Bifrost isn't a new version of Naiad. It's a combination of a new evaluation engine and the next iteration of the Naiad solvers expressed into this new architecture. The evaluation engine is based on a compiler technology, and the solvers are expressed as a collection of low-level nodes (like "add", "if" and "get data") assembled together in compounds the way that Lagoa was made with the built-in ICE node. That means you have both high level nodes and a low level visual languages to make your own stuff or modify what you get. But the data that can flow in the graph is opened, it isn't just a fix set of types and arrays like ICE, and the data changes during evaluation, and not just at "execute" points.
In my opinion it's a huge waste of time to go on analytical speculations based strictly based on wishful failure fantasies when all of these speculations will be disproven anyway in the medium future. Btw, if I read correctly you got your timeline wrong on ICE. We worked on it for between 2 and 3 years before its release XSI 7.0, not necessarily full time. Then we did a bit more work, like adding ICE modeling, with a few months of work here and there in the year that followed, with a skeleton team. We didn't work 6 years on it. On Tue, Aug 19, 2014 at 3:37 PM, Jason S <jasonsta...@gmail.com> wrote: > And I think it's more than enough to extrapolate that, unless things change, > 'Bifrost' is, and will quite likely remain like an elaborate Naiad > simulation engine.