Hi Daniel, patch is attached. I couldn't really figure out how to work the
format-patch thing, but did the next best thing.
On Wed, Aug 23, 2017 at 10:30 AM, Daniel Ferreira (theiostream) <
[email protected]> wrote:
> Hi Stefan, sorry for taking a day to answer; I was absent yesterday.
>
> Can you generate a patch for your changes with `git format-patch` and
> send them to me? I can embed them into my pull request and make them
> look nice with Git history. Afterwards, you could merge it.
>
> -- Daniel.
>
> On Tue, Aug 22, 2017 at 4:56 PM, Stefan Bidigaray <[email protected]>
> wrote:
> > I have what I believe is a solution to this PR, however, I do not
> actually
> > know how to go about fixing the PR and pushing the changes. I was
> > successfully able to get Daniel's branch and made the modifications. Now
> I'm
> > not sure how to commit and push the changes to the repo. Can anyone give
> me
> > nudge in the right direction?
> >
> > On Tue, Aug 22, 2017 at 2:41 PM, Stefan Bidigaray <[email protected]>
> > wrote:
> >>
> >> So CFBridgingRelease() essentially calls -autorelease, and
> >> CFBridgingRetain() -retain? I kind of get why, but only after thinking
> about
> >> it for an hour.
> >>
> >> On Tue, Aug 22, 2017 at 12:56 PM, Fred Kiefer <[email protected]>
> wrote:
> >>>
> >>>
> >>> > Am 22.08.2017 um 17:15 schrieb Stefan Bidigaray <
> [email protected]>:
> >>> >
> >>> > Hi everyone,
> >>> > This message is mostly for Daniel, but anyone with some experience
> >>> > using these functions should not hesitate to add their thoughts.
> >>> >
> >>> > Daniel, I just noticed your pull request on github last night. I'm
> good
> >>> > with most of it, my only concern is the bridging functions. I made a
> comment
> >>> > last night, but later discovered I misunderstood their purpose. I'd
> like to
> >>> > get your pull request merged, and want do discuss it before we move
> forward.
> >>> >
> >>> > My understanding of the documentation for CFBridgingRelease and
> >>> > CFBridgingRetain is that they are essentially no-ops, unless ARC is
> in
> >>> > effect. And even when ARC is being used they do not do anything
> other than
> >>> > tell the compiler that the object is now controlled by ARC. With
> that in
> >>> > mind, these two "functions" should do nothing other than cast the CF
> object
> >>> > as an ARC aware object, and nothing at all if ARC is not
> implemented. Does
> >>> > this sound reasonable? Another part of this conversation is where do
> these
> >>> > functions belong. According to the Apple documentation, they are
> part of
> >>> > Foundation, so should they not be in Base, instead?
> >>> >
> >>> > As for CFAutorelease, I think this is the function where
> >>> > objc_autoreleaseReturnValue() needs to be used. The documentation
> for this
> >>> > function is nonexistent, so I'm just guessing at what it really does.
> >>> >
> >>> > Any thoughts? Anyone?
> >>>
> >>>
> >>> Stefan,
> >>>
> >>> I think you are mostly correct. Apple has this stuff in Foundation, but
> >>> we don’t build base on top of CoreFoundation, so we need to have it the
> >>> other way around. In ARC-mode these two functions should call the
> compiler
> >>> primitives David pointed to. Where you are incorrect is the
> non-ARC-mode.
> >>> There Apple seems to implement these functions as retain and
> (auto-)release.
> >>> Don’t ask me why, your interpretations makes more sense to me.
> >>>
> >>> Fred
> >>>
> >>>
> >>>
> >>> _______________________________________________
> >>> Gnustep-dev mailing list
> >>> [email protected]
> >>> https://lists.gnu.org/mailman/listinfo/gnustep-dev
> >>
> >>
> >
> >
> > _______________________________________________
> > Gnustep-dev mailing list
> > [email protected]
> > https://lists.gnu.org/mailman/listinfo/gnustep-dev
> >
>
diff --git a/Headers/CoreFoundation/CFBase.h.in
b/Headers/CoreFoundation/CFBase.h.in
index 47fdead..8d71a44 100644
--- a/Headers/CoreFoundation/CFBase.h.in
+++ b/Headers/CoreFoundation/CFBase.h.in
@@ -451,14 +451,14 @@ CF_EXPORT void *_CFBridgingRelease (CFTypeRef cf);
CF_EXPORT CFTypeRef _CFBridgingRetain (void *obj);
#if __has_feature(objc_arc)
-#define CFBridgingRetain(x) _CFBridgingRetain((__bridge void *)(x))
-#define CFBridgingRelease(x) (__bridge id)(_CFBridgingRelease(x))
+#define CFBridgingRetain(x) (__bridge_retained CFTypeRef)(x)
+#define CFBridgingRelease(x) (__bridge_transfer id)(x)
#elif __OBJC__
#define CFBridgingRetain(x) _CFBridgingRetain((void *)(x))
-#define CFBridgingRelease(x) (id)_CFBridgingRelease(x)
+#define CFBridgingRelease(x) (id)_CFBridgingRelease((x))
#else
#define CFBridgingRetain(x) _CFBridgingRetain((void *)(x))
-#define CFBridgingRelease(x) _CFBridgingRelease(x)
+#define CFBridgingRelease(x) _CFBridgingRelease((x))
#endif
#endif
/** \} */
diff --git a/Source/CFRuntime.c b/Source/CFRuntime.c
index 875c9ba..55f7d4e 100644
--- a/Source/CFRuntime.c
+++ b/Source/CFRuntime.c
@@ -387,20 +387,23 @@ CFRetain (CFTypeRef cf)
CFTypeRef
CFAutorelease (CFTypeRef cf)
{
- // FIXME: this is leaking
- return cf;
+#if __has_feature(objc_arc)
+ return objc_autoreleaseReturnValue(cf);
+#else
+ CF_OBJC_FUNCDISPATCHV(CFTypeRef, CFTypeRef, cf, "autorelease");
+#endif
}
void *
_CFBridgingRelease (CFTypeRef cf)
{
- return objc_autoreleaseReturnValue(cf);
+ CF_OBJC_FUNCDISPATCHV(CFTypeRef, void *, cf, "autorelease");
}
CFTypeRef
_CFBridgingRetain (void *obj)
{
- return objc_retain(obj);
+ CF_OBJC_FUNCDISPATCHV(void *, CFTypeRef, obj, "retain");
}
const void *
_______________________________________________
Gnustep-dev mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/gnustep-dev
