Driving towards Bugzilla 3

[David Miller]

Vlad Dascalu wrote:

> So the main issue is that when the rewrite idea dropped, the 3.0 idea 
> dropped as well. Due to the 6 months release cycle, there will be no 
> other pair of releases with a considerable feature set difference in 
> them that will be greater than the 2.16 versus 2.18 thing (those covered 
> 2 years of development). Since Dave rejected 2.18 as 3.0, we're back 
> where we started.

I've continued to be open to the idea of changing a version to 3.0 at 
some point in the near future, and I've even actively campaigned for it, 
but every time I've brought it up, nobody could come to a consensus of 
what would constitute sufficient change in a version to call it 3.0.

I think what we've more or less decided at this point is that our CVS 
repository is very poorly laid out at the moment, and needs to be 
reorganized (as well as the installation location of files so the stuff 
that the webserver isn't supposed to see doesn't have to be under the 
docroot of the webserver).  When this reorganization happens, it's more 
than likely it will break people being able to upgrade a live production 
installation via CVS in place (it would have to be moved out of the 
docroot then upgraded, then reinstalled back to the docroot).  That in 
itself seems like a very good reason to bump the version number, since 
people expect major installation process changes when a major version 
number changes.

> The feature set for 3.0 seemed reasonable ok. The issue is that we kinda 
> dropped those aims when we dropped the rewrite thing. The question now 
> is how to make a roadmap that will allow us to have the 3.0 feature set 
> without doing the 3.0 rewrite thing.

What was the proposed feature set for 3.0?  I honestly don't recall 
seeing anything other than the design documents for PLIF...  It seemed 
more of an idea of completely abstracting everything.

> The bigger picture might be: are we still considering the 3.0 feature 
> set in the long term, without having to do a rewrite this time? How do 
> we plan to achieve that with the current code-base? On what steps/bugs 
> should we focus in order to improve modularity/abstractization layers 
> needed for the 3.0 goals?

Ian himself has admitted that some of his abstraction ideas resulted in 
completely unacceptable performance problems, and thus would be unusable 
in a production environment anyway.  (That's among the reasons he gave 
up on the rewrite).

-- 
Dave Miller      Project Leader, Bugzilla Bug Tracking System
http://www.justdave.net/             http://www.bugzilla.org/