2023-06-10 04:55:53 <dave0       > veltas: this is a very good document on the z80!    http://www.z80.info/zip/z80-documented.pdf
2023-06-10 10:38:49 <KipIngram   > I've still got a Z80 book up on my bookshelf, I believe.  I'm kind of a packrat when it comes to books.
2023-06-10 10:49:10 <thrig       > I have a Z3 pdf somewhere. not sure what happend to Z4..Z79
2023-06-10 10:50:22 <KipIngram   > So just to be sure I've got this right.  In that map example we were talking about yesterday, that I wrote as <[ 1 2 3 ]> <{ 1 + }> , <{ would just compile a jump and leave the address of the distance field of that jump on the stack, and then }> would compile a return and fix up that distance field to make the jump land just past that return, right?
2023-06-10 10:51:07 <KipIngram   > And <{ would compile, just before the jump, a runtime word that would push an xt for that little section of code onto the stack.
2023-06-10 10:54:02 <KipIngram   > So we wind up with
2023-06-10 10:54:50 <KipIngram   > <xt pushing runtime> <jump> <to target> | ... ret ||
2023-06-10 10:55:05 <KipIngram   > so the xt pushed is for | and the target is set to || ?
2023-06-10 10:55:42 <KipIngram   > And that xt pushing runtime is really just (lit) with a proper value.
2023-06-10 11:01:12 <KipIngram   > I'm debating whether or not I actually want to bother with that.  It's appealing from a "just because I can" perspective, but in practice I'd be very likely to factor that right out into its own word to shorten my definitions, so I'd wind up not using it.
2023-06-10 11:01:39 <KipIngram   > Just as I use my conditional returns to factor the code that would otherwise be in IF ... THEN sections.
2023-06-10 11:03:34 <KipIngram   > If I didn't want that name in the dictionary I'd define it with .:
2023-06-10 11:10:10 <KipIngram   > I'm thinking maybe instead of having both append and overwrite options on that "paste into block" stuff, I may keep just append, and if I want to overwrite I do that by erasing first, with <block#> block off.
2023-06-10 11:10:26 <KipIngram   > So block< will always append.
2023-06-10 11:10:34 <KipIngram   > Less chance of fouling up that way.
2023-06-10 12:06:56 <KipIngram   > Ok, I finally wound up with this:
2023-06-10 12:06:59 <KipIngram   > https://pastebin.com/e9nH2qqL
2023-06-10 12:07:38 <KipIngram   > Gives me two words, ram< and block<.  Using ctrl-z to terminate instead of escape, so I could type escape sequences directly into RAM if I want to.
2023-06-10 12:08:17 <KipIngram   > Uses a double return to finish up.
2023-06-10 12:26:07 <KipIngram   > So, the way I'm set up, I can key zero with ctrl-space, and ctrl-a through ctrl-z keys 1-26, except for ctrl-v which I have set for pasting in my terminal emulator.  Then ctrl-[ or esc gets 27, ctrl-] gets 29, and ctrl-\ gets 28.  Anyone know of things that give 30 and/or 31?
2023-06-10 13:14:17 <thrig       > control-@ is typical for '\0', following the columns in the ASCII manual
2023-06-10 13:20:27 <Riviera     > KipIngram: did not read the backlog, so i don't know which environment you're talking about; classically Ctrl might be understood as either anding or xoring, with which 32 would then not be possible, as it'd "and" for keeping the lowest 5 bits. anyway, try Ctrl-_ for 31.
2023-06-10 13:20:54 <Riviera     > (and C- is not an "escape" sequence)
2023-06-10 13:21:01 <Riviera     > (not a "sequence but a single character)
2023-06-10 14:00:27 <KipIngram   > ctrl-_ is indeed 31.  Thanks.
2023-06-10 14:00:58 <KipIngram   > It's a Fedora 36 environment, using terminator as my terminal app.
2023-06-10 14:01:05 <Riviera     > it's not so easy
2023-06-10 14:01:24 <KipIngram   > I can switch psate back to shift-ctrl-v instead of ctrl-v if I want that one back enough.
2023-06-10 14:01:29 <Riviera     > there's several layers between keyboard and application running in the terminal, and even those can be configured
2023-06-10 14:01:50 <KipIngram   > I don't have any immediate need for this, but it would just be kind of cool to know I could produce all 128 bytes from the keyboard.
2023-06-10 14:01:52 <Riviera     > so while "in theory" C-_ *is* ASCII character 31 it does not necesarily have to be that in a given setup.
2023-06-10 14:02:10 <KipIngram   > It is in this one, though.
2023-06-10 14:02:14 <Riviera     > yup
2023-06-10 14:02:21 <Riviera     > "just saying"
2023-06-10 14:02:21 <Riviera     > :)
2023-06-10 14:02:25 <KipIngram   > Did you mention 30 at all?
2023-06-10 14:02:36 <KipIngram   > Yeah - I don't fret a whole lot over portability.  :-)
2023-06-10 14:03:00 <Riviera     > ah right, 30
2023-06-10 14:03:02 <Riviera     > C-^
2023-06-10 14:03:20 <KipIngram   > Perfect - thanks again.
2023-06-10 14:03:23 <Riviera     > :)
2023-06-10 14:03:57 <KipIngram   > Now if I can come up with some streamlined way to toggle that high bit (maybe after the byte's already in RAM), I'll be able to type in arbitrary stuff.
2023-06-10 14:05:10 <thrig       > (this is easy if you `man ascii` and then follow G to bel and then you can figure out the rest of the control keys)
2023-06-10 14:05:19 <Riviera     > The 8th bit? IIRC that's what the meta key once was, but i'm not sre.
2023-06-10 14:05:20 <Riviera     > sure.
2023-06-10 14:05:39 <Riviera     > Keep in mind, please, that you're talking about terminal contexts, and in these text and character encodings matter
2023-06-10 14:06:14 <Riviera     > simplest way might just be to configure the keyboard in your linux environment to allow for entering arbitrary values.
2023-06-10 14:06:27 <Riviera     > like a modifier key and a byte value or stuff like that.
2023-06-10 14:06:51 <KipIngram   > Yeah, could be.
2023-06-10 14:06:56 <Riviera     > it's not really an actual problem you have, right, it's just something that you think is cool? perhaps ... re-evaluate your life choices. :) just saying. :p
2023-06-10 14:07:18 <KipIngram   > And I can't really use this to type ANYTHING, because I am using ^Z to terminate entry.
2023-06-10 14:07:26 <KipIngram   > So that one's unavailable.
2023-06-10 14:07:59 <Riviera     > to terminate entry?
2023-06-10 14:07:59 <KipIngram   > And I'm detecting 13 and changing it to 10, because that's what I need in order to have pastes from vim wind up right in my blocks.
2023-06-10 14:08:08 <Riviera     > what does that mean?
2023-06-10 14:08:45 <KipIngram   > I wrote a snip of Forth yesterday and today that will let me get from the keyboard or paste buffer into my source code blocks.
2023-06-10 14:09:06 <KipIngram   > That was the original purpose at least - to be able to copy source code I wrote in vim and paste it into my system.
2023-06-10 14:09:34 <KipIngram   > In a way that saves it somewhere instead of immediately executing it.
2023-06-10 14:09:48 <KipIngram   > So that I can then edit it within the system with my block editor.
2023-06-10 14:09:55 <KipIngram   > Or LOAD it or whatever.
2023-06-10 14:50:26 <KipIngram   > The whole "every value" thing was just a side excursion.
2023-06-10 16:04:08 <KipIngram   > So I was just thinking about parallelizing matrix operations.
2023-06-10 16:04:29 <KipIngram   > I think something to try would be to store the matrix by columns, and divide responsibility for the columns up among your cores.
2023-06-10 16:05:26 <KipIngram   > The each core would get its columns into its cache, and if done right they wouldn't interfere wit one another.  Use a CSP type of thing to dispatch each row operation to the worker pool; wait till they all finished their share of the work, move on to the next operation.
2023-06-10 16:05:37 <KipIngram   > Of course I know GPU is the real way to do that.
2023-06-10 17:25:54 <KipIngram   > So the way my system works is that if a run-time error like a bad memory reference, divide by zero, etc. occurs (generally speaking things that would normally segfault), my error handler catches it and returns control to the interpreter.
2023-06-10 17:26:41 <KipIngram   > But I guess I only want that for the "root" task that's launched at startup, right?  If a thread I launch later encounters such an issue, I should just kill it, I'd think?  Unless I've provided some sort of "exception handler"?
2023-06-10 17:36:44 <KipIngram   > I think that means I have a user variable that holds the xt for the error recovery "exit strategy."
2023-06-10 17:36:56 <KipIngram   > Default when a task is created will be a "suicide" word.
2023-06-10 17:38:09 <thrig       > sepuku! harakiri!
2023-06-10 17:39:54 <KipIngram   > Heh.  I thought of the harakiri name, but I'd forgotten sepuku.
2023-06-10 17:40:04 <KipIngram   > Probably ought to rerad Shogun again.  :-)
2023-06-10 17:40:11 <KipIngram   > Been... a long long time.
2023-06-10 17:40:54 <KipIngram   > sepuku onerror !
2023-06-10 17:41:15 <thrig       > having five rings would be a nice musashi reference
2023-06-10 17:41:28 <KipIngram   > You're beyond me now.
2023-06-10 17:44:07 <KipIngram   > I think if a thread goes that route, I'll not immediately deallocate it's space - I'll put a reference on some list so I can find it and inspect it if I want to, and then if it's still around later then deallocate it.
2023-06-10 17:44:35 <KipIngram   > Window of opportunity to take a look at it, but still not a "requirement" to manually do anything with it.
2023-06-10 17:45:12 <KipIngram   > Might be interesting to look at the call sequence that led up to the failure.
2023-06-10 23:00:10 <drakonis    > http://web.archive.org/web/20030710020735/http://4th.allthingsgo.biz/~sye/BOOST.pdf