It's a nice proof-of-concept, but the more that I think about it the more that I'm convinced that latency will completely kill it for anything other than local network use. I think something like the 280North model is a lot nicer, going right back to the NeWS model of views on the client, models on the server.
This weekend I've been playing with the clang libraries and working on compiling Objective-C to JavaScript. It's actually a lot easier than I was expecting - the clang AST is very easy to work with. There are some things in C that won't map at all to JS, but I think we can get most of ObjC working without a huge amount of effort. Ideally, I'd like to then implement DO top of WebSocket (maybe with some fallback to async XML requests), so we can just compile view classes to JS and run them in the browser, and run model classes on the server, without developers having to think too hard about where the separation goes. David On 17 Mar 2011, at 20:41, Quentin Mathé wrote: > Hi, > > Alexander Larsson (GNOME developer) has implemented a very cool idea to > support remote applications. The basic idea is to stream drawing commands > from a running application over the network, then on the other side to render > them in a web browser with a canvas element. > > It's something Nicolas, David and I have discussed several times and we'd > like to implement either at Étoilé or GNUstep level. > > I read about it on back in November, but it was on Slashdot today, so I > thought it was good idea to post the link were this GTK web backend is > discussed: http://blogs.gnome.org/alexl/ > > Cheers, > Quentin. > > > > _______________________________________________ > Etoile-dev mailing list > Etoile-dev@gna.org > https://mail.gna.org/listinfo/etoile-dev -- Sent from my IBM 1620 _______________________________________________ Etoile-dev mailing list Etoile-dev@gna.org https://mail.gna.org/listinfo/etoile-dev
