2021-12-21 04:47:52 <neuro_sys_  > veltas: Thanks, I read it in the crc's occasionally crashing logs.
2021-12-21 04:48:08 <veltas      > No problem
2021-12-21 04:48:18 <veltas      > There is a way of posting messages too but I can't be bothered lol
2021-12-21 04:48:37 <neuro_sys_  > potatoalienof13: About your question of using linked list. It's only relevant for the compile time speed.
2021-12-21 04:49:27 <neuro_sys_  > Just in case you might have overlooked it as I did when I first read about Forth, and was apalled of using a linked list to look up each word.
2021-12-21 04:50:12 <neuro_sys_  > Yeah there's memoserv but I think I have not registered my nickname on this server yet.
2021-12-21 04:53:02 <neuro_sys_  > After solving more Advent of Code with "modern" languages very fast and comfortably, now I feel like experimenting with making Forth similar to them in syntax and paradigm.
2021-12-21 04:53:29 <neuro_sys_  > I wonder how much of a similarity could be achieved in parsing a C-like language syntax.
2021-12-21 04:53:48 <neuro_sys_  > Without writing an AST and in-fix parser
2021-12-21 04:57:15 <veltas      > Locals get you pretty close
2021-12-21 04:57:43 <veltas      > You don't miss in-fix at all, you can write essentially postfix C code with just locals
2021-12-21 04:58:25 <veltas      > But you feel gross and all your methods are larger, sometimes it's valid though TBH, if you allow yourself
2021-12-21 04:58:32 <veltas      > It's definitely not very 'pure'
2021-12-21 04:59:04 <veltas      > And alternative that people probably used to use was putting stuff on return stack and accessing it via I, I', J (equiv of R@, R2@, R3@)
2021-12-21 04:59:36 <veltas      > No harm in playing around with it, but you'll come back. They all do!
2021-12-21 05:09:02 <neuro_sys_  > That was my first attempt in learning Forth. I basically wrote C without types. And indeed it did not feel right.
2021-12-21 05:10:13 <neuro_sys_  > I like the retro's approach though, but still I haven't yet enough enjoyed the traditional Forth to dissent much.
2021-12-21 05:12:55 <neuro_sys_  > There are a couple talks on Youtube about concatenative languages which are inspiring. I still want to be as fluent or more in Forth than I am with other languages. It takes practice for sure, but let's see.
2021-12-21 05:17:02 <neuro_sys_  > https://www.youtube.com/watch?v=TcveznrwYtk
2021-12-21 05:17:05 <neuro_sys_  > Hm, I had not seen this
2021-12-21 05:21:05 <veltas      > SectorLISP is now smaller than SectorFORTH https://justine.lol/sectorlisp2/
2021-12-21 05:21:48 <neuro_sys_  > Lol indeed
2021-12-21 05:21:53 <neuro_sys_  > I wonder if it does a GC
2021-12-21 05:22:00 <veltas      > Yeah it has GC
2021-12-21 05:25:14 <veltas      > I think it might just be reference counting, not sure, but depending on how dumb the Lisp is that would be sufficient
2021-12-21 05:25:26 <veltas      > Well sufficient 90% of the time in a lisp that fits in one sector, anywya
2021-12-21 05:26:03 <veltas      > Looks like it's more usable than sectorFORTH too, not sure I'd want to write lisp anyway but most people probably would prefer that over FORTH
2021-12-21 06:55:53 <dave0       > i find it insulting!
2021-12-21 06:56:00 <dave0       > forth is clearly smaller than lisp :-)
2021-12-21 06:56:42 <f-a         > hopefully
2021-12-21 07:20:08 <veltas      > I wouldn't guess lisp was able to be smaller but right now they're ahead
2021-12-21 07:20:32 <veltas      > I guess we'll see if sectorforth catches up, or another contender enters the ring
2021-12-21 07:34:04 <crc         > I suspect sectorforth could be made smaller, but is that necessary?
2021-12-21 07:35:05 <crc         > given that an x86 boot sector is required to be 510+2 bytes, I'd only focus on shrinking it further if there was something else I could use the extra space for.
2021-12-21 07:52:38 <veltas      > Got plans to do something like this in FORTH at some point https://zig.news/xq/a-challenger-to-the-throne-of-vector-graphics-svg-is-dead-long-live-tinyvg-4on8
2021-12-21 07:53:03 <veltas      > crc: It's about being the smallest though, Justine's called the FORTH community out
2021-12-21 07:53:07 <veltas      > It means war!!!
2021-12-21 07:53:37 <veltas      > Just a bit of friendly competition really, I wonder what sectorforth can cdo
2021-12-21 07:53:56 <veltas      > I can't even wrap my head around how a FORTH or any interpreter can be that small! I'll have to read in more depth at some point
2021-12-21 07:57:00 <crc         > I've done a couple of forth systems that fit in a boot sector back when I was into x86 coding
2021-12-21 07:57:31 <crc         > you can't do a lot, but it's possible to squeeze an interpreter and basic compiler to fit
2021-12-21 08:09:06 <crc         > stripping out some of the nicer backspace handling in output drops sectorforth to 470 bytes; not hiding definitions until ; saves another 10 bytes
2021-12-21 08:12:19 <veltas      > Maybe you should do a follow-up to that article then
2021-12-21 08:12:57 <veltas      > A bit of good humoured competition
2021-12-21 08:13:39 <crc         > I might...
2021-12-21 08:14:01 <veltas      > The old vi text editor doesn't really handle backspace properly, it's pretty much baked into the POSIX standard for it
2021-12-21 08:14:32 <veltas      > Yeah go for it, will probably get some attention on HN
2021-12-21 08:18:31 <crc         > this could be made much smaller. TIB, >IN, STATE are at fixed addresses, so can be defined at the interpreter. Removing these gets it down to 424 bytes, which is smaller than sectorlisp.
2021-12-21 08:38:04 <neuro_sys_  > Wasn't the sectorforth guy lurking here
2021-12-21 08:39:54 <veltas      > How can you define those at the interpreter if you don't have them?
2021-12-21 08:40:14 <veltas      > Surely those are needed for the interpreter to function
2021-12-21 08:40:51 <crc         > they are at fixed positions in memory, so : tib 0 ;  : state 4096 ; etc
2021-12-21 08:40:56 <neuro_sys_  > It's also important to stress that the rest of the Forth (like ANS Forth) can be built from that small core. When I was watching that sectorlisp video, the guy mentioned that as something unique to sectorlisp that you can compile the lisp 1.5 or something. I suppose you can do that same with sectorforth.
2021-12-21 08:42:08 <crc         > veltas: they need to exist, but the names for them don't need to be in the initial dictionary
2021-12-21 08:50:48 <potatoalienof13> how much bootstrapping is required for sectorforth compared to sectorlisp?
2021-12-21 08:58:34 <veltas      > crc: Oh right, of course
2021-12-21 08:58:58 <veltas      > potatoalienof13: I think a lot compared to sectorlisp which supposedly passes some lisp tests out the box
2021-12-21 08:59:22 <veltas      > But you'd have to see for yourself really
2021-12-21 09:03:11 <veltas      > Maybe space can be saved with a TOS register?
2021-12-21 09:03:43 <veltas      > Will change the bootstrapping code but would probably save some space
2021-12-21 09:04:44 <veltas      > Hmmm maybe that's not an option though, can't access TOS if it's in a register without another word, probably wastes more than it saves
2021-12-21 14:06:46 <xyh         > i share a tool i write for sharing docs: https://readonly.link
2021-12-21 14:24:01 <crc         > xyh: does it only work with github?
2021-12-21 15:22:49 <veltas      > crc: I don't think so
2021-12-21 15:23:20 <veltas      > github.com is in the URL so I think you can put other hosts in there?
2021-12-21 15:24:15 <veltas      > xyh: Hey this is really nice https://readonly.link/articles/github.com/veltas/zenv/-/README.md
2021-12-21 15:24:24 <veltas      > That was a really basic README and it looks quite good in there
2021-12-21 15:24:55 <veltas      > xyh: What did you make that with? I'm guessing this has nothing to do with FORTH?
2021-12-21 15:26:36 <veltas      > Does it do pretty much all the work clientside?
2021-12-21 15:30:54 <veltas      > xyh: You need to make the navigation of books more obvious for dumb people like me
2021-12-21 15:31:08 <veltas      > Took me a while clicking around to figure out I needed to click on the title
2021-12-21 15:32:37 <xyh         > veltas: yes, i will support other hosts, such as gitlab & npm, and custom gitlab host.
2021-12-21 15:33:29 <xyh         > veltas: yes, i build it in typescript and vue, it is not about forth, i just want to share it.
2021-12-21 15:34:11 <xyh         > yes, it fetch files from github repo, and render it at runtime.
2021-12-21 15:35:06 <veltas      > I like it, but I wouldn't go sharing it in random channels like this
2021-12-21 15:35:25 <xyh         > i will improve it, if you want some new features, let me konw :)
2021-12-21 15:35:45 <xyh         > sorry, i was rude, i will not do it again.
2021-12-21 15:37:02 <veltas      > Looks good to me, maybe get someone to post it on hackernews. It's a bit slow on my laptop but if it's clientside that's why
2021-12-21 15:37:24 <veltas      > My laptop is quite old so probably not an issue
2021-12-21 15:38:16 <xyh         > i can improve the slow loading by not using github api but using CDN by default.
2021-12-21 15:39:44 <crc         > I use git.sr.ht for git hosting; so will look at this more in the future when it supports more hosts.
2021-12-21 15:41:33 <xyh         > hi crc!
2021-12-21 15:41:37 <xyh         > long time no see
2021-12-21 15:42:22 <xyh         > i remember your retro
2021-12-21 15:50:54 <xyh         > i tried to render this: https://readonly.link/articles/github.com/crcx/retroforth/-/doc/RETRO-Book.md
2021-12-21 15:51:14 <xyh         > the rendering is really slow for big page
2021-12-21 15:54:30 <xyh         > a small page is cool :) https://readonly.link/articles/github.com/crcx/retroforth/-/doc/QuotesAndCombinators.md
2021-12-21 15:56:46 <crc         > Looks good
2021-12-21 16:16:25 <veltas      > xyh: Have you tried profiling it? To see where you can make it faster?
2021-12-21 16:17:53 <veltas      > "xyh has become unresponsive, do you want to wait?"
2021-12-21 16:19:13 <veltas      > xyh: Have you tried profiling it? To see where you can make it faster?
2021-12-21 16:19:50 <swissChili  > Hi I'm new-ish to forth, what to you all use forth for nowadays? It's certainly an interesting language, just wondering if it's used practically
2021-12-21 16:20:22 <veltas      > crc: ^
2021-12-21 16:21:00 <veltas      > swissChili: I use it as a glorified desk calculator at work, although I'm hoping to get more tools working in forth
2021-12-21 16:21:40 <veltas      > It's highly portable, FORTH is on some systems where literally the only alternatives are assembly and C, but FORTHs tend to come with a REPL built-in which is quite cool
2021-12-21 16:21:48 <veltas      > So it has embedded uses
2021-12-21 16:22:39 <crc         > swissChili: I do most of my programming in Forth.
2021-12-21 16:23:31 <crc         > for work, I develop/maintain an in-house, browser based order management system and some related tools
2021-12-21 16:23:40 <swissChili  > A big part of it is probably just being unfamiliar with the language, but I've had a lot of trouble dealing with any sort of complicated data (even just something tabular) in forth. That's been the main thing keeping me from making anything useful with it ;)
2021-12-21 16:24:03 <veltas      > It can't be overstated how important being tiny is. FORTHs are so small that you or any half-decent programmer can learn everything about them, if they want, in not that much time.
2021-12-21 16:24:17 <swissChili  > crc: What forth dialect to you use? If that is in ANS forth consider me very impressed
2021-12-21 16:24:19 <veltas      > It makes business sense to me, I know my team can be their own support if they need to be
2021-12-21 16:24:31 <crc         > swissChili: retroforth
2021-12-21 16:25:04 <veltas      > Do you think you could do your work in ANS forth instead crc?
2021-12-21 16:25:07 <veltas      > I'd like to think so
2021-12-21 16:25:32 <swissChili  > Retro looks cool, I skimmed the docs for it a few months ago. It looks like it addresses a lot of the difficulties I was having with managing more complicated data in ANS forth
2021-12-21 16:26:00 <crc         > I probably *could*, but I think what I end up with would be similar to retroforth after I define the words I want
2021-12-21 16:26:14 <veltas      > That's what I would expect yeah
2021-12-21 16:28:32 <swissChili  > Do you use CGI or have you implemented an HTTP server in Forth? In case of the latter, how do you go about concurrency? Just curious
2021-12-21 16:30:42 <crc         > for work, CGI.
2021-12-21 16:30:58 <crc         > for static sites/pages, I use http://retroforth.org/examples/Casket-HTTP.retro.html
2021-12-21 16:31:37 <crc         > (I have a version of casket with limited cgi support, but this isn't released yet)
2021-12-21 16:33:42 <swissChili  > Cool stuff, thanks
2021-12-21 16:34:40 <crc         > casket runs under inetd, so is basically uses a separate process for each connection. not massively scalable, but sufficient for my current needs
2021-12-21 16:39:18 <swissChili  > A quick grep of handbook.txt reveals some multitasking support, so I guess a threaded web server could be possible as well
2021-12-21 16:40:10 <crc         > maybe; I haven't worked on that though, so can't say for certain
2021-12-21 16:42:10 <crc         > the multitasking stuff is very new (it's a port of the system used in new retro implementation for my new vm), and hasn't been tested or used extensively so far