The meaning of STABLE (was: X server 1.9 release thoughts)

[olafBuddenhagen at gmx.net]

Hi,

On Tue, Apr 06, 2010 at 02:43:01PM -0700, Jeremy Huddleston wrote:
> On Apr 6, 2010, at 11:47, Daniel Stone wrote:

>> Er, is there no reason hardware enable (even if it's not entirely
>> fully-featured) can't be done in point releases?
>
> Yeah, I thought the 6-week point release schedule was mainly to
> address this very concern amongst the drivers developers.

I firmly believe that point releases should *never* add any new
functionality. Ever.

The whole idea of a stable branch is, well, that it's supposed to be
stable... Upgrading to a newer point release should *fix* bugs, not add
stuff that is likely to introduce new ones. Avoiding regressions is the
top priority here.

If more upstreams would commit to such a policy, distributions could
more realistically upgrade to point releases within a stable release
branch, instead of just manually picking patches for the most critical
bugs, for fear of regressions...

I'm afraid this will sound a bit pathetic: but I believe the fact that
for non-savvy users, the only way to deal with a problem in one Ubuntu
release is to wait for the next one (introducing new bugs in turn), is
the single most disturbing technical problem with using a free operating
system.

Note that I picked Ubuntu as the example here, because that's the one
for which I have first-hand experience of the unbearable troubles this
causes -- to the point where I no longer feel I can recommend people to
use Ubuntu with a clean consience... But I very much doubt that other
distributions are considerably better in this regard. No distribution
has the means to manually go through tens of thousands of changes for
thousands of packages, and deciding which are safe to include in a
stable release. It's what upstreams are for -- if only they were more
conscientious about it :-(

-antrik-