Path: csiph.com!x330-a1.tempe.blueboxinc.net!newsfeed.hal-mli.net!feeder1.hal-mli.net!news.linkpendium.com!news.linkpendium.com!newsfeeds.ihug.co.nz!lust.ihug.co.nz!ihug.co.nz!not-for-mail From: Lawrence D'Oliveiro Newsgroups: comp.lang.java.programmer Subject: Re: Java generics and type erasure Followup-To: comp.lang.java.programmer Date: Thu, 26 May 2011 13:59:25 +1200 Organization: Geek Central Lines: 18 Message-ID: References: <9d4c2b16-beb5-40b1-87a2-f03e971efeed@k17g2000vbn.googlegroups.com> NNTP-Posting-Host: 118-92-95-178.dsl.dyn.ihug.co.nz Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 8Bit X-Trace: lust.ihug.co.nz 1306375165 26884 118.92.95.178 (26 May 2011 01:59:25 GMT) X-Complaints-To: abuse@ihug.co.nz NNTP-Posting-Date: Thu, 26 May 2011 01:59:25 +0000 (UTC) User-Agent: KNode/4.4.7 Xref: x330-a1.tempe.blueboxinc.net comp.lang.java.programmer:4590 In message , Susan Calvin wrote: > On Thu, 26 May 2011 10:18:20 +1200, Lawrence D'Oliveiro wrote: > >> In message , Susan Calvin wrote: >> >>> Perhaps there should be a compile flag that turns on the >>> legacy-compatible behavior for use when compiling 1.4 and older >>> sources, but which is off by default? >> >> What happens when you mix code compiled with that flag, with code that >> was compiled without? > > Why, nothing, of course, since generics don't exist at run-time. If theat were the case, then there would be no backward-compatibility issue, would there? And the whole machinery of “raw” types could just disappear in a puff of un-necessity.