Legacy in Build Roots

[Toshio Kuratomi]

On Sat, 2006-08-05 at 11:16 +0200, Nicolas Mailhot wrote:
> Le vendredi 04 août 2006 à 17:50 -0700, Toshio Kuratomi a écrit :
> 
> I think both you and the Legacy people misunderstand the meaning of the
> split for packagers. It's not "since there is a split I'll only maintain
> current releases" but "I'll only maintain current releases and since
> there is a split I'll do so till the legacy handover".

I see your difference in emphasis.  This is a policy difference, though.
If we define what a packager's responsibilities are (as Michael Schwendt
has been pushing for ages) then we will have expectations that align
with one of those phrases.

The important thing is in drafting that policy we need to be sure we're
confirming what Extras Packagers already want.  Since packagers are
volunteering to do the work the policy has to address what work they are
willing to volunteer to do.

> Without the split I for example would not have been maintaining my
> packages forever but would have stopped maintenance much sooner (only
> doing rawhide + current).
> 
Is this true even if there was a split in the level of support?  The
same packager owns the package from inception to True EOL but there's a
place in the middle where maintenance goes from "new features, bugfixes,
upgrades" to "fix security issues only"?  There is definitely a
different mindset and motivation for maintaining packages on each side
of that divide; I'm wondering if the difference in audience and mindset
is large enough that most packages will require two owners or if "fix
security issues" is a small enough amount of work and important enough
issue that packagers for current releases would be willing to be
responsible for that.

In other words what should be our default expectations?  Do we need an
FE-Legacy group and any current packagers that want to can step in to
help or do we have the kind of setup we have now: Package owners are the
primary line of defense in fixing security issues (and breakage that
falls out of that) and other groups (Security team and Dennis Gilmore
currently) step in when the owners bow out?

Also note: If Extras packagers are expected to fix issues on Legacy
builds Extras packagers should get input on how long Legacy releases
were to last.  Legacy is not forever, there is a true EOL.  If Extras
packagers are doing work on Legacy, then they need to have a voice in
how long Legacy lasts.

> The split is not a "stop there" but "do at least so much"
> 
I think this depends on the packager.  That said, I also view it as "do
at least so much".

> > Long term we
> > might want to get rid of a separate Fedora Legacy -- instead maintainers
> > of both Core and Extras packages (and their teams of co-maintainers)
> > will continue to provide security and major bugfixes to the Legacy
> > releases.
> 
> The current situation has nice commitment limits (granted legacy should
> move closer to extras but that's a legacy not extras problem). I'll
> personally oppose at my low maintainer level lifting them.

Could you sum up your reasons?  I have:
1) It takes energy away from maintaining packages for the current
releases.
2) It is well placed to make maintainers feel comfortable with the level
of commitment.

> However should they be lifted I think you totally misunderstand what the
> effect will be. People committed to maintaining every Fedora release
> under the Sun will continue to do so. Other people (because more
> releases is more work, which sucks) will probably shift their
> commitment, either doing less updates to address more releases or less
> updated releases than now. Either way you look at it Extras quality will
> go down since maintainers won't be focused on the same releases anymore.
> 
I don't understand your two alternate scenarios for "other people".

> In a volunteer project rules do not affect the amount of work people do.
> They only affect how the volunteers whose to allocate it. Spreading the
> available work budget over more releases will only benefit Legacy. The
> Legacy people should remember however a smaller Extras will mean less
> Fedora users and less Legacy users.

I agree with this.  So another way of phrasing this debate would be: Do
Extras packagers want to spend their time fixing older releases or
creating more packages?

-Toshio
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://listman.redhat.com/archives/fedora-maintainers/attachments/20060805/66126710/attachment.sig>