commit 80fb20b69eaae6ed4cf1f3c869632d01448559c0
author gwt.team.scottb <gwt.team.scottb@8db76d5a-ed1c-0410-87a9-c151d255dfc7> Thu Jun 21 21:01:30 2007 +0000
committer gwt.team.scottb <gwt.team.scottb@8db76d5a-ed1c-0410-87a9-c151d255dfc7> Thu Jun 21 21:01:30 2007 +0000
tree 7f4505fcab9d507fd419d5519f51c0fe5c0c7130
parent d66e92e856c4a517637f85732906ad82696a908d

Fixes issue #1132.  In r956 I put in a fix to make static initializers chain to their superclass static initializers.  Unfortunately, the change I put in resulted in the wrong evaluation order.  I'm backing out much of that change, and instead fixing the same problem a different way.

This change adds an explicit chaining call from a subclass clinit to a superclass clinit very early in AST generation.  It's actually a much more straightforward fix. This may produce slightly more code in the short term, because the old implementation would essentially elide empty subclass ctors.  When we we get better inlining in place any differences should go away.

Found by: ispeters, jcai99
Review by: mmendez


git-svn-id: https://google-web-toolkit.googlecode.com/svn/trunk@1197 8db76d5a-ed1c-0410-87a9-c151d255dfc7