> That's a matter of doing update releases on the stable dist, which we
> are now doing.
Will we have xfree4.0 for slink?
> > * we see a light at the end of the tunnel, have not so big goals,
> > better quality managment,
>
> Why better QM?
1 year cycle: 1000 or all packages must be checked
3-4 months : 100 (+/-) must be checked
The QA-Team can work intensivly on those packages, and so better quality
and easier QM.
> > * i hope we start important things at the start of a new cycle and not
> > at the end
>
> You've got cause and effect reversed here. We've never planned on
> waiting a year between releases. The fact that we delay significant
> decisions until the 3 months into the developement cycle is what causes
> the delays, not the other way around.
No; we had no development stop after three months in the past. So everyone
ignore it and all maintainers think year-wise and not month-wise. The
delay comes then automatic.
> > * work on 'new things' are prefered by many peoples
>
> Those people will never be interested in making stable releases, because
> they are bored by quality control and documentation.
Right. Do you upload to woody (only woody i mean)? Why, if so? Why not helping
in quality control and documentation?
> > * easier for test-utilities
>
> Why?
Because we have less packages to test.
> > * less bug possiblities, because of not so big package version updates
>
> Uhh, this makes no sense. For the most part, packages track the upstream
> releases. The fact that there is a big jump in versions between stable
> and unstable doesn't mean the itermediate versions weren't packaged and
> distributed.
But we have then not so often updates per release.
> > * we must not wait for the next version of gcc, libc, kernel, x11,
> > mesa, ...
>
> Were not waiting for them (or maybe I don't understand what you're
> saying).
kernel-image-2.2.15 is such an example, xfree-3.3.3 another one in the past.
> > * hopefully less RCB's
>
> Why?
50 or 100 new(er) packages have less RCBs then 1000.
MfG,
Hartmut
Reply to: