I also remembered this morning that I speculated on the impact on compiler performance of replacing Jburg with a FalconJX “back-end” that outputs SWFs. It isn’t on my list of priorities since we have a compiler that works, but it just occurred to me that maybe Michael Schmalle and some others might find that interesting.
Just thinking out loud, there would be two or three proof points, one when you first get enough code to produce a SWF, another when you can produce the same SWF that the Jburg reducer/generator does, so you know you are taking into consideration some of the optimizations that the reducer can currently do (I believe it does constant folding) and much later, if ever, when other optimizations are warranted (tail-call, in-lining, etc). I wouldn’t be surprised if Jburg loses at the first proof point because I’ve stepped through the code and watched it make several function calls to output a single ABC sequence, but IIRC, the promise of a BURM is supposed to be in the optimizations (the rewrites). On the JS side, we send the JS to the Google Closure Compiler which is also doing rewrites. -Alex On 4/27/15, 10:35 PM, "Alex Harui" <aha...@adobe.com> wrote: > > >On 4/27/15, 10:20 PM, "Christofer Dutz" <christofer.d...@c-ware.de> wrote: > >>Hi, >> >> >>has anything significantly changed with FlexJS and Falcon from a >>packaging point of view? > >They are still two separate packages. The list of SWCs in FlexJS has >changed significantly as well as where the source is located. > >> >>I had thought that I head Alex say that JBurg was taken out of Falcon to >>make it more understandable ... is that true or was that a memory of a >>dream? I could see the jburg jar is still downloaded ... > >Jburg was and still used by the SWF compiler. Jburg was in the original >cross-compiler called FalconJS, but some folks thought it was too hard to >learn and created FalconJX that we use today. But it has been that way >since before the 0.0.1 releases. > >> >> >>I guess with FlexJS a lot has changed, but did that have an effect on the >>generated artifacts or are the changes related to the artifacts content >>and how the compiler deals with it? > >Not sure what you mean. > >-Alex >
