Regardless of the outcome we arrive at, I am planning on doing more tests next week too see close they are, though remains to be answered the question of what we do if they start drifting apart I have a build of ie9 being handled as safari running...
On Mar 4, 2011 6:22 PM, <[email protected]> wrote: Having IE9 fallback to Safari sounds dangerous to me. If I put a fix in a Safari Impl class, am I supposed to know that the change will also apply to IE9? Normally, I don't test other browsers if I'm working on a permutation specified fix. And just because Showcase runs doesn't mean that all of the IE workarounds have gone away, or that all of the Safari workarounds make sense in IE9. Our plan of record is to have IE9 fallback to IE8. That way, IE9 will continue to work as it does today. In GWT user code, we will update deferred bindings to fix all GWT bugs, which should cover most of the low level stuff. User code is more likely to be making a view decision, such as using a different look in IE versus Safari. I don't like the idea of using a soft perm by default because it can't be undone by users who are willing to wait out a full compile for smaller code size. I'd be happier if GPE offered a fast compile option where users could specify user agents to compile if they just want to test in web mode. We just got rid of the gecko permutation, so we can replace it with IE9 and be back to where we were. Users can choose to soft perm however they want. And, once IE9 is released, it probably makes more sense to soft perm IE6/IE8 to reduce permutations. Lets get the best possible experience on modern browsers. Certainly we don't want to increase Safari code size to carry IE9! On 2011/03/04 20:56:48, fabiomfv wrote: http://gwt-code-reviews.appspot.com/1369807/diff/1/dev/... -- http://groups.google.com/group/Google-Web-Toolkit-Contributors
