2024-08-10 00:49:47 <rendar      > https://ibb.co/YRG470Y -- this is a preview of my Forth language
2024-08-10 00:49:50 <rendar      > do you like it?
2024-08-10 01:50:37 <lf94        > 'This page doesnt exist'
2024-08-10 01:51:51 <rendar      > sorry: https://ibb.co/WFXzYwR here we go
2024-08-10 02:33:38 <neauoire    > looks funny with # for keywords but I like it
2024-08-10 03:00:11 <lf94        > rendar: reason for not just using : and ; for def and enddef
2024-08-10 10:38:16 <veltas      > I've started ##roff on this network for discussing roff and related programs, if anyone is interested
2024-08-10 11:14:53 <rendar      > lf94, it must be used by people more accustomed with a def/end structure, but it suports also #: and #;
2024-08-10 11:15:17 <rendar      > neauoire, those # are special keywords that cannot be redefined
2024-08-10 18:21:34 <xentrac     > I wrote the minimal roguelike in Forth last night: http://canonical.org/~kragen/sw/dev3/wander.fs
2024-08-10 18:22:22 <xentrac     > I'm interested in suggestions for improvement, though I might only implement them in whatever I implement next instead of in this program
2024-08-10 18:23:08 <xentrac     > there isn't any way in Forth 2012 to find the current time, is there?  For random seeding
2024-08-10 18:24:15 <GeDaMo      > There's https://forth-standard.org/standard/facility/TIMEandDATE
2024-08-10 18:35:09 <xentrac     > oh, thanks!
2024-08-10 18:37:38 <xentrac     > that lets me eliminate one of three Gforthisms from the program.  I somehow didn't realize time&date was in the Forth standard
2024-08-10 18:37:55 <xentrac     > the other two being `require` and `#!`
2024-08-10 18:38:37 <GeDaMo      > I normally check https://forth-standard.org/standard/alpha
2024-08-10 18:40:35 <GeDaMo      > #! is just to allow it to be run directly from the shell, I'd assume
2024-08-10 18:41:01 <unjust      > xentrac: you could eliminate the shebang/#! by invoking it as gforth wander.fs instead
2024-08-10 18:41:18 <xentrac     > yes
2024-08-10 18:41:38 <xentrac     > or from things that aren't the shell
2024-08-10 18:41:44 <xentrac     > unjust: yes, I did do that previously before I learned Gforth supported #!
2024-08-10 18:42:19 <GeDaMo      > : #! POSTPONE \ ; immediate
2024-08-10 18:42:29 <GeDaMo      > That's gforth's definition
2024-08-10 18:43:19 <xentrac     > yeah, that makes sense
2024-08-10 18:43:37 <xentrac     > but unfortunately there isn't a way to use that information to make #! portable to other Forths
2024-08-10 18:43:53 <xentrac     > because to work as a Unix magic number it has to be the first two bytes of the file
2024-08-10 18:44:05 <GeDaMo      > postpone, \ and immediate are all in the standard
2024-08-10 18:44:16 <xentrac     > so you can't put that definition before it
2024-08-10 18:44:22 <GeDaMo      > All it's doing is treating the rest of the line as a comment
2024-08-10 18:44:25 <xentrac     > I know
2024-08-10 18:44:29 <GeDaMo      > Ah, I see what you mean
2024-08-10 18:45:05 <xentrac     > to eliminate `require` I guess I'll add my own definition of `random`
2024-08-10 18:45:09 <GeDaMo      > Does the shebang have to be the first line in the file?
2024-08-10 18:45:29 <xentrac     > not just the first line, the first two bytes
2024-08-10 18:45:34 <xentrac     > you can't even put a space before it
2024-08-10 18:47:18 <xentrac     > ugh, https://en.wikipedia.org/wiki/Linear_congruential_generator#Parameters_in_common_use used to have information in this table about which LCG parameters were good ones
2024-08-10 18:48:32 <xentrac     > now instead it refers you to primary sources
2024-08-10 19:11:44 <xentrac     > MINSTD seems to be adequate for my purposes:
2024-08-10 19:11:45 <xentrac     > 1 value seed  \ note that this LCG PRNG assumes at least a 32-bit Forth:
2024-08-10 19:11:45 <xentrac     > : minstd ( -- rand ) seed 48271 m* 2147483647 sm/rem drop dup to seed ;
2024-08-10 19:18:35 <xentrac     > so now http://canonical.org/~kragen/sw/dev3/wander.fs works properly in both Gforth and PFE
2024-08-10 19:19:04 <xentrac     > thanks, GeDaMo!
2024-08-10 19:19:14 <GeDaMo      > :)
2024-08-10 19:20:14 <xentrac     > I'd first implemented MINSTD just using `mod` rather than `sm/rem` but I realized that that meant I was assuming a 64-bit Forth
2024-08-10 19:20:56 <xentrac     > since `seed 48271 *` is going to produce the wrong answer on a 32-bit Forth in the cases where `2147483647 mod` matters
2024-08-10 19:23:25 <xentrac     > I suspect the program might be better if I switched to representing (x, y) positions as a single offset
2024-08-10 19:27:50 <xentrac     > I also wrote a C version in http://canonical.org/~kragen/sw/dev3/wander.c because Gforth can't build a standalone executable
2024-08-10 19:29:29 <xentrac     > Because I'm not doing anything especially Forthy the C and Forth versions are about the same amount of code, but the C is formatted much more vertically, while the Forth is formatted more horizontally, arguably excessively so
2024-08-10 19:33:50 <xentrac     > heh, I was going to say "but the Forth code is probably buffering its output" but strace says it isn't
2024-08-10 19:34:16 <xentrac     > so redrawing the dungeon on the screen involves calling write(2) 48 times
2024-08-10 20:05:39 <unjust      > xentrac: do you prefer the single line style in wander.fs nowadays vs. what you did with midpoint.fs?
2024-08-10 20:05:55 <unjust      > s/single line/horizontally-biased/
2024-08-10 20:55:09 <xentrac     > hmm, well
2024-08-10 20:58:13 <xentrac     > I think the style in midpoint.fs is *clearer*
2024-08-10 20:58:23 <xentrac     > but where Forth really excels is in interactivity
2024-08-10 21:02:45 <xentrac     > you don't normally have a single-stepping debugger like you do in C or C++ or Pascal
2024-08-10 21:03:16 <xentrac     > in BASIC you have line numbers and you can interactively do things like "RUN 70" or sometimes "GOSUB 70"
2024-08-10 21:03:33 <xentrac     > in Forth instead you can *name* your lines of code
2024-08-10 21:04:02 <xentrac     > so you can run them one at a time by naming them, which is a lot like single stepping in a debugger
2024-08-10 21:06:46 <xentrac     > and interactive testing and debugging is more important than in C because (at least so far for me) the code is harder to read, and you don't have a type system to catch your bugs
2024-08-10 21:08:14 <xentrac     > so I tend more toward writing lots of small definitions so I can test them interactively now
2024-08-10 21:09:08 <xentrac     > in https://asciinema.org/a/621404 I did a screencast of debugging a buggy implementation of the Babylonian square root algorithm
2024-08-10 21:10:50 <unjust      > that makes sense
2024-08-10 21:11:14 <unjust      > have you used dbg in gforth-itc?
2024-08-10 21:11:59 <unjust      > https://www.complang.tuwien.ac.at/forth/gforth/Docs-html/Singlestep-Debugger.html
2024-08-10 21:12:35 <unjust      > to debug wander for example: page dbg wander
2024-08-10 21:12:47 <xentrac     > I haven't!  thanks!
2024-08-10 21:13:02 <unjust      > i like that approach of stepping through words, word-by-word
2024-08-10 21:13:56 <xentrac     > Code Complete recommended stepping through all the code you write in a debugger at least once
2024-08-10 21:14:04 <xentrac     > which at the time sounded like a big hassle
2024-08-10 21:14:19 <xentrac     > to me
2024-08-10 21:14:36 <xentrac     > I still don't always do it in languages like C, but often I do
2024-08-10 21:14:45 <xentrac     > I never do it in Python and JS
2024-08-10 21:15:12 <unjust      > if you're comfortable with your debugger of choice (or circumstance) it's worth it, at least with languages that end up being translated in byte or machine code
2024-08-10 21:15:27 <unjust      > i'm biased though, and sort of like debugging more than programming though
2024-08-10 21:16:04 <xentrac     > I feel like the Forth style of very small definitions makes it less necessary
2024-08-10 21:18:08 <unjust      > i guess the trick is being able to think through your problem well enough to ultimately factor your solution into concise definitions that are ideally self-evident enough that it becomes obvious how everything should be threaded together - but i haven't mastered that ideal