00:00:00 --- log: started retro/17.07.17
09:25:12 --- join: FreeFull (~freefull@defocus/sausage-lover) joined #retro
09:32:58 --- join: kisspunch (~za3k@smtp.za3k.com) joined #retro
09:33:22 <kisspunch> Not a fan of the new white-on-black website
09:39:53 <kisspunch> Also thanks for the excellent Retro documentation--I'm using it as base to write a Forth in native code
09:40:38 <crc> kisspunch: I'll be having the colors selectable shortly. (Working on some stuff in the server to allow this)
09:41:00 <kisspunch> will this require javascript or cookies? won't work for me then
09:41:06 <kisspunch> :P
09:41:22 <crc> no :)
09:48:00 <kisspunch> terrifying. very good then
09:49:48 <kisspunch> crc: as long as you're here can you point me at an overview of parable slices?
09:50:00 <kisspunch> i'm trying to understand if this is something i can use in 64K
09:51:27 <kisspunch> the thing with redirecting the input steam in forth /also/ bugs the hell out of me
09:51:45 <crc> https://github.com/crcx/parable/blob/master/source/docs/Parable.md, under Memory Model IIRC
09:53:22 <kisspunch> also i'm unclear on the idea of 'quotations' but i'm pretty sure i should just read :)
09:53:39 <kisspunch> e.g. what type of scoping they use / macro they correspond to
09:55:20 <crc> quotations are just anonymous, nestable functions
09:56:00 <crc> There's not a lot of documentation on slices. Basically in Parable memory is allocated as a set of slices; each of which can grow / shrink independently.
09:56:13 <crc> Fetches and stores require a slice number and an offset into the slice
09:56:49 <crc> Parable has a very conservative garbage collector to identify and reclaim unused/no longer needed slices
09:56:49 <kisspunch> how do you manage both offsets and independent growth?
09:56:51 <FreeFull> Does Parable have its own virtual machine?
09:57:11 <kisspunch> also maybe i shouldn't go into a long conversation about parable in #retro :)
09:57:35 <FreeFull> kisspunch: I disagree
09:57:49 <FreeFull> I haven't heard about parable until now
09:57:56 <kisspunch> crc: sure but the question is about how quotations deal with things like variable/function names and redefinitions
09:58:30 <crc> parable is a more experimental language than retro. I tried lots of things with it, and it's influenced my recent work on Retro in various ways
09:58:32 <kisspunch> is it just blind bytecode copying, which seems the forth-iest?
09:58:45 <crc> it never took off though
09:59:04 <crc> kisspunch: there's a single global dictionary, no locals, etc. A lot like forth in that regard
09:59:15 <crc> redefinitions just copy the bytecode, yes
09:59:30 <kisspunch> does quotation definition compile then or later?
09:59:35 <crc> unlike forth, parable does not allow reuse of a name. redefinining it replaces the original definition
09:59:58 <crc> kisspunch: in parable, everything is compiled prior to running
10:00:16 <kisspunch> yeah so imagine something like: <define X> <define a quotation using name X> <refine X> <pass the quotation "X">
10:00:21 <kisspunch> this kind of edge case :P
10:01:28 <kisspunch> anyway sounds pretty straightforward
10:02:13 <crc> a recursion example:
10:02:18 <crc> [ ] 'fib' :
10:02:18 <crc> [ "n-n" dup #1 gt? [ [ #1 - fib ] sip #2 - fib + ] if-true ] 'fib' :
10:03:59 <kisspunch> oh, does this mean you can do forward references or is recursion special-cased?
10:04:48 <kisspunch> e.g. every function call is a dictionary lookup
10:05:42 <crc> you can define a function, then implement it later via redefinition
10:06:13 <kisspunch> ohhh, didn't understand the ":" syntax was the end, got it :)
10:07:05 <kisspunch> that was in the examples so i probably should have
10:07:15 <crc> parable is rigidly rpn; so functions take all arguments from the stack
10:07:38 <kisspunch> yeah that's more compelling to me than typed cells (for 64K)
10:07:50 <kisspunch> I think I probably don't have enough memory to try out parable
10:08:19 <crc> (retro is currently similar, though retro allows for new prefixes, whereas the prefixes in parable are part of the compiler)
10:08:25 <kisspunch> typed heaps and garbage collection are excellent features, just playing around with some really old hardware
10:09:08 <kisspunch> yeah i'm going to go with traditional forth total-compilation-without-dictionary-lookup
10:09:09 <crc> I don't think parable will run in 64k. I *think* the lowest memory space I was successful in was 256k with a somewhat stripped down stdlib, using micropython
10:09:27 <kisspunch> well i probably mean something like 'parable re-written in assembler'
10:09:54 <kisspunch> parable-in-spirit
10:10:02 <crc> I'd love to see that. Writing an embedded-capable version is on my long-term todo list.
10:10:24 <kisspunch> so to me, having a very bare-metal language like forth that runs in a emulation-layer VM is sort of missing the point
10:11:19 <kisspunch> i think you should target a version of Nga/Ngaro which avoids the overhead of instruction lookup ("emulation")
10:11:24 <kisspunch> using standard VM techniques :)
10:11:40 <kisspunch> i mean if yer bored
10:12:00 <crc> it's more a matter of time than boredom.
10:12:23 <crc> I switched to working on an emulated architecture to reduce the time I spent testing each change.
10:12:26 <kisspunch> anyway thanks for all the help
10:12:34 <kisspunch> yeah makes sense
10:12:47 <kisspunch> Retro is pretty polished
10:12:51 <crc> even when I was using x86 assembly directly, supporting 10-12 OSes lead to a lot of time testing/hunting down bugs
10:13:11 * crc is interested in writing something similar to retro in arm assembly at some point though
10:13:25 <crc> thanks :)
10:14:29 <kisspunch> so the advantage of binary translation is then you only have to debug nga, and you can default to a version that works, but is slower
10:14:39 <kisspunch> and nga is much simpler and more static
10:15:52 <kisspunch> i guess there's still the "get anything to run in an OS at all" which seems like more work
10:16:09 <kisspunch> but as far as instruction sets
10:16:39 <kisspunch> do you want a copy of retro-on-Z80 when I give up on it much before it's done?
10:16:49 <crc> yes
10:17:05 <kisspunch> aight, ta
10:17:05 --- part: kisspunch left #retro
12:59:53 --- quit: docl (Read error: Connection reset by peer)
13:04:15 --- join: docl (~docl@159.203.115.16) joined #retro
17:49:30 --- quit: trn (Excess Flood)
17:49:51 --- join: trn (jhj@prone.ws) joined #retro
18:13:03 --- join: kisspunch (~za3k@smtp.za3k.com) joined #retro
18:13:12 --- part: kisspunch left #retro
21:15:35 --- join: kisspunch (~za3k@smtp.za3k.com) joined #retro
21:15:45 --- part: kisspunch left #retro
23:05:51 --- quit: FreeFull ()
23:59:59 --- log: ended retro/17.07.17