I see 2 areas that coud use some effort 1. Upgrade the GWT widget set to the latest GWT 2.5 2. Work on the compiler optimzations/redesign on pys or pyjaco - an orthogonal question 3. Work on the java2python translator to fully automate and/or generate more optimal python code.
I am leaning towards tasking the resource to 1 above and maybe 3( towards more complete automation if possible). And from what I am hearing, 2 above is usable now, and I could wait to see where pyjaco heads before deciding on what needs to be done, at a later date. What do you think? Sarvi On Friday, December 21, 2012 7:15:44 AM UTC-8, Lex Berezhny wrote: > > On Fri, Dec 21, 2012 at 9:35 AM, Sarvi Shanmugham > <[email protected]<javascript:> > > wrote: > >> So I am trying to understand what drove pyjaco to fork off and not build >> on the compiler that comes with pyjamas?? >> > > My hunch and something I've experienced myself is that the pyjs compiler / > runtime has a lot of hacks and patches to make it work. This is partly > because as I said, people are actually using pyjs in production systems. > There is a lot of effort into making sure we don't brake existing APIs and > functionality. It's very difficult to overhaul sections of the > compiler/runtime without breaking that compatibility so instead of doing > major rewrite/refactorings we end up patching things. > > It wouldn't surprise me (especially with the comments I saw on their > mailing list regarding a rewrite) that they would rather start new and > designed the system the way they want than to spend time trying to fix an > older system. > > The race becomes between how quickly they can get their > new, hopefully much better designed compiler, to the level of pyjs vs how > quickly we can refactor and add new features to pyjs while maintaining > backwards compatibility. > > It seems you have the following choice to make: 1) go with pyjs, which > works right now, something people are already using in production but with > a complex code base that needs refactoring or 2) jump on the pyjaco > bandwagon and help to get their compiler and tools to a working state with > better design and hopefuly this will happen within your required timeframe > to start your own project. > > If you need something now, you don't have much of a choice, pyjs is used > in production systems today. If you can put a few months of full time > development towards either framework then you can decide between helping to > refactor pyjs or working on a new pyjaco compiler. > > That's my estimation of this situation. > > - lex > --
