2016-04-28 7:20 GMT+02:00 Tudor Girba <tudor@tudorgirba.com>:

Hi Nicolai,

> On Apr 27, 2016, at 11:12 PM, Nicolai Hess <nicolaihess@gmail.com> wrote:
>
>
>
> 2016-04-27 23:07 GMT+02:00 Thierry Goubier <thierry.goubier@gmail.com>:
> Le 27/04/2016 22:55, Esteban Lorenzano a ��crit :
>
> On 27 Apr 2016, at 22:52, Thierry Goubier <thierry.goubier@gmail.com
> <mailto:thierry.goubier@gmail.com>> wrote:
>
> Hi Doru,
>
> Le 27/04/2016 22:38, Tudor Girba a ��crit :
> Hi,
>
> On Apr 27, 2016, at 10:17 PM, Thierry Goubier
> <thierry.goubier@gmail.com <mailto:thierry.goubier@gmail.com>> wrote:
>
> Le 27/04/2016 21:26, Hilaire a ��crit :
> Now I remember I already asked several months ago, and it does not
> work.
>
> Editing on the value does not work for me.
>
> http://forum.world.st/GL-inspector-editing-attribute-td4837704.html
>
> The same mis fortune is encountered with Pharo5
>
> I don't imagine how it can be like that and I fell unproductive now
> with
> Playground and GTInspector, althought I acknowledge there are nice
> ideas
> in these new tools but it can't be at the price of productivity.
>
> Hopefully you can switch to Workspace and EyeInspector.
>
> With the help of Nicolai Hess, we worked a bit on improving syntax
> colouring for the EyeInspector and this has been integrated. Maybe
> someone can look into doing the same with GT (to correctly set
> #doItReceiver, #doItContext and a few other things related to syntax
> highlighting).
>
> What exactly is the problem in GT regarding syntax highlighting?
>
> If you inspect a morph (say GTInspector inspect: Morph new), when you
> type bounds (one of Morph instance variables) in the text pane below,
> you get a red == erroneous / undefined var.
>
> but that��s not a syntax highlighting problem: is a bindings problem :P
>
> both
>
> binding, because glamours smalltalk code mode automatically creates bindings (when OCASTSemantic analyzer tries to
> lookup a variable) (see glamours workspacebinding strategy)
>
> and styling
> because for a "raw"-tab inspector pane, glamour does not set the classOrMetaClass attribute of the styler.

I am looking at this issue right now:
https://pharo.fogbugz.com/f/cases/17948/Bindings-in-the-evaluator-pane-in-Inspector

Could you explain why setting the classOrMetaClass is important in this context if we add the bindings to the instance variables?

That depends on how you want to add bindints for instance variables.

For evaluating expressions with instance variables, OCASTSemantic analyzer already knows how to lookup the

variable binding from the OCInstanceScope (it does not work at the moment, because inspect (the requestor) is asked first, and

it creates a binding on the fly).

For syntax highlighting, the styler needs to know if a classOrMetaClass attribute is set, otherwise it does not create a compilation

context that can lookup up with the correct instance scope. (see SHRBTextStyler>>#privateStyle: )

If you omit to create bindings on the fly for instance variables, you need to set the classOrMetaClass attribute

but if you include the instance variable bindings and the GT inspector (the requestor in #lookupVar: ) is responsible for finding the correct

var, it may work without setting the classOrMetaClass attribute - I don't know for sure.

��

Cheers,
Doru

>
> I can let you try to solve it by playing with the bindings ;)
>
> Thierry
>
> Esteban
>
>
> In the latest 5.0, in EyeInspector, it will correctly highlight bounds
> as a defined variable.
>
> Moreover, and the source of the first complaint, in GTInspector,
> inspecting bounds will give a nil answer instead of the morph bounds.
> Whereas EyeInspector will properly answer (0@0) corner: (50@40).
>
> Thierry
>
> Cheers,
> Doru
>
>
> By the way, would someone know how to force the styler to re-style a
> text? When selecting another element in for example a
> EyeTreeInspector, this changes the reference class for syntax
> highlighting (and the styler correctly picks that) but the existing
> text isn't re-colored.
>
> Thierry
>
>
> Hilaire
>
>
> Le 27/04/2016 15:51, Sean P. DeNigris a ��crit :
> HilaireFernandes wrote
> instance variables evaluate to nil in the bottom area of the
> integrated
> inspector.
> There is no direct inst var access from the playground. I was
> initially
> shocked by this as well and have had to resort to #instVarNamed:
> on several
> occasions. On the bright side, you can edit the values in place in the
> 'Value' column above.
>
>
>
>
> --
> www.tudorgirba.com <http://www.tudorgirba.com>
> www.feenk.com
>
> "Value is always contextual."

--
www.tudorgirba.com
www.feenk.com

"Problem solving efficiency grows with the abstractness level of problem understanding."