[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: EOKeyValueCoding
From: |
Richard Frith-Macdonald |
Subject: |
Re: EOKeyValueCoding |
Date: |
Sun, 24 Feb 2002 16:18:12 +0000 |
On Sunday, February 24, 2002, at 03:16 PM, Manuel Guesdon wrote:
On Sun, 24 Feb 2002 15:46:24 +0100 David Wetzel <address@hidden>
wrote:
| The problem we had is fixed with the code you send me. Thanks.
| Do you really think EOKeyValueCoding should be in EOF? I think it
should be in the base lib.
|
| Why? Imaginge someone wants GSWeb programs without EOF.
| Or tools without GSWeb...
There's a discussion about this on gnustep-dev mailing-list.
I'd wanted to avoid to have KV implementations parts in too much
libraries (one part in base, some additions in GSWeb, some
others in EOF,...). I thing it's better to have all additions related
to base objects (NSObject, NSArray, NSDictionary) in
one place (_with_ all the needed handlers to enable derived objects
customization and minimize code+possible bugs in
these customizations) and put specific (EOF, GNUstepWeb,... related)
parts in there own libraries.
I was thinking about GNUstep additions and MacOSX portage. May be it
could be better to put all 'base' GNUstep additions in a separate
library
(not in base, not in gsweb, not in EOF, not in extensions) so we have a
base which have same features as MacOsX and
people who want to use GNUstep additions for GNUstep and/or MacOSX
applications can use this library. As you say, it
could be a good place to put some code so GNUstepWeb and others may
work without EOF.
Anyway, I'm not maintainer of core so I let concerned people choose.
I'd just like to know the final decision soon so I can finish
modifying GNUstepWeb and gdl2 for these points.
I think we have a very simple rule for where to put apple stuff ...
maintain compatibility with apple.
If we put code in the wrong places, so it's either in the wrong headers
or linked in as part of the
wrong frameworks (as far as apple compatibility is concerned), then we
are making extra work for anyone
porting the code, and any end users could resonably report it as a bug
and ask us to fix it.
A problem then only arises where apple have abandoned an API, but we
want to keep it ... in which case
I think we should probably aim to keep it where they had it.
I don't want to put every API apple drop into the base and gui
libraries ... that would just result in
bloated libraries with lots of APIs that most people don't want. In
fact, far from adding to the base
library, I'd rather try to factor out some GNUstep code (MIME, XML, SSL)
into separate bundles
(or something similar) included as part of the base package but capable
of being used independently.
I think this means the basic KVC stuff should be in the base library,
with webobjects extensions in gsweb
and eof extensions in gdl2.
That being said, since I'm still not sure of the status of gdl2, it may
be that some of the stuff from
there has nowhere to go but the base library.
- Re: EOKeyValueCoding, Manuel Guesdon, 2002/02/24
- Re: EOKeyValueCoding,
Richard Frith-Macdonald <=
- Re[2]: EOKeyValueCoding, Manuel Guesdon, 2002/02/24
- Re: Re[2]: EOKeyValueCoding, Helge Hess, 2002/02/24
- Re[4]: EOKeyValueCoding, Manuel Guesdon, 2002/02/24
- Re: Re[4]: EOKeyValueCoding, Helge Hess, 2002/02/24
- Re[6]: EOKeyValueCoding, Manuel Guesdon, 2002/02/24
- Re: Re[6]: EOKeyValueCoding, Philippe C.D. Robert, 2002/02/24
- Re: Re[6]: EOKeyValueCoding, Richard Frith-Macdonald, 2002/02/24
- Re: EOKeyValueCoding, Adam Fedor, 2002/02/24
- Re: EOKeyValueCoding, Nicola Pero, 2002/02/24
- Re[2]: EOKeyValueCoding, Manuel Guesdon, 2002/02/25