2010/3/12 Edgar J. De Cleene <edgardec2005 at gmail.com>:
>
>
>
> On 3/12/10 1:49 PM, "Ron Teitelbaum" <Ron at USMedRec.com> wrote:
>
>> Maybe in the future when the fury of
>> work that is going on in Squeak right now settles down we will need a less
>> risky method of gathering and releasing changes.
>
> It's on my agenda as soon 4.0 is out and I begin to manage the Release.
>
>> Maybe I'm just missing something here but I don't want to miss the work that
>> Andreas and Eliot and a host of others are offering and I don't want to
>> scare off non core developers from making changes if they are inspired to do
>> so via mantis.
>
> We don't lose they.
> We won more people as I wish go to two ways model, plain old .cs and trunk.
> All going to 4.1 must follow Ralph quality rules for 3.10.
>

Ouch! Then don't load closures, they break a bunch of tests ;)

I thought the model was more like:
- take a trunk mcm as starting point
- concentrate on fixing failures and errors
- mark irreductible as knowFailures
- select which pachages to remove to make a core image
- try fixing cycle again after unloading many packages -> Core
- select which pachages to add to make a full image
- try fixing cycle again after loading many packages -> Full

No ?
That said, I don't see so many tests failing in trunk.

Nicolas

>
>> Ron Teitelbaum
>> (Disclosure: I work for Teleplace but the opinions are my own)
>
> Lucky you !!
> If Teleplace hope to win, should think in talented students here in Rosario.
> I have some prospects...
>
> Edgar
>
>
>
>