Does anybody know roughly how long it will be till Twisty 2.0 is ready for the Android market?
Pasting my reply from the other thread:
"I’m the main maintainer for Twisty, and I apologize for its anemia. We’re in the middle of a complete rewrite – instead of a poor port of the java ‘zplet’ to Android, the new version is actually an attempt to write a clean GLK layer using Java’s JNI interface. The rewrite is actually using ‘nitfol’ (C language) as its backend interpreter, and will eventually use glulxe or git for glulx games. I simply need more round tuits to get this thing done. :-/ "
Until they can run Glulx I view them all as rudimentary experiments with great promise that is not yet realised. (With the exception of that TADS thing which is an interesting tangent. Seems like it’s a pretty big task to implement a Glulx interpreter on top of Z-code, if adding TADS instead is the easier target; but I’m just guessing at the motivation for that.)
I’m pleased to hear about the roadmap for Twisty. I would send you a round tuit, if I had got same. 8)
Glulx, Z-Machine, and TADS all run under different virtual machines, so you need a different interpreter for each. The main opportunity for savings comes from reusing the platform specific code for input and output.
How easy that is depends on how many assumptions in the platform code are tied to the original VM, and how many of the unique / incompatible features of the other formats you intend to support.
Dannii’s IFVM and Wei-ju’s ZMPP are exceptions to the “one interpreter per VM” rule - they have abstracted both common I/O code and common VM code, and can plug in modules that support other VMs.