http://www.mikeash.com/?page=pyblog/friday-qa-2011-05-06-a-tour-of-mablockclosure.html#commentsmikeash.com Recent CommentsTue, 12 May 2026 00:37:31 GMTPyRSS2Gen-1.0.0http://blogs.law.harvard.edu/tech/rsshttp://www.mikeash.com/?page=pyblog/friday-qa-2011-05-06-a-tour-of-mablockclosure.html#commentsThe numbers in method signatures are the offsets within the argument storage. A '+' preceding the number indicates the argument belongs in a register.0233819b1ff557322476c1527120ff4eTue, 10 May 2011 15:16:37 GMThttp://www.mikeash.com/?page=pyblog/friday-qa-2011-05-06-a-tour-of-mablockclosure.html#commentsThat's interesting to know. This seems like a case of the perfect being the enemy of the good. Yeah, @encode is pretty awful, but it's better than nothing, and how long would we have to wait for Apple to come up with their less-horrible replacement? I'd love to see one, but in the meantime, @encode gets the job done.6bfc5ff4c0de1365e06456c575239c9eFri, 06 May 2011 18:45:39 GMThttp://www.mikeash.com/?page=pyblog/friday-qa-2011-05-06-a-tour-of-mablockclosure.html#commentsAs I understand it, Apple didn’t add type metadata to blocks – David Chisnall did, for Étoilé’s scriping and Smalltalk layer. The Apple guys didn’t want to do it because @encode() sucks and they want to eventually add something less horrible instead. <br /> <br />A consequence of this is that the metadata isn’t generated by apple-gcc.fab620a951cc6db34bc54a9185c8a21dFri, 06 May 2011 18:42:40 GMT