Update the compatibility version check for the ObjC API's InitConstructorSupport to use dyld_get_program_sdk_version().

[Mark Lam]

Patch coming.

[Mark Lam]

<rdar://problem/25093101>

[Mark Lam]

Created attachment 277419 [details]
proposed patch.

[Geoffrey Garen]

Comment on attachment 277419 [details]
proposed patch.

r=me

[Mark Lam]

Thanks for the review.  Landed in r200114: <http://trac.webkit.org/r200114>.

Comment on attachment 277419 [details]
proposed patch.

View in context: https://bugs.webkit.org/attachment.cgi?id=277419&action=review

> Source/JavaScriptCore/API/JSWrapperMap.mm:50
> +#if __has_include(<mach-o/dyld_priv.h>)
> +#include <mach-o/dyld_priv.h>
> +#endif
> +extern "C" uint32_t dyld_get_program_sdk_version();

It’s incorrect to use __has_include to check for private headers. Instead, check for USE(APPLE_INTERNAL_SDK).

> Source/JavaScriptCore/API/JSWrapperMap.mm:55
> +static const uint32_t webkitFirstSDKVersionWithInitConstructorSupport = 0x80000; // iOS 8.0.0
> +#elif PLATFORM(MAC)
> +static const uint32_t webkitFirstSDKVersionWithInitConstructorSupport = 0xA0A00; // OSX 10.10.0

It doesn’t make sense to use “webkit” in these names, which aren’t WebKit versions.

[Mark Lam]

(In reply to comment #5)
> It’s incorrect to use __has_include to check for private headers. Instead,
> check for USE(APPLE_INTERNAL_SDK).
> 
...
> It doesn’t make sense to use “webkit” in these names, which aren’t WebKit
> versions.

I'll address these in https://bugs.webkit.org/show_bug.cgi?id=157096.

[Alexey Proskuryakov]

Comment on attachment 277419 [details]
proposed patch.

View in context: https://bugs.webkit.org/attachment.cgi?id=277419&action=review

> Source/JavaScriptCore/API/JSWrapperMap.mm:665
> +    static uint32_t programSDKVersion = 0;
> +    if (!programSDKVersion)
> +        programSDKVersion = dyld_get_program_sdk_version();

Do we need to keep trying if we get a zero value once? If not, this should be written as:

static uint32_t programSDKVersion = dyld_get_program_sdk_version();