2022-06-05 00:16:46 <tabemann    > klys: nibble code is a bad idea unless if you're very short on space
2022-06-05 00:16:57 <tabemann    > because it's more expensive to decode than bytecode
2022-06-05 00:31:05 <klys        > tabemann, https://paste.debian.net/1243080/
2022-06-05 00:31:13 <klys        > I been working on it some
2022-06-05 00:43:21 <klys        > needs width fields tho
2022-06-05 01:23:08 <klys        > https://paste.debian.net/1243082/
2022-06-05 01:25:43 <klys        > (errata: I merged line 5 into line 4 there, whoops.)
2022-06-05 11:13:07 <klys        > https://paste.debian.net/1243112/
2022-06-05 11:13:15 <klys        > fixed it up a tidge
2022-06-05 12:03:07 <KipIngram   > Couldn't you implement it so that you had a "next unit" routine that handed you the next nybble, byte, word, cell, or whatever?  Then you
2022-06-05 12:03:22 <KipIngram   > coud support any design you wanted, and that "next unit" thing would be more or less expensive?
2022-06-05 12:03:45 <KipIngram   > Seems like if that were a macro it would be a zero-cost flexibility.
2022-06-05 12:04:17 <KipIngram   > Then most of your system wouldn't even need to know how the code was stored.
2022-06-05 12:05:07 <KipIngram   > In fact, you could absorb the lookup in there too, so that what popped out was *always* the address of the next code to execute.
2022-06-05 12:05:41 <KipIngram   > So every bit of code would end with "next_unit; jump."
2022-06-05 12:13:11 <KipIngram   > Maybe I'm overlooking something (only one cup of coffee...), but it seems like that would cover more or less everything.  For a code threaded system, the whole thing would be "nothing" - just keep executing forward.  It would get direct threading, indirect threading, byte code, nybble code, etc.
2022-06-05 12:16:21 <Zarutian_HTC> KipIngram: this on archs where there is some register or accunulator based code implementing a Forth vm?
2022-06-05 12:28:11 <KipIngram   > I don't know - I came in mid-stream.
2022-06-05 12:28:23 <KipIngram   > Shot my mouth off anyway.  :-)
2022-06-05 12:29:11 <Zarutian_HTC> oh, same here ;-þ
2022-06-05 13:02:55 <KipIngram   > At some point, not in the *too* distant future, I hope I'll get around to beginning a re-write of this system, using the system itself.
2022-06-05 13:03:30 <KipIngram   > And I'm going to pay as much attention as possible to trying to roll in that kind of flexibility.  I'd like to wind up with something that will let me experiment with as many different types of system as possible.