Re: Re(2): Strange crash, CFBundle*
Re: Re(2): Strange crash, CFBundle*
- Subject: Re: Re(2): Strange crash, CFBundle*
- From: Dan Wood <email@hidden>
- Date: Tue, 11 Dec 2001 11:23:10 -0800
Douglas, is this also the case with NSBundles? I'm currently
working on a problem where I'm trying to completely deallocate
an NSBundle (because the version on disk has been updated to a
new version), and then load it again. Maybe this "unique-ing"
is getting in the way. Any way to cause a bundle to be
re-fetched from disk?
Dan Wood
On Tuesday, December 11, 2001, at 09:16 AM, Douglas Davidson wrote:
On Tuesday, December 11, 2001, at 12:31 AM, Jens Bauer wrote:
Here's a shortened down version of code I did not write myself:
static bool myfunction()
{
CFBundleRef bundle;
CFURLRef resources;
bundle = CFBundleGetMainBundle();
if(!bundle)
{
return(FALSE);
}
resources = CFBundleCopyResourcesDirectoryURL(bundle);
--> CFRelease(bundle);
if(!resources)
{
return(FALSE);
}
...
...
}
You mention that I should not release something that was
returned to me by a
"get" function. Is the above CFRelease(bundle) wrong then ?
Yes! You will see random crashes if you release the result of
CFBundleGetMainBundle().
Remember the rules: exactly one release for each API call with
"create", "copy", or "retain" in it.
One further point is that you don't actually need to release
bundles that you don't want to be unloaded--and the main bundle
would fall into this category. Bundle instances are uniqued,
so not releasing a bundle doesn't cause a leak if that bundle
wasn't going to go away.
Douglas Davidson
_______________________________________________
cocoa-dev mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/cocoa-dev
Do not post admin requests to the list. They will be ignored.
--
Dan Wood
email@hidden
http://www.karelia.com/
Watson for Mac OS X:
http://www.karelia.com/watson/