2024-08-14 10:58:02 <hello-operator> unjust: Sorry we had some network issues
2024-08-14 15:24:31 <siak        > Is Forth is in use anywhere?
2024-08-14 15:25:25 <siak        > I remember seenig it in an old book as a modern language compared to C and Fortranc in some drwings.
2024-08-14 15:25:50 <siak        > Never encountered its name after that until now. Brought back memories.
2024-08-14 15:28:42 <GeDaMo      > https://en.wikipedia.org/wiki/Open_Firmware used it
2024-08-14 15:29:13 <GeDaMo      > Also some NASA craft
2024-08-14 15:29:22 <GeDaMo      > I think most uses nowadays are in embedded systems
2024-08-14 15:30:45 <dlowe       > one of my favorite old games for the Mac was written in forth and I never knew until recently
2024-08-14 15:31:07 <GeDaMo      > https://www.forth.com/resources/space-applications/
2024-08-14 15:31:15 <dlowe       > I read that something like 92% of the disk image was pointers from indirect threaded words
2024-08-14 15:32:12 <unjust      > FreeBSD loader
2024-08-14 15:33:02 <unjust      > at least as a scripting option, https://man.freebsd.org/cgi/man.cgi?query=loader&sektion=8
2024-08-14 15:33:51 <unjust      > https://man.freebsd.org/cgi/man.cgi?query=loader_4th&sektion=8 is probably better at a glance
2024-08-14 15:41:38 <unjust      > also some PLC applications like (use a translator) https://es.ua/forthlogic
2024-08-14 15:51:32 <veltas      > Some game used it I heard
2024-08-14 15:51:52 <veltas      > crc uses it professionally
2024-08-14 15:51:54 <veltas      > In here it's mostly just a hobby
2024-08-14 15:53:27 <dlowe       > I would need a very compelling reason to invest in it professionally
2024-08-14 16:00:24 <veltas      > Vender lock-in
2024-08-14 16:00:27 <veltas      > Vendor*
2024-08-14 16:00:58 <dlowe       > that's compelling :)
2024-08-14 16:01:33 <veltas      > I can't think of any reason, other than personal interest
2024-08-14 16:01:43 <veltas      > Creativity etc
2024-08-14 16:01:45 <dlowe       > modern tech can afford to spend a lot of cpu and ram on guardrails
2024-08-14 16:01:46 <veltas      > Not business reasons
2024-08-14 16:02:13 <dlowe       > even the embedded space is getting luxurious amounts of ram
2024-08-14 16:02:23 <dlowe       > (comparatively speaking)
2024-08-14 16:02:41 <veltas      > In embedded I'd rather use C, can use a lot less ROM/RAM with C than Forth
2024-08-14 16:02:43 <unjust      > now watch the code quality dwindle even further in the coming years
2024-08-14 16:02:55 <veltas      > Unless I'm using it as a powerful assembler
2024-08-14 16:03:01 <veltas      > But then development time is much much higher
2024-08-14 16:03:57 <dlowe       > I haven't done my own comparison but I thought forth compacted much better than object code
2024-08-14 16:04:10 <veltas      > Depends on arch
2024-08-14 16:04:16 <dlowe       > I will say that if I wanted some very low level code that also needed to be portable, I might reach for forth
2024-08-14 16:04:35 <veltas      > There's no modern embedded archs that are much denser than Forth
2024-08-14 16:04:41 <dlowe       > just for the number of underlying words I would have to redefine
2024-08-14 16:04:42 <veltas      > much larger I mean
2024-08-14 16:05:01 <rendar      > historically forth has never supported recursion, right?
2024-08-14 16:05:57 <GeDaMo      > Yeah it has
2024-08-14 16:06:03 <dlowe       > there's a recurse word for it
2024-08-14 16:06:27 <GeDaMo      > Older Forths had different ways of making words visible while they were being compiled
2024-08-14 16:06:42 <veltas      > More hacky but possible
2024-08-14 16:06:54 <GeDaMo      > The term "smudge bit" comes to mind :P
2024-08-14 16:08:09 <rendar      > oh the recurse word, right
2024-08-14 16:09:00 <veltas      > The oldest ref to Forth I have seems to use a model that supports recursion
2024-08-14 16:09:20 <dlowe       > not sure if there's any benefit at all over just having a loop in your word?
2024-08-14 16:09:21 <veltas      > Being the "problem oriented language" paper or whatever it's called
2024-08-14 16:09:49 <veltas      > Since Forth is very stack-based, no there's no benefit in Forth
2024-08-14 16:10:23 <veltas      > Unless you want actions on other end?
2024-08-14 16:10:23 <veltas      > So there is benefit I guess lol
2024-08-14 16:42:35 <veltas      > Yup
2024-08-14 18:09:35 <crc         > siak: re: "Is Forth is in use anywhere?", a couple of examples from my current/past projects are listed at http://forth.works/share/HoM0qVZ00t.txt
2024-08-14 18:13:58 <dlowe       > crc: thanks for sharing this!
2024-08-14 18:14:20 <dlowe       > crc: how much byte-munging did you have to do yourself for stuff like http and image formats?
2024-08-14 18:14:25 <xentrac     > you can even do mutual recursion in Forth; the classic stupid example is
2024-08-14 18:14:41 <xentrac     > defer odd  : even dup 0= if drop 1 else 1- odd then ;  :noname dup 0= if drop 0 else 1- even then ; is odd
2024-08-14 18:15:25 <dlowe       > it just occurred to me that you could do trivial tail-call optimization by jumping to the last word in a definition instead of calling it
2024-08-14 18:15:39 <xentrac     > yes, that's what ColorForth does
2024-08-14 18:15:40 <dlowe       > assuming there was one
2024-08-14 18:16:03 <xentrac     > I'm not sure if that's ANS-compliant what with the different return stack effect
2024-08-14 18:17:23 <GeDaMo      > Colorforth has a word -: for turning the call to the previous word into a jump
2024-08-14 18:17:33 <GeDaMo      > Er, -;
2024-08-14 18:18:16 <xentrac     > note though that my odd/even example above only does tail recursion and wouldn't be helped by turning the lexically last word into a jump
2024-08-14 18:25:26 <xentrac     > if you rewrite it like this:
2024-08-14 18:25:51 <xentrac     > defer odd  : even dup 0= if drop 1 else 1- r> drop odd then ;  :noname dup 0= if drop 0 else 1- r> drop even then ;
2024-08-14 18:26:06 <xentrac     > it becomes tail-recursive and still mostly works
2024-08-14 18:26:20 <xentrac     > but you can no longer type 53 even .
2024-08-14 18:26:25 <xentrac     > you have to type
2024-08-14 18:26:26 <xentrac     > 53 even
2024-08-14 18:26:27 <xentrac     > .
2024-08-14 18:26:40 <xentrac     > in the outer interpreter
2024-08-14 18:27:18 <crc         > dlowe: images were a lot of work. I'll be redoing a lot of this in the not-too-distant future using my current systems.
2024-08-14 18:27:35 <xentrac     > crc: which image formats did you write support for?
2024-08-14 18:28:07 <crc         > http isn't too bad. I stopped with basic http/1.1 support though, so there's still a lot that I didn't actually do on that
2024-08-14 18:28:37 <xentrac     > for http://canonical.org/~kragen/sw/dev3/server.s I just did HTTP/1.0
2024-08-14 18:30:45 <crc         > I needed 1.1 for vhost support
2024-08-14 18:31:41 <crc         > a slightly older copy of my server: https://retroforth.org/examples/Casket-HTTP.retro.html
2024-08-14 18:32:15 <crc         > for images, I mostly work with BMP, XBM, and XPM formats, though I've done a little with PNG in the past
2024-08-14 18:32:33 <xentrac     > that makes sense
2024-08-14 18:32:41 <xentrac     > BMP, XBM, and XPM aren't bad at all
2024-08-14 18:33:08 <xentrac     > I wrote a BMP output function in Python on a ferry across the river once
2024-08-14 18:33:28 <xentrac     > but that's less of a brag than it sounds like because it was the Rio de la Plata
2024-08-14 18:33:47 <xentrac     > so I was on the ferry for 4 hours
2024-08-14 18:39:20 <darkling    > I wrote a PPM output for my current project.
2024-08-14 18:39:45 <darkling    > (Because it's arguably the simplest possible output format).
2024-08-14 18:40:04 <xentrac     > yeah, PPM P6 is also pretty great that way
2024-08-14 18:40:20 <xentrac     > also SVG for output
2024-08-14 18:44:47 <unjust      > xentrac: interesting approach to writing macro-laden assembly, and nice be macro - did you ever consider getting it to test if src is within {b,w,l} bounds for mov suffix selection post-clear-by-self-xor? to shave off a few of those immediate zero-padding bytes?
2024-08-14 18:48:04 <xentrac     > thanks!
2024-08-14 20:28:40 <xentrac     > there are clearly constant values for which that would be a win
2024-08-14 20:28:49 <xentrac     > I don't think I ever thought of that
2024-08-14 20:57:08 <xentrac     > we should have a Forth IRC bot that evaluates Forth
2024-08-14 21:00:09 <lispmacs[work]> xentrac: somebody wrote one once. But it self-destructed, erasing its own code. The last string in the log was "A real forth programmer would code their own bot, not use somebody else's."
2024-08-14 21:32:06 <xentrac     > lispmacs[work]: haha
2024-08-14 23:31:32 <crc         > xentrac: we've tried this a couple of times, many years back. It never really got used.
2024-08-14 23:48:53 <crc         > 
2024-08-14 23:50:20 <KipIngram   > lispmacs[work]: lol... Nice.
2024-08-14 23:52:08 <xentrac     > crc: hmm, what other kinds of problems did people run into?
2024-08-14 23:56:18 <crc         > not really any problems (when a little work was done to prevent full host system access & endless looping things), but in the ~20 or so years I've followed this channel, none of them ever really got used for anything useful
2024-08-14 23:58:13 <crc         > I'm open to having one in the channel again, but it probably won't be something I develop/run
2024-08-14 23:59:22 <crc         > I have thought about writing a documentation bot, that could return information about words from various standards/implementations
2024-08-14 23:59:59 <xentrac     > IRC bots often have the problem of being inherently kind of silly