I would hope to motivate you to NOT delay the testing and release. First, Gordon did announce to the list his return from vacation, but poor soul, he was stuck with having to try to swallow all those weeks of svn-2-get migration and it's not clear from the postings that he ever succeeded in getting his own development and test methodologies to work in that brave new world.
Second, even at the reduced commitment that Adobe made to one-day-per-week, all of you involved know full well that if it's going to work, it's going to work when Gordon makes it work, and that really means that the round-the-clock testing framework that is now in place has to include the new compiler. Third, I understand your long commitment to this framework, and it is admirable that you make sure that every innovation meets your personal standard of trying very hard not to break anything that is out there. That is not, however, the approach that will help the community move forward with the ability to generate tighter code with the new compiler. Whenever you make it available, it will for as long as you like, be an option, not a replacement, so it cannot break anything. Anything that needs to be maintained with the mxmlc compiler will be maintained that way, but anything newly developed will be free to use the best compiler that works closer to the engine, that works best with the telemetry, that allows us to create ANE-integrated solutions with inlining, etc. Fourth anyone wanting to use these features will not give a tinker's damn about the compiler supporting some nice-to-have auto-binding mechanism that can be driven by drag-and-drop application development. It's not the UI where our optimization takes place, its in the interface to Workers, to Sockets, to ANE-supported device features, to the GPU. If you can give me the new compiler tomorrow and it completely fails to know how to auto-bind some mxml property sheet to my application architecture, I will take in in a heartbeat and spend my idle time writing binding to the UI the old-fashioned way -- like Mr. Houseman used to say, "They Earned It". It should be at the top of the priority list, not buried where it is as something nice to have circa 2015. On Wed, Jul 24, 2013 at 9:30 PM, Alex Harui <aha...@adobe.com> wrote: > > > On 7/24/13 6:45 PM, "Terry Corbet" <terry.cor...@gmail.com> wrote: > > What has become of > >anything having to do with Gordon's getting it possible to compile an > >ApacheFlex/AIR application with the new compiler so that we can take > >advantage of the performance optimizations it enables? With all the > >coverage that has beena dded to the nightly builds, where is the one that > >shows what happens when you attempt to do it all with the new compiler? > > I believe Gordon is still on vacation. He is not full-time on the > compiler anyway, so his progress is limited by time. > > Meanwhile, I am shoving a decently sized production application through > the new compiler. The AS handling is pretty sound, but the MXML handling > definitely needs work, especially involving databinding. Once I can get > the new compiler to generate a SWF that runs it will be time to consider > shoving our entire mustella suite through the new compiler. But that's a > ways off. And then we should try to integrate it as best as possible with > Flash Builder, which is even more work. > > -Alex > >
