Picked up last rounds of feedback from Joel (EmptyBinder weirdness), Thomas and James. Also checkstyle fixes.
Barring further objections, I'll submit this in the morning. There are more tests that could be implemented (Joel may have written them already), but I'd like to get in what I have. http://gwt-code-reviews.appspot.com/127801/diff/2005/2013 File reference/Microbenchmarks/src/com/google/gwt/reference/microbenchmark/client/TestDom.java (right): http://gwt-code-reviews.appspot.com/127801/diff/2005/2013#newcode58 Line 58: e.removeAttribute("id"); If I delete it I'll have to generate unique ids at runtime, and do string concatenation then too. Hard to imagine that being a win. I like the move-it-around idea, though. Now doing it after detach. Can't say I see a big difference that way, get by id still clearly loses. http://gwt-code-reviews.appspot.com/127801/diff/2005/2021 File reference/Microbenchmarks/src/com/google/gwt/reference/microbenchmark/client/WidgetCreation.java (right): http://gwt-code-reviews.appspot.com/127801/diff/2005/2021#newcode58 Line 58: return (new Date()).getTime(); Thanks, didn't know about that. http://gwt-code-reviews.appspot.com/127801/diff/2005/2021#newcode157 Line 157: widgets[widgets.length - 1].getElement().getOffsetTop(); On 2010/01/02 09:01:46, jamesr wrote: > You should query offsetTop on the <body> element instead of trying to grab some > child element - it'll still do layout on everything in the document. > Another problem with setTimeout(0) is that the the '0' will get rounded up to a > larger value that's not consistent between browsers - one of 4, 10, or 15ms. Done. http://gwt-code-reviews.appspot.com/127801 -- http://groups.google.com/group/Google-Web-Toolkit-Contributors
