Re: Clang extension for BOOL and nil

[David Chisnall]

On 24 Aug 2013, at 12:13, "Vincent R." <address@hidden> wrote:

> Le 24.08.2013 13:02, Vincent R. a écrit :
>> Le 23.08.2013 12:32, David Chisnall a écrit :
>>> I've been meaning to make this change for a little while.  It makes a
>>> few bits of static analysis easier.  I don't have time to do this next
>>> week, but if you send a patch I'd be happy to commit it.
>> 
>> Please find a small patch, tested with clang but not gcc.
> 
> Don't know if it's related but when I launch Gorm on the terminal I have some 
> warning like :
> 
> address@hidden ~/objc2cs/gnustep/libobjc2/build $ Gorm
> Calling [NSBrowserDelegate -browser:numberOfRowsInColumn:] with incorrect 
> signature.  Method has address@hidden:address@hidden, selector has 
> address@hidden:address@hidden
> Calling [NSBrowserDelegate -browser:titleOfColumn:] with incorrect signature. 
>  Method has @address@hidden:address@hidden, selector has 
> @address@hidden:address@hidden
> Calling [NSTableViewDataSource -numberOfRowsInTableView:] with incorrect 
> signature.  Method has address@hidden:address@hidden, selector has 
> address@hidden:address@hidden
> Calling [NSOutlineViewDataSource -outlineView:numberOfChildrenOfItem:] with 
> incorrect signature.  Method has address@hidden:address@hidden@24, selector 
> has address@hidden:address@hidden@24
> Calling [NSOutlineViewDataSource -outlineView:child:ofItem:] with incorrect 
> signature.  Method has @address@hidden:address@hidden@28, selector has 
> @address@hidden:address@hidden@32
> Calling [NSOutlineViewDataSource -outlineView:numberOfChildrenOfItem:] with 
> incorrect signature.  Method has address@hidden:address@hidden@24, selector 
> has address@hidden:address@hidden@24
> Calling [NSOutlineViewDataSource -outlineView:child:ofItem:] with incorrect 
> signature.  Method has @address@hidden:address@hidden@28, selector has 
> @address@hidden:address@hidden@32
> Calling [NSOutlineViewDataSource -outlineView:child:ofItem:] with incorrect 
> signature.  Method has @address@hidden:address@hidden@28, selector has 
> @address@hidden:address@hidden@32
> Calling [NSOutlineViewDataSource -outlineView:child:ofItem:] with incorrect 
> signature.  Method has @address@hidden:address@hidden@28, selector has 
> @address@hidden:address@hidden@32
> Calling [NSOutlineViewDataSource -outlineView:child:ofItem:] with incorrect 
> signature.  Method has @address@hidden:address@hidden@28, selector has 
> @address@hidden:address@hidden@32
> Calling [NSOutlineViewDataSource -outlineView:numberOfChildrenOfItem:] with 
> incorrect signature.  Method has address@hidden:address@hidden@24, selector 
> has address@hidden:address@hidden@24
> Calling [NSOutlineViewDataSource -outlineView:child:ofItem:] with incorrect 
> signature.  Method has @address@hidden:address@hidden@28, selector has 
> @address@hidden:address@hidden@32
> Calling [NSOutlineViewDataSource -outlineView:child:ofItem:] with incorrect 
> signature.  Method has @address@hidden:address@hidden@28, selector has 
> @address@hidden:address@hidden@32

No, this is unrelated.  You should get compile-time warnings from Gorm as well 
- it still hasn't been completely updated to reflect the NS[U]Integer changes 
in the most recent release of -gui.  It's passing int where the receiver 
expects NSInteger (which, on LP64 platforms, means long or long long).  This 
mostly works correctly, because these values typically fit in an int and are 
passed in registers, so they're equivalent, but if any of the arguments needs 
to go on the stack (x86-64 passes up to 4 arguments in registers, including 
self and _cmd) then it's a problem.

David

-- Sent from my IBM 1620