Re: STL revisited

On Mon, 2006-09-04 at 02:25 +0200, Henrik Nordstrom wrote:
> mån 2006-09-04 klockan 09:21 +1000 skrev Robert Collins:
>
> > Right now, 3.0 is the top priority. However, I think the big step is
> > saying 'using the STL is allowed' - we should always be considering any
> > individual change with rigour - and doing a single large branch is going
> > to be much harder than just having a series of branches which say 'here
> > is an improvement X, and it happens to use this STL feature.'
>
> Again, I am fine with forking 3.0 pretty soon to allow for these things
> to get started in an incremental manner if you prefer doing further
> refactoring instead of focusing on bugfixes.

I think the best use of the time I have available is in improving the
testing & development facilities for squid, rather than working on new
features. So yes, refactoring is part of that.

> I very much prefer this kind of transition to be done incrementally over
> time instead of a big batch.

Agreed.

> I don't agree that forking would seriously further delay 3.0 unless the
> developers wanting to see 3.0.STABLE can't focus. People wanting to work
> on new features is likely to do that anyway rather than fixing bugs, and
> I rather have those efforts collected incrementally (with due review) in
> HEAD than floating around and bitrotting elsewhere like we have done in
> the past..

I'm not claiming that forking for 3.1 as a new trunk will delay 3.0
unless everyone switches ;).

What I'm really trying to say is 'lets get 3.0 out the door!'.

-Rob

-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.