Hi, AbstractEnvironmentTestCase exists only in Pharo 8, tracking this with STON (from changes in Pharo 8) gives me tests that can no longer be loaded in Pharo 7 (or versions before that). At least AbstractEnvironmentTestCase should be back ported to Pharo 7. Apart from that, I don't think this is a good change (since it breaks backwards compatibility too much). Why do all Smalltalk>>#at: references have to be removed ? The way I see it, Smalltalk>>#at: could be rewritten to dynamically use a different environment, if that is set up earlier, using a DynamicVariable. Similarly, #asClass could be implemented to follow the same path. Or am I wrong ? Sven
On 19 Nov 2019, at 11:57, Sven Van Caekenberghe <sven@stfx.eu> wrote:
Hi,
AbstractEnvironmentTestCase exists only in Pharo 8, tracking this with STON (from changes in Pharo 8) gives me tests that can no longer be loaded in Pharo 7 (or versions before that).
At least AbstractEnvironmentTestCase should be back ported to Pharo 7.
Apart from that, I don't think this is a good change (since it breaks backwards compatibility too much).
I do not agree.
Why do all Smalltalk>>#at: references have to be removed ? The way I see it, Smalltalk>>#at: could be rewritten to dynamically use a different environment, if that is set up earlier, using a DynamicVariable. Similarly, #asClass could be implemented to follow the same path.
Because if one day we want to have modules and environment better to have a system that does not point to global variables. We should not refer to global variables. Stef
Or am I wrong ?
Sven
On 19 Nov 2019, at 12:27, ducasse <stepharo@netcourrier.com> wrote:
On 19 Nov 2019, at 11:57, Sven Van Caekenberghe <sven@stfx.eu> wrote:
Hi,
AbstractEnvironmentTestCase exists only in Pharo 8, tracking this with STON (from changes in Pharo 8) gives me tests that can no longer be loaded in Pharo 7 (or versions before that).
At least AbstractEnvironmentTestCase should be back ported to Pharo 7.
Apart from that, I don't think this is a good change (since it breaks backwards compatibility too much).
I do not agree.
Well, external code cannot write tests that confirms to the requirements of both versions.
Why do all Smalltalk>>#at: references have to be removed ? The way I see it, Smalltalk>>#at: could be rewritten to dynamically use a different environment, if that is set up earlier, using a DynamicVariable. Similarly, #asClass could be implemented to follow the same path.
Because if one day we want to have modules and environment better to have a system that does not point to global variables. We should not refer to global variables.
But eventually, the reference has to be resolved and it has a name. I am familiar with packages in common lisp, is it like that ? Is there a document that describes these issues more clearly ? The plan going forward ?
Stef
Or am I wrong ?
Sven
participants (2)
-
ducasse -
Sven Van Caekenberghe