some simple (hopefully!) visualgst ideas

> All,
>
> looks like visualgst is healthy and is introducing some interesting
> improvements to Smalltalk browsing:
>
> - sidebar sender/implementor
> - compilation errors in an infobar
> - tabbed browsing
>
> Here are some more ideas to improve VisualGST with respect to this points:
>
> 1) finish the transition to BrowserState objects by:
> - adding ContextState for the debugger
> - adding #state: to the browsers replacing the complicated #select...
> methods; setting a state automatically updates the listboxes
> - rationale: this should simplify fixing
> http://smalltalk.gnu.org/project/issue/505 by reverting to the previous
> state upon a compilation error
>

> Hi Paolo,
>
> On Sun, Sep 26, 2010 at 12:09 PM, Paolo Bonzini <[hidden email]> wrote:
>> All,
>>
>> looks like visualgst is healthy and is introducing some interesting
>> improvements to Smalltalk browsing:
>>
>> - sidebar sender/implementor
>> - compilation errors in an infobar
>> - tabbed browsing
>>
>> Here are some more ideas to improve VisualGST with respect to this points:
>>
>> 1) finish the transition to BrowserState objects by:
>> - adding ContextState for the debugger
>> - adding #state: to the browsers replacing the complicated #select...
>> methods; setting a state automatically updates the listboxes
>> - rationale: this should simplify fixing
>> http://smalltalk.gnu.org/project/issue/505 by reverting to the previous
>> state upon a compilation error
>>
>
> Good idea
>
>> 2) merge the two notebooks
>> - rationale: the required refactoring will ultimately make it easier to
>> provide pluggable tools (e.g. sunit browser within GtkLauncher) and multiple
>> GtkLaunchers
>>
>
> That should be pretty simple: create a new widget with both notebooks.
>
>> 3) improve RBParser to provide fine-grained Exception objects; calling the
>> parse error block should simply be the default action of the exceptions
>> - rationale: fine grained exceptions are an easy way to customize the
>> compilation error infobar and provide squeak's "declare undefined variable
>> as..." functionality
>
> Yes, we should put it in the to do list.
>
>> - could also be used for incremental parsing. making an incomplete but valid
>> parse tree can help autocompletion
>
> For the autocompletion there is OCompletion feel free to port it ;-P.
>
> Your idea are good, but also we should focus on these points:
>
> - finish the package builder
> - versionning support. Do you want to have everything on your image? I
> don't, what
> happens if there is a crash ...
> - finish the debugger
> - test everythings since the soft becomes bigger and bigger everything
>  should be tested.
>
> Gwen
>
>>
>> Paolo
>>
>> _______________________________________________
>> help-smalltalk mailing list
>> [hidden email]
>> http://lists.gnu.org/mailman/listinfo/help-smalltalk
>>
>