Re: [Gnustep-cvs] r31864 - in /libs/gui/trunk: ChangeLog Source/NSSavePanel.m

[Gregory Casamento]

One other thing I would like to say on this is this:

Currently the code which implements different kinds of menus is full
of if/then/else clauses in methods.
if it's win95 do this, otherwise do that.  That's a clear indication
to me that a refactorization is sorely needed.

We need different menu classes for different menu types i.e.
NSMenuView should be an abstract class.    GSGNUstepMenuView (or
whatever) should draw the floating menus...  GSWin95MenuView should
draw the in-window menus...  GSMacMenuView should handle the Mac like
menus.
Something like that.

The trouble with how it's been done lays in it's reliance on the old
paradigm of handling events for the floating menu.  It's difficult
having so many paths through the same code to handle things
differently in this case or that.  I propose that we not only rethink
how we're handling events for in-window menus, but also how menus are
implemented at a very fundamental level.

GC

On Tue, Jan 11, 2011 at 2:51 AM, Wolfgang Lux <address@hidden> wrote:
> Gregory Casamento wrote:
>
>> Fred,
>>
>> I understand what you're saying, but I'm not sure what to do to
>> correct it.  I admit I had a feeling when I made this commit that
>> there must be a better way of doing this.  Something, perhaps, more
>> centralized is needed.    NSMenuView, as you suggested, does seem like
>> the right place.
>>
>> After some research it looks like _executeItemAtIndex:removeSubmenu:
>> might be a good candidate for this to be located centrally.
>>
>> Let me know what you think.
>
> I think the main problem is that NSMenuView's event tracking loop is too
> autistic. On one hand it accepts only mouse events during event tracking,
> but on the other hand it does not communicate this information to the window
> manager and so the user can freely move or close windows while the loop is
> active. This is absolutely inconsistent. Either we must also grab the mouse
> at the level of the X server during the event tracking, or the event
> tracking loop must handle AppKit generated events like window close and
> window move events. Personally, I'd prefer the latter, since grabbing would
> make it impossible to enter a debugging session in the event loop (at least
> without a second machine), but the former may be easier to implement.
>
> Wolfgang
>
>



-- 
Gregory Casamento - GNUstep Lead/Principal Consultant, OLC, Inc.
yahoo/skype: greg_casamento, aol: gjcasa
(240)274-9630 (Cell)