Questions/Ojections about FAI 2.9 upcoming release

[Henning Sprang]

Hi,

On Mon, 2005-12-12 at 15:27 +0100, Thomas Lange wrote:
> [...]
> 
> some weeks ago I send an email, that I like to prepare FAI 2.9. See
> http://www.uni-koeln.de/bin2/maillist/linux-fai-devel/20051125.175955/170489
> [...]

Oops, I remember that one - at least the question for testing and more
patches and cleaning up branches - I didn't have the announcement of 2.9
in mind anymore, and when I asked on irc oz_ also said he cannot
remember a 2.9 announcement.

Seems I am not concentrated enough and overreacting...

> [...]
> Since most people on IRC are active working on FAI, I think it's a
> good idea to ask people (after the mail several weeks ago) what to
> care about for FAI 2.9.

My concern wasn't about too much has been said on irc, I just had the
impression/fear things being _only_ said on irc... Sure irc should be
used, and it's a good place for ad-hoc discussions, but it should also
be kept in mind, that some people don't go there, so all important
announcements should also be made on the list. 

> 
>     > - There wasn't even publications of beta *.debs for this release, which
>     > can increase the likelyhood of more testing
> We had some discussion when I create some *.debs before, and then
> someone says it does not make sense to create debs that may be out of
> sync with the svn trunk. Therefore everybody who likes to test the new
> version can build it from the svn sources.

Yes, in some of the 2.8* release cycles we had too many beta packages,
also some who weren't correctly versioned IIRC, at least there was some
chaos.
But it could be useful to have some days/weeks of a freeze/betatest
state, where testing takes place, and for such a time a deb could be
made.

> [...]
> 
>     > - I doubt that this code has been tested enough. we had quality problems
>     > before because of releases in a hurry. 
> I can't tell I you how much I tested this version:-). I also think
> that it's not that important to have a bug free version, since we are
> not two weeks before the next Debian release. 

Generally I think each new version should be better than the last one,
not less.
Every release should be as bug free as possible. But that's a big goal.
Maybe we won't achieve that without automated (from which we are far
away - /me prays to find somebody sponsoring me for FAI work) or very,
even too long beta testing periods.


> [..]
>     > I there an important need to release right now, and no way to wait til
>     > january?
> Without a new version in unstable, I can't upload the sarge bugfig release.

Oh, really? I did not know that and don't understand why this is the
case, say, what the debian bureaucracy wants to achieve with this. If we
are forced that way, that is a totally different situation and a good
reason. I understand much better then!

> So I like to upload FAI 2.9 and after that I can upload the sarge
> bugfix release. I'm also not sure if I will get more test feedback
> until january. 

At least I hope to be able to do more testing, I am currently putting
lot of energy into my new job, so when I come home, I am most often only
able to reply some mails.

> Also, for planing my timetable it the best to release soon.
>
> Henning, I know you have objections with this release, but keep in
> mind that it's only an upload to unstable. AFAIK this will not make it
> into testing, because of the FHS pathes bugs.

Then I don't understand even less what strange rules force us to do such
a release... but that's not our fault then.

Henning