On Thursday 03 October 2002 9:40 am, Rob Lahaye wrote:
> Angus Leeming wrote:
> > Rob, here's an updated patch to replace what you have
> > already. It incorporates some bugfixes, your suggestions and
> > most of those made by Allan (off-list).
> >
> > It also gives you a checkLyXLength() function for ... well,
> > you know.
> >
> > I suspect that you'll be able to simplify those input()
> > methods considerably.
>
> Thanks Angus!
>  From now on, please don't send me patches pulled from your
> previous ones, since I made modifications here and I'll have
> difficulties to merge them.

I'll leave it in your tender care ;-) One thing I'd like you to 
change is:

ButtonControllerBase.h:
struct CheckedWidget {
-       virtual bool check() = 0;
+       virtual bool check() const = 0;

You should also add a const to the declaration and definition of 
the check() methods in checkedwidgets.[Ch]. These methods do not 
alter the list that stores them and so should reflect that in 
their definition. It doesn't matter that they might (do) alter 
the objects pointed to.

> My problem is: I can't see where the Paragraph and Graphics
> differ. Do you know?

I do.

Oh, Ok. ControlParagraph derives from ControlDialog but 
ControlGraphics derives from ControlInset.

You should add
+
+       // The widgets may not be valid, so refresh the button 
controller
+       bc().refresh();
to the end of methods
void ControlInset<Inset, Params>::show(Params const & params)
void ControlInset<Inset, Params>::update()
to be found in controllers/ControlInset.tmpl

> Another thing I don't understand: I ran the debugger (gdb)
> with a breakpoint at "CheckedGlueLength::check". Start LyX, do
> "File->New". You can mouse-click anywhere on the canvas, no
> problem. Now open a dialog that uses the length check and
> close it again; you'll have to continue several times at the
> breakpoint. Now all of a sudden the leftmouse and middlemouse
> clicks on the canvas stop at the breakpoint; the middlemouse
> click even more often than the leftmouse click. What's going
> on here? Why does a canvas click need a length check? It
> appears to me that the canvas click somehow is intertwined
> with the dialog stuff, but only after a length-check-dialog
> has been used. Any idea what's wrong here?

None. ;-)
Angus

Reply via email to