ducasse at iam.unibe.ch
Mon Apr 7 08:24:25 UTC 2003
> Explicitly modelling instance variables would also be useful, at least
> from a versioning standpoint. You'll notice that in your snippet of
> declarative specification you have InstanceVariable as a separate
> declaration, whereas in mine (for now) inst vars are part of the class
> declaration. Among other things, this doesn't merge properly (if Colin
> and I concurrently add an inst var to the same class, the conflict
> have to be manually resolved).
So having explicit instance variable is really simple to add.
> The hard part here is that if variables (class, pool, instance) are
> treated separately from their class declaration, they can be in
> packages - and currently in Squeak there's no way to mark that (for
> methods we can abuse method categories, but no equivalent metadata
> for variables).
This is clear but we can imagine a normal use for now and see later.
Once you have
explicit support for package then the package becomes the palce where
you want to attached this information. Note that having a package
object would solve a lot of problems that will arrive for Pool and
Global for example
> If anyone has a brilliant suggestion of how to resolve this with
> changes to the kernel, I'll be delighted to modify PackageInfo and
> Monticello to suit.
Just having a package, bundle classes containing other declaration
would be enough.
I will discuss with roel and alex because we need also a declarative
syntax for the classBoxes.
Prof. Dr. Stéphane DUCASSE
"if you knew today was your last day on earth, what would you do
different? ... especially if,
by doing something different, today might not be your last day on
"The best way to predict the future is to invent it..." Alan Kay.
Open Source Smalltalks: http://www.squeak.org,
Free books for Universities at
Free Online Book at
More information about the Squeak-dev