This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: FSF GCC, #import and #pragma once

> Has Zack succeeded in changing the language definition to not include 
> #import?

Language definition ?  The first time I heard of such a thing. :-)

Apple has got his own private "language definition", if this is what you
are referring to, and of course nobody else but Apple can change that.  
Which makes sense, as that "language definition" actually describes the
Apple GCC compiler shipped on Apple machines, and only Apple can change

As a matter of fact, your "language definition" is full of references to
your own proprietary development environment.  It implies the NeXT runtime
(it also describes in full details the NEXT runtime in the same document).  
It implies the Apple/Cocoa foundation library.  If you ever read the
document, you can't seriously call that a vendor-neutral language
definition.  It's a manual documenting part of your proprietary Cocoa
development environment, and can't really be used as a vendor-neutral
language definition.  Last time I looked at it, the name of the document
had even explicitly something like 'Inside MacOSX' in the title (I don't
know if that has been changed).

The best language definition for the GNU Objective-C language/runtime is
unfortunately obtained looking at its reference implementation, which is
the GCC Objective-C compiler/runtime itself. :-)

I suppose we need a GNU language definition for the GNU Objective-C
language, to drop this dependency on your own proprietary literature.  
Our users, after all, also deserve a free manual describing our free
compiler/runtime.  It would probably be a ten/twenty page document, so it
shouldn't take long to write from scratch - a week maybe.  I'll try to
find the time to produce such a document in the next weeks, as this issue
seems to pop up over and over again.

> [ ... talking of gnustep users ... ]
> And what % of Objective-C users does the above represent? 90%, 99%?  I 
> suspect not.

Ok - fair enough - yes, that's a minority of the whole Objective-C users.  

Yet, it's almost all of the GNU runtime / free software Objective-C users
except for the Swarm project.

> I think #import removal has to be OKed by Stan and Zem before it is 
> done.

I found the other reply, saying the GCC SC need to decide that, much more
balanced and acceptable, as the GCC SC is supposed to represent a variety
of interests, rather than just Apple's own interests.

But let me say that there is another subtle way of removing #import, which
is simply by not maintaining it.  If you have a feature which doesn't
work, that's basically equivalent to not having it.

As far as I know, nobody is willing to maintain #import working across all
platforms / filesystems that GCC is working on.  Which at the end of the
day is why the precise reason why I agree with dropping it.

Apple is willing to maintain in working on Apple, but what about all the
other platforms / filesystems ?

No matter what all you powerful decision makers at Apple decide, unless
someone executes them by actually porting and maintaining #import, they
are pointless.  As far as I know, nobody is willing to do that work except
yourself on your own platform.

Maybe we could keep #import for the NEXT runtime, since you are willing to
maintain it, and drop it for the GNU runtime, since nobody is willing to
maintain it on GNU.

I don't care what it is decided, I'm happy with whoever powerful authority
is supposed to decide that, but I'd like to state clearly that I'm not
willing to do any work on #import (as far as I'm concerned, it has no
relevance to the GNU Objective-C language) and I don't know anyone willing
to do that work.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]