Am 10.12.2004 um 21:50 schrieb stéphane ducasse:
>
> Normally if I remember what marcus told me is that he introduced
> a preference to be able to switch between full and not full closures 
> but normally
> we will use it in not full. Then as Etoy depends heavily on the old 
> compiler we will
> keep it until a good soul change etoy to use the new compiler but all 
> the other method
> compilation will use the new compiler). Marcus can you tell me whether 
> I'm right?
>

The new compiler will be introduced in a way that nobody feels any 
pain. That is,
at first the pref to use it will not be enabled. After some (a lot) of 
testing, and making
sure that eToys uses the old compiler automatically, this will change.

At some point, after making sure that nobody uses the old compiler 
anymore, it will
be put (in a working state) on SqueakMap.

I think eToys are in 'Hard Maintanance" mode (so I infered trying to 
"read mind over
distance"). Of course nobody thinks it might be important to 
communicate such stuff to
anyone...

Nobody will put any time into them other then fixing bugs for 
squeakland, so we keep
them till we throw all that stuff  (morphic, eToys) together. I guess. 
Maybe. Very likely.

Building an eToy like system should be much cleaner with the 
RB-Parsetrees... but I'm
sure that even eToy2 will be a brave hack using the old compiler 
instead.

>> For System-stuff you mentioned, how about SmaCC?
>
> Why not as a package. Good idea.

I don't think that it makes sense to add the development package to the 
full image.
the SmaCC-Compiler-Compiler is only needed if you want to change the 
Scanner
or Parser. Somebody doing that can then install it via SqueakMap. (the 
runtime
is already in for some time)

        Marcus