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

Re: [debian-devel] Re: [debian-devel] Re: The stable/testing/unstable branches not a solution ?



A levelezőm azt hiszi, hogy Enrico Zini a következőeket írta:
> On Mon, Jan 26, 2004 at 04:39:24PM +0000, Magos?nyi ?rp?d wrote:
> 
> > > I quite admire the way GNOME are doing releases these days; "we release
> > > on this date, if your code isn't in, it's in the next release."
> > > Seems to be working well for them.
> > Let's do the same.
> 
> I think we're doing the same already, and that's why there is software
> in stable that's obsolete even before stable releases.

No, we are _way_ beyond schedule.

> 
> You also have the problem that sometimes you _can't_ release without
> some software having RC bugs (like, sometimes glibc or d-i have RC bugs :)

Having RC bugs in essential packages is a special case, which might
warrant a one or two days slip in schedule. But not months.

> Anyway, there is a big difference between Debian and Gnome to be taken
> into account: Debian sid is at the moment something like 13882 packages.
> Sarge is less, but the order of magnitude is the same.  Gnome is _so
> much smaller_ (and completely included in the Debian package set :).

Let's compare the size of Gnome to the essential package set then.
If we have a complexity which cannot be handled in the ways we
try to handle them, then we should not handle it in the ways we handle
them.

> 
> Making 10/20/50/100 (how many are they?) gnome applications work
> together is a radically different task than making thousand or tens of
> thousand packages work together.

Agreed. It might lead to some insight if someone could point out
in which ways exactly it is different. I see that the number of
interdependencies is growing exponentially in theory, but fail
to catch qualitative differences (which do not mean there
aren't any).

> 
> That is what makes us unique, what is showing scalability issues at the
> moment and what requires us to have some new, innovative ideas.

Until someone can come out with the holy grail, I would stick to
being extremely cruel in release measures, and try to find ways
to further minimize interdependencies.
I'm afraid even these ideas count as too new and innovative
in the current state of affairs.

-- 
GNU GPL: csak tiszta forrásból



Reply to: