Smart on Fedora and openSUSE

[Gustavo Niemeyer]

> The main reason for doing a new interim release would be to integrate
> various patches already applied downstream in the distros packaging.

Agreed.  That's one of the reasons we need a new release very shortly.

> Maybe the current "trunk" series should be split anyway, to allow
> for faster integration of more experimental features while still
> offering a completely stable main branch for legacy deployments ?
> And there the "1.x" versus "2.x" naming split felt rather natural...

Having an unstable branch for stabilizing new features as they are
developed is nice.  That said, there's no difference in terms of
development practices for either branch.  In other words, if it's not
properly tested, it shouldn't go into 1.X *or* 2.X.

I'm pointing that out because good part of the current contention is
related to lack of proper testing in suggested changes, or in some
cases because the changes break backwards compatibility in an
unnecessary way.  Whenever I go into a round of integration of
external patches I end up writing many of these tests myself, and that
means less integrated patches per time spent reviewing changes.

> For instance, you might want one "stable" version for servers
> and one "testing" version for hobbyists and enthusiasts [sic] ?

I'd rather see effort being put on producing one line of development
which is updated more often.

-- 
Gustavo Niemeyer
http://niemeyer.net