2022-07-19 11:22:33 <KipIngram   > I only discovered mecrisp recently (someone here brought it up), but I'm quite impressed with how it works.  That's a nice architecture.
2022-07-19 11:22:51 <KipIngram   > Well, specifically mecrisp-across.
2022-07-19 11:23:30 <KipIngram   > The undeniable truth is that we Forth folk are a fairly small minority of the population, unfortunately.
2022-07-19 11:31:25 <MrMobius    > the optimizing of mecrisp is what blows me away. most forths have no real optimization to speak of
2022-07-19 11:53:17 <forthmacs   > I've found with Mecrisp Stellaris, that it is fun to try to figure out ways to reduce code size:
2022-07-19 11:53:19 <forthmacs   > https://portal.mozz.us/gemini/gem.librehacker.com/gemlog/tech/20220702-0.gmi?inline=1
2022-07-19 11:55:03 <forthmacs   > Stellaris has a disassembler and also a word that prints all the dictionary and code addresses, so it is easy to experiment
2022-07-19 11:56:09 <forthmacs   > Stellaris however is hard coded, by default, to inline all code less than 10 instructions, but that could be tweaked
2022-07-19 11:57:20 <forthmacs   > or, I assume that could be tweaked - haven't tried it myself
2022-07-19 11:58:15 <KipIngram   > I've been straining at the code size reduction thing on this boot sector thing I'm writing right now.
2022-07-19 11:58:38 <KipIngram   > One method that paid off well was to look for primitives with the same ending sequence - keep it on one and jump to it form the others.
2022-07-19 11:58:40 <forthmacs   > KipIngram: making a boot sector program?
2022-07-19 11:58:48 <KipIngram   > And if you make sure those are close together, you get a two-byte jump.
2022-07-19 11:59:00 <KipIngram   > Yeah, I'm kind of trying to merge sectorForth and PlanckForth.
2022-07-19 11:59:21 <KipIngram   > It's a PlanckForth style thing - uses a bootstrap file, but I'm trying to squeeze it into a boot sector.
2022-07-19 11:59:39 <forthmacs   > for x86, I presume?
2022-07-19 11:59:47 <KipIngram   > The init code loads a 4kB tib with the first part of the bootstrap, and then off it goes.
2022-07-19 12:00:00 <KipIngram   > Yeah - x86 16-bit real mode with a standard BIOS.
2022-07-19 12:00:51 <KipIngram   > I started out by kind of directly implementing the PlanckForth primitives.  Dropped a few that are inapplicable to the boot envrironment.
2022-07-19 12:01:09 <forthmacs   > KipIngram: which text mode do you initialize to?
2022-07-19 12:01:29 <forthmacs   > 40x25?
2022-07-19 12:01:35 <KipIngram   > I don't, in the boot sector.  I provide a general interface to the BIOS, and bootstrap.fs will take care of that.
2022-07-19 12:02:16 <KipIngram   > The only bios call it actually makes in the startup is an int 13h call to read the first eight sectors of bootstrap.fs.
2022-07-19 12:03:03 <KipIngram   > There's a primitive, 'b' with stack effect ( ax bx cx dx -- dx cx ax )
2022-07-19 12:03:31 <KipIngram   > And there's only one of those - to use it I will have to plug the interrupt number as desired before making the call.
2022-07-19 12:03:43 <KipIngram   > In the bootstrap I'll establish new more "proper" bios access.
2022-07-19 12:04:37 <KipIngram   > That is, in the 'int ??' instructin - I will have to write a byte to the ?? for what I want.
2022-07-19 12:05:07 <KipIngram   > There just wasn't room to have separate words for video, disk, keyboard, etc.
2022-07-19 12:05:32 <KipIngram   > That first int 13h is in the initialization - that one is hard coded.
2022-07-19 12:06:30 <KipIngram   > Haven't tested it yet - I wrote it over the weekend (mostly Sunday), and yesterday I started at it, fixing things, recovering more of my familiarity with what can and can't be done in that instruction set, and shaving out bytes.
2022-07-19 12:06:42 <KipIngram   > At the moment I have about 30 unused bytes; thinking about what to do with them.
2022-07-19 13:26:34 <thrig       > can it make coffee?
2022-07-19 13:38:16 <KipIngram   > Not yet - maybe someday.