Questions/Ojections about FAI 2.9 upcoming release

[Henning Sprang]

Hi,

On Mon, 2005-12-12 at 15:35 +0100, Michael Tautschnig wrote:
> Hi all!
> 
> [...]
> > 
> >     > - this release has not been announced on any mailing list before, and I
> >     > don't think it's a good idea to make the IRC the premier information and
> > 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.
> Still, I'd like to have found some mail on this list that you are going to build
> the package *now* (I'm not on IRC...). I'm pretty much aware of the fact that
> there was no response to your request for bugfixes that should make it into the
> new release, but at least you changed quite a lot...

Yes, even if most of my concerns are gone in the meantime, I agree it
was not a completely clear announcement.

Maybe it would be cool to announce such things on the mailing list and
also write a plan in the wiki like: from december 10 to december 20 beta
test, planned release of 2.9 on december 22 or something.

> [...] 
> - The fact that this is an upload to unstable is really important here - and we
>   should try to encourage people to test these new packages; AFAIK only very few
>   build packages from the current trunk, but it is a lot easier to simply fetch
>   the packages from the repository...
>   At first I really intended to support Hennings opinion that there has been too
>   little testing, but now I'd say it is up to unstable-users to test the
>   packages... 

Yeah - both views are right.
I sometimes think that we could have stable releases which not must be
in snyc with Debian releases. I don't want to force FAI users to wait in
such long cycle to get the benefit of new cool FAI features. But I am
O.K. for now - it seems anyway to be a very special situation.

> 
> Still, we should think about some way to record tests, i.e., who tested which
> version doing what. At least it would be useful if there were some notes on
> "patch xyz reviewed by..." - but I have no idea, where this should be done. The
> wiki? The mailinglist? The svn?

We could start with the wiki. I already wrote some notes what would be
the minimal needs to be tested in the wiki - see
http://faiwiki.informatik.uni-koeln.de/index.php/Testing

We should log there who has tested what with which beta release, and
only get a release out if each important feature has been tested
successfully by, say 5 people.

If we see that the wiki isn't enough, could come up with some proposal
for a web-based tool for managing test protocols.

Henning