Re: Creating a GNUstep desktop on Solaris

[Adam Fedor]

On Jan 6, 2005, at 9:35 AM, Andreas Höschler wrote:
> > Yes it is standard behavior, and ObjC does ignore a message to a nil 
> > object and just returns 'NULL'. The problem is Solaris chokes when 
> > you try to assign a 'NULL' to a structure. Even this should crash 
> > Solaris (if you could actually compile it, which you can't since the 
> > compiler is smart enough for this):
> >
> > struct foo {int i; int j;};
> >
> > struct foo bar;
> >
> > bar = (struct foo)0;
>
> Haven't tried to compile this example on the Solaris yet, however I 
> would consider the above code to be illegal, whether this works on 
> some platforms or not. Assign Null to a structure simply makes no 
> sence. I would expect that the compiler, or if it goes through the 
> runtime does not know how to handle this.

Perhaps I should be more specific, much as I love to delve into 
compiler and language internals :-)
Of course the above code will not compile (I said so, in fact). It's 
really just pseudo-code for a similar problem that occurs in 
Objective-C due to function pointer indirection. Thus when the compiler 
sees:

struct foo {int i; int j;};

@interface Foo(Category)
- (struct foo) structReturn;
@end

int main()
{
struct foo bar = [(Foo *)nil structReturn];
}

(which is perfectly legal code), it converts the method call into 
something like (warning: pseudo-like code here):

struct foo bar = objc_msg_send(nil, @selector(structReturn), args);

which in turn performs some magic to make sure the proper type gets 
returned for the selector. Most operating systems seem to handle this 
correctly even though it's certainly not clear what happens when you 
assign nothing (0, NULL, whatever) to a struct.


> IMHO this should be fixed in the TextEdit.app and GNUMail.app sources 
> and coders should make sure to always check a pointer for NULL before 
> assigning it to a struc, shouldn't they?

Yes.