Re: [gtk-osx-devel] [PATCH] gtk-mac-integration reference counting

[John Ralls <jralls ceridwen us>]

On Jan 28, 2012, at 2:09 PM, Richard Procter wrote:

> Hi John, 
> 
> I ran the latest clang static analyzer [1] over gtk-mac-integration - here's a patch for a couple of ref-counting issues it found. Nothing that looks crucial but nice to have all the same. I've compiled it but have not tested it and as I rarely use objective-c it's worth casting your eye over it. 
> 
> My understanding is that class methods return an autoreleased object, so, e.g. [NSString stringWithUTF8String:] doesn't require an explicit release [2]. 
> 
> I've also renamed gtk_mac_image_from_pixbuf() -> gtk_create_cgimage_from_pixbuf() to indicate the returned CGImage is owned by the caller. This function isn't exposed in the bindings. 
> 
> Lastly, their scan-build tool is very useful but doesn't quite run out-of-the-box with gtk-mac-integration as I couldn't find a way to force it to parse objective-c, except for renaming the source files .objectivec, which automake then choked on. So I just ran: 
> 
> $ for x in *.c; do ./checker-260/bin/clang --analyze -x objective-c [whole bunch of includes] $x; done 
> 
> ... manually, and did without the pretty HTML reporting. 

Richard,

Cool, thanks, I'll have a look at it -- But:

What SDK are you analyzing against? Apple has some new memory-management features in 10.7, and they did some major cleanup for 10.6. All well and good, but gtk-mac-integration has to work correctly with 10.4.

Functions need to keep the gtk_mac name (in fact, 1.1 will rename them to gtkmac_) for compatibility with 
gobject-introspection.  

checker-260/bin/clang? You're not using Apple's?

Regards,
John Ralls