tag:blogger.com,1999:blog-876358347971598886.post6702744752656545652..comments2009-09-21T07:16:00.502+03:00Comments on nothingmuch's perl blog: Method Style Callbacksnothingmuchhttp://www.blogger.com/profile/03975438115490089158nothingmuch@woobling.orgBlogger2125tag:blogger.com,1999:blog-876358347971598886.post-6308562332723878142009-09-21T07:16:00.502+03:002009-09-21T07:16:00.502+03:00I think it's a side effect of the implementation.
...I think it's a side effect of the implementation.<br /><br />Method calls are two opcodes, one opcode to find a code ref, and then a normal subroutine invocation.<br /><br />The "method" opcode will simply do nothing if a the method being invoked is already a code reference and the "entersub" opcode doesn't really care what the arguments to the code ref are.<br /><br />BTW, the way that <a href="http://search.cpan.org/perldoc?autobox" rel="nofollow">autobox</a> works is by overriding the definition of that opcode in a lexical scope making method calls on non objects no longer be an error.nothingmuchhttp://www.blogger.com/profile/03975438115490089158noreply@blogger.comtag:blogger.com,1999:blog-876358347971598886.post-51184192271852670602009-09-20T21:31:13.618+03:002009-09-20T21:31:13.618+03:00Very interesting. I had never seen this $data->$co...Very interesting. I had never seen this $data->$coderef() notation before (when $data is not an object); but indeed it works! However, I couldn't find any mention of it in perldoc. So is this an undocumented side-effect of the current implementation, or is it really a Perl feature ?damihttp://www.blogger.com/profile/05175840149225722194noreply@blogger.com