bad kCFTaggedObjectID_Integer definition ?

[tshirtman]

https://github.com/kivy/pyobjus/blob/master/pyobjus/pyobjus_conversions.pxi#L188

defines it as (3 << 1) + 1

while all references i seem to find defines it as (1 << 1) + 1

http://www.crest.iu.edu/projects/conceptcpp/docs/html-ext/CGObjCMac_8cpp_source.html#l01623
http://bou.io/talks/TaggedPointersTests.html#4

--- Want to back this issue? **[Post a bounty on it!](https://www.bountysource.com/issues/7064670-bad-kcftaggedobjectid_integer-definition?utm_campaign=plugin&utm_content=tracker%2F77137&utm_medium=issues&utm_source=github)** We accept bounties via [Bountysource](https://www.bountysource.com/?utm_campaign=plugin&utm_content=tracker%2F77137&utm_medium=issues&utm_source=github).

[tshirtman]

Apparently while this value is indeed (1 << 1) + 1 on 10.7.5, it changed to (3 << 1) + 1 in 10.8.5, so the value probably needs to be imported from the right place at compile time.

[tshirtman]

After reading on tagged pointers, it seems any info about them−aside the general idea−is fragile and undocumented, because intended to be private, so I'm wondering if the solution wouldn't be just to instanciate an NSNumber at import time, with a very low value (i guess 1 would do), and take the kCFTaggedObjectID_Integer value from the LSB of the returned pointer. Since there are a few other types that can be represented as tagget pointers, (a potential list of which can be found here http://stackoverflow.com/questions/20362406/tagged-pointers-in-objective-c), we could instanciate values of them that should fit and check for their last bit to get the correct class, at runtime.