Post AQHqCVDjodeiFvS9fU by jakintosh@merveilles.town
 (DIR) More posts by jakintosh@merveilles.town
 (DIR) Post #AQHqCVDjodeiFvS9fU by jakintosh@merveilles.town
       2022-11-24T01:57:18Z
       
       0 likes, 0 repeats
       
       got my little virtual CPU to a baseline state of having instructions that can be assembled from text and executed. now we get to the interesting parts... #theFurnace
       
 (DIR) Post #AQHqCWAEJARZBKorq4 by jakintosh@merveilles.town
       2022-12-05T05:22:21Z
       
       1 likes, 0 repeats
       
       I've been working on expanding my assembly language, both in terms of design and also implementation. The language borrows from forth, but splits between callable routines and somewhat powerful macros that allow label parameters for hopefully very readable but very reusable code segments.It is almost completely assembling, but I'm working to unravel the mess of text and semantic parsing... #theFurnace
       
 (DIR) Post #AQHqCXLxt1QOs1Jklk by jakintosh@merveilles.town
       2022-11-24T02:20:09Z
       
       0 likes, 0 repeats
       
       some of the main inquiries this project is exploring threaded below:
       
 (DIR) Post #AQHqCZIUetputpi0dU by jakintosh@merveilles.town
       2022-11-24T02:20:58Z
       
       0 likes, 0 repeats
       
       1. is it interesting to have access to a third stack on a stack machine? what about if instead of a third stack, it was a queue? does this open interesting new doors for elegant assembly code, or is it just unnecessary hardware/instruction complexity that could just be some functions instead?
       
 (DIR) Post #AQHqCb425dSQNYxUNE by jakintosh@merveilles.town
       2022-11-24T02:25:32Z
       
       0 likes, 0 repeats
       
       2. what if we used unused machine code bytes for assembly syntax? is it possible to build an instruction set that also makes room for assembly niceties, removing the need for a separate "textual" assembly? so that you could share executable binaries that also allow for editing without loss of "programmer" context (and thus, removing the size overhead of a text encoding)?
       
 (DIR) Post #AQHqCcoVaKEBnzi7SC by jakintosh@merveilles.town
       2022-11-24T02:33:30Z
       
       0 likes, 0 repeats
       
       3a. harder to describe, but going off the last one: what if we encode "hash-references for the routines being called" into the assembly instead of jump addresses, which are more of a marker for the CPU to interpret when loading the routine into program memory so that it can resolve the hash-id to a "real address" at runtime?
       
 (DIR) Post #AQHqCedansgVSicQiW by jakintosh@merveilles.town
       2022-11-24T02:35:11Z
       
       0 likes, 0 repeats
       
       3b. would this even work? if it did, could it enable the ability for people to easily share machine code without the need for package management? could we nearly completely eliminate routine duplication even in a multi author/networked environment?