http://www.javablogging.com/puzzler-the-catalyst-method-does-not-work/ Tracking surprises, features and bugs Sun, 05 Feb 2012 15:43:49 -0500 http://wordpress.org/?v=2.8 hourly 1 http://www.javablogging.com/puzzler-the-catalyst-method-does-not-work/comment-page-1/#comment-924 Yardena Sun, 08 Nov 2009 19:42:14 +0000 http://www.javablogging.com/?p=731#comment-924 And you can make the puzzler work again if you change "E extends Serializable" to "E extends Object & Serializable". And you can make the puzzler work again if you change “E extends Serializable” to “E extends Object & Serializable”.

]]> http://www.javablogging.com/puzzler-the-catalyst-method-does-not-work/comment-page-1/#comment-659 IBBoard Fri, 16 Oct 2009 14:21:37 +0000 http://www.javablogging.com/?p=731#comment-659 Is it really not because of the use of List or other collections? From an initial look it appears that a naive assumption of "multi-pass" erasure would change the previous example's methods to count(List) and count(List) before erasing down again to count(List) for both. In this example the erasure appears to convert the generic parameters to the only class it knows fits (Serializable for "E extends Serializable" or Object for "E"), and since they're the parameters directly then that's what is in the byte code. Is it really not because of the use of List or other collections? From an initial look it appears that a naive assumption of “multi-pass” erasure would change the previous example’s methods to count(List) and count(List) before erasing down again to count(List) for both. In this example the erasure appears to convert the generic parameters to the only class it knows fits (Serializable for “E extends Serializable” or Object for “E”), and since they’re the parameters directly then that’s what is in the byte code.

]]>