gnustep-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[GDL2/base] RFC: NSControl classes


From: David Ayers
Subject: [GDL2/base] RFC: NSControl classes
Date: Fri, 02 May 2003 19:01:06 +0200
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3) Gecko/20030312

Hello,

OK, now let's get back to using the undocumented "NSControl" classes:
NSFault
NSFaultHandler
NSClassDescription
to be the superclasses of the coresponding EOClasses, i.e. ...

http://mail.gnu.org/archive/html/bug-gnustep/2003-04/msg00034.html

As NSClassDescription is the only documented class, it happens to be only class currently available from gnustep-base.

- I'm very unsure, whether we should support these classes in GDL2 at all, but I'm willing to try. Manuel?
- If we support it, should we add the NSClasses to gnustep-base? Richard?
- Independent of where they would be added, should GDL2 always use the superclasses or should we try to configure GDL2 to use them explicitly and not use them by default?

My personal opinion is:
- GDL2 is trying to be WO45 compatible and these NSClasses do not exist in that scope.
- adding support for them will complicate the code.
- the NSClasses implementation can't use any of the GDL2Classes/features. This can make the implementation rather awkward. - GDL2 looses access to some of the cached information in the NSClasses (EO/NSClassDescription)

And most of all, I don't see any real merit in using NSClasses. I acknowledge the fact that some might believe that running GDL2 on OS X makes them feel better when using as many Foundation classes as available. But because we don't know what these classes really do and how the would fit into a GDL2 mechanics, I believe the opposite is true, because we loose control of what is going on.

So I would vote against adding these classes, but if overruled, I would volunteer to try to integrate them so they are as little invasive as possible.

In the (as I fear almost likely) case that in June Apple will present EOF 4 which makes use of these classes (and adds documentation in Foundation for these and possibly other classes), we may have to consider this again. My vote would be leave GDL2 WO4.5 compatible and possibly start a new GDL3 library to be EOF current compatible. Personally I would probably focus on GDL2 rather than GDL3. But I must admit that if gnustep-base tracked Cocoa once these classes are documented and used them internally, keeping GDL2 stable (once we get that far :-) ) with gnustep-base might be a daunting task. Also explaining to users why gnustep-base uses NSClasses/Exceptions/Notification/OtherKeys and GDL2 uses the EOVersions and figureing out how to write code that supports both simultaniously, won't make it any eaiser. But WO 4.5 compatibility is very important to me. (And others porting WO4.5 projects.)

I think we should postpone any changes in that direction until the end of June and then rediscuss this.

Cheers,
Dave






reply via email to

[Prev in Thread] Current Thread [Next in Thread]