Re: [orca-list] Structural HTML navigation by lists

[Joanmarie Diggs <joanied gnome org>]

Hey Marcus.

> The new pop-up menu is using the criteria/predicate (for searching) and
> present (for presenting) properties of the corresponding
> StructuralNavigationObject instance.

Awesome. It sounds like you are (already were) heading in the same 
direction I was envisioning. Nice!

> This safes many lines of code. The current changes will result into a
> patch with ~ 850 lines. My last patch on the other hand has more than
> 1500 lines.

Also nice.

> This could be done if you add two lines like above to the corresponding
> method (_spellingErrorsBindings or so). As I said the search and
> presentation is done by the *Criteria/*Predicate and *Presentation methods.

Two lines ain't bad. :-)

> joanie, I have tested the guessTheLabel method some days ago to save
> more code. But this method has returned mostly empty labels. And
> currently it seems to me that this method is only executed on form
> fields. Could this be true?

Not just could; definitely is. What label guessing* does is just that: 
guess** the label for unlabeled form fields. Ignoring Orca for a 
moment... A label is not the same as a name; it's a separate object 
(GtkLabel, <label>foo</label>, etc.) that... labels... another object. 
And that typically only occurs in form fields (and dialog boxes): 
specifically in widgets which expect data to be provided or altered.

Getting back to Orca.... If a widget is properly labelled, then then 
that widget should obtain its (accessible) name via the label 
automagically. Thus there is no reason for us to guess a label because 
we have a name we can trust. If the thing isn't a widget, but is instead 
some sort of text object, then it doesn't make sense to guess as we're 
interested in the displayed text.

(Make sense?)

--joanie

* soon to be renamed label inference -- Ale doesn't like 'guess' ;-)
** infer ;-)