00:00:00 --- log: started forth/17.04.03 00:00:04 Well, that was why I was hoping someone out here already knew freeforth 00:00:04 : five-x: create, does> @ 5 * ; seems to do just what you need if you had been using ans 00:00:15 perhaps though it is a small channel it could be so 00:00:52 generally speaking there are 8 words in the above statement, so you simply must transalte 1 word at a time until you reach the end of the definition 00:00:58 freeforth has some cool stuff, but the original author seems to have gone dark. No recent postings, no response to email, etc... 00:01:08 For example what is the equivelant of : in freeforth? 00:01:18 Its still : 00:01:36 I know freeforth pretty well, but this stupid little thing has me baffled 00:01:38 and ; is still ; correct? 00:01:39 kevinfish: was going to download it and have a go, but apparently it is not 64-bit clean 00:02:02 and i don't feel like setting up an i386 environment right now 00:02:07 koisoke: yeah, I think its only 32 bit. There is a freeforth2 on github that might be 64 00:02:22 the latter is linux only tho 00:02:32 is there a particular reason why you have chosen freeforth? 00:02:45 I like some of the performance aspects of it 00:03:23 for example, it uses the cpu's status registers for tests rather than popping a t/f value off the stack 00:03:28 here is my reasoning. If freeforth is indeed abandoned then in a sense you are the developer in charge of keeping freeforth working and stable. And if you are using your own forth you will make a much faster forth by writing it yourself from scratch in c 00:03:38 hand written forths are generally always faster than forths written by others 00:03:44 at least according to moore 00:04:40 Well, I may be supporting it then. Me and the guy who's doing freeforth2. I don't think I'm currently at the level of expertise to whipe one up from scratch that's up to par with freeforth 00:04:49 looking at freeforth2 and i see a bunch of 32-bit asm 00:05:09 yeah, I haven't looked closely at freeforth2 cuz I'm doing an app in windows 00:05:15 if you understand forth to the point where you know reverse polish notation and understand macros then you should be able to write your own forth 00:05:29 I've written several already 00:06:03 If you have another person helping you with this 2nd version of freeforth then that should improve your odds 00:06:04 Its just this one has be interested in mastering cuz its so different 00:06:41 I think what I'll have to do is instead of trying to extend existing defining words, write them again from scratch until I figure this out 00:06:53 this is perfectly doable in forth. 00:07:09 redefining to make optimizations is a good strategy 00:07:23 and if you are indeed rewriting the freeforth "standard" into your own thing, this may be necessary 00:07:38 like for example, here's a definition: : fun:` :` lit` lit` #call ' call, ;` ; 00:07:59 fun: creates a word that will call an internal function call (like in a dll) 00:08:47 the :` postpones the colon def 00:09:07 it takes 2 args: the entry address (to be compiled by the first lit`) and the number of args (the 2nd) 00:09:47 do does fun:` postpone the function definition? 00:09:48 the #call ' call, compiles a call to #call which calls the address in the defined word 00:10:10 defining fun: with the trailing backquote makes it an immediate word 00:10:20 I see 00:10:32 so the ` character is used for a different purpose in that situation 00:10:49 but if you did something like : foo fun:` ... ; that would postpone fun: 00:11:07 he he he in your version of freeforth you may want to make that two separate symbols to avoid confusion 00:11:27 its a bit of a mind bender but there is a certain elgance to it 00:11:50 want to have some real fun look at the literal compiler and control flow in freeforth 00:12:14 almost all the words in the dictionary are macros to expand as much into inline code as possible 00:12:32 I haven't had much time to play with it yet but it must really scream 00:12:36 is there a freeforth equivelant to the ans forth word create 00:12:44 yes, it has create 00:12:48 does it have , 00:12:52 yes 00:13:19 here's constant: : constant` create` H@ ! anon:` ; 00:13:47 anon:` restarts it compiling an anonymous definition (its equivalent of interpretation) 00:14:48 so basically 00:14:50 --- quit: dys (Read error: Connection reset by peer) 00:14:53 I think H points to the memory location where the header gets compiled in the symbol table 00:14:54 in terms of the first ' 00:14:58 as in : myword' 00:15:03 that is the same as using the word immediate on that word 00:15:11 yes 00:15:12 and in terms of putting ` at the end of the word 00:15:17 it is like saying ' thatword 00:15:28 as in : pointer-to-create ' create ; 00:15:30 the opposite 00:15:38 similar 00:15:38 : myword` is immediate 00:15:47 strange 00:15:52 and foo ' is similar to ['] foo 00:15:54 so it does not delay it from being evaluated then 00:16:07 ' turns the previous call into a lit 00:16:14 or rather it delays it from being evaluated once but after compilation it gets evaluated each time 00:16:17 it splices the previously compiled code 00:16:22 ^ ? 00:16:56 words that do not have ` are run as normal words during compilation but words that have ` are quoted 00:17:27 words that have a trailing backtick in their dictionary header are immediately executed 00:17:46 but if they have a trailing backtick as they are being copiled into the body of a definition are postponed 00:17:51 Let me show you some ans forth 00:18:10 s" hello" s" hello" .s will show you two different string pointers even though they are the same string 00:18:24 but if I run : mystring s" hello" ; mystring mystring = I get true 00:18:31 or rather 00:18:32 he he 00:18:41 mystring mystring .s will give you two copies 00:18:50 right 00:18:59 that is because s" is sort of a macro and has different compile time behavior than runtime behavior 00:19:34 but that is as far as I understand it because I do not understand the word immediate, or the word create, or the word , 00:19:57 freeforth is pretty bizarre, but is very forth-like in the concept that if you can tweak things a little bit and get a performance gain and shrink the compiler, or speed it up, then go for it. 00:20:56 Like Chuck says: "confusion only exists once, but cost exists forever" 00:22:17 My intuition tells me that there are 9 words in the working ans code and we have already translated 7 of the words 00:22:20 well, most languages like forth, lisp, etc... that heavily rely on macros can twist your brain trying to figure out what is happening when 00:22:31 and so logically if we can translate the other 2 words then we will ahve solved it 00:22:44 I don't think its that easy 00:22:57 first of all in our ans forth version we did not use a backtick in the word name 00:23:08 its not a direct translation thing, cuz freeforth compiles an anon definition and executes it instead of intrepreting 00:23:20 does freeforth have a version of the word does> ? 00:23:25 no 00:23:32 do you know what does> does? 00:23:37 yes 00:23:40 what does it do? 00:23:44 and therein lies the rub essentially 00:24:01 I do not know what does> or create or , do 00:24:02 does is the runtime part of the defined word giving the address to the compiled data 00:24:32 so the part on the left side of does> is the compile time behavior 00:24:40 with create/does you have the compiling part and the runtime part 00:24:44 yes 00:24:56 is compile time behavior immediate? 00:25:04 you craft your data between create and does and then you say what you want to do at runtime on the right 00:25:27 hmmm.... have to think about that for a min. I don't think it has to be. 00:25:27 oh I see 00:25:43 --- join: dys (~dys@ip-109-40-2-79.web.vodafone.de) joined #forth 00:25:51 what is the difference between how the compile time code is handled and how the runtime code is handled 00:25:56 create/does is like the mold that makes the mold that makes the object 00:26:33 : test create 2 + does> 2 + ; 00:26:46 you have to be more specific. The compile time code is taking stuff off the stack that you can tweak then comma into the run time version's data area 00:27:02 see, that wouldn't work cuz you didn't compile any data 00:27:04 My stack is empty 00:27:13 I typed gforth RETURN : test create 2 + does> 2 + ; 00:27:22 you'd have to do something like : test create 2 + , does> @ 2 + ; 00:27:30 what does , do? 00:27:52 comma places the value on the top of the stack here in the dictionary and advances the dictionary pointer 00:28:06 essentially, comma is forth's compiler 00:28:15 here is a word which produces a free pointer correct? 00:28:24 , requests a new free pointer 00:28:25 here is the pointer 00:28:35 here produces a new pointer if you type , 00:28:57 so comma is something like : , ( n) here @ ! 1 cell here !+ ; 00:29:12 what does !+ do? 00:29:12 oops, I mean +! 00:29:29 s/!+/+! 00:29:50 +! is like : +! ( n a) dup >r @ + r> ! ; 00:30:16 it just increments the value at addr a by val n 00:31:13 John[Lisbeth]: have you tried reading starting/thinking forth yet? 00:31:21 here returns the pointer to the start of free data space. it doesn't return a pointer to the pointer, so you can't ! to it like that 00:31:24 I prefer to skim 00:31:44 I am sufficiently certain my assumption abut what , does is correct 00:31:47 ok, yeah, its been a few years since I've used a standard forth 00:32:03 , seems to just put a new pointer in here 00:32:21 so comma is more like : , here cell allot ! ; 00:32:33 so if you did something like here . 100 , here . you'd get a printout of two numbers advanced by one cell 00:32:43 is here a pointer to an index in an array? 00:32:47 kevinfish: yes 00:33:12 well, if you call the unallocated forth dictionary space an array, then yes 00:33:18 but it usually isn't thought of like that 00:34:23 traditionally forth is just a big block of memory with the code/headers at the bottom growing up, the disk buffers at the top, and the stacks at the top growing down 00:34:38 what if I say : anonymous-variable here cell allot ; 00:35:06 that will leave here on the stack 00:35:10 yes 00:35:16 because it is anonymous and not a variable 00:35:25 not a named variable 00:35:37 yeah, I guess that would work 00:35:41 I see 00:35:49 so then variables in forth are actually quite literally just forth words 00:35:53 but then 00:36:17 when I allot cells to a variable that certainly does not add to the dictionary does it? 00:36:43 John[Lisbeth]: ?? 00:36:48 allot cells to a variable? 00:36:49 usually you can do something like variable foo 0 , if you want a double sized var 00:36:54 variable foo 10 cells allot 00:37:06 but that may be breaking the standard 00:37:11 John[Lisbeth]: that is not guarnateed to give the behavior you want 00:37:20 you want create there 00:37:33 variable foo create 10 cells allot 00:37:38 right, some forth's may create those with malloc 00:37:46 in which case they wouldn't be contigious 00:38:12 John[Lisbeth]: Create foo 10 cells allot 00:38:54 why is create foo 10 cells allot better than variable foo 10 cells allot ? 00:39:15 John[Lisbeth]: simply because the standard doesn't guarentee they would be contigious with variable 00:39:29 I see so create is ans standard. Fine works for me 00:39:31 that may call two calls to its heap manager 00:39:33 and even if they were, you would be allocating 11 cells there 00:40:13 it would work fine with traditional forths but the standard allows for non-traditional memory maps/allocations 00:40:26 so lets say I do create foo 10 cells allot 00:40:31 Is that being added to my dictionary? 00:40:41 or are those 10 cells somewhere else 00:40:48 well, depending on how the forth is implemented 00:41:00 does ans specify? 00:41:11 most common forths use a single upwardly-growing heap where the dictionary and alloted data space are just appended onto the end 00:41:13 see, like freeforth actually has what it calls a "heap" where data/code goes and a symbol table where word headers go 00:41:28 but it's implementation-dependent 00:41:55 but according to the standard, for example, data and programs could go in different places, or each call to allot could get allocated to a completely different place 00:41:58 --- quit: dys (Read error: Connection reset by peer) 00:42:56 or they could go the same place and the implementation could add other stuff after the variable's data cell 00:42:58 John[Lisbeth]: what language(s) are you sued to? 00:43:09 s/sued/used 00:43:39 I learned c++ then bash then haskell then lisp then javascript then forth 00:43:58 approximately 4.5 years of programming. 00:44:17 ya, sux how most people start out with an algol variant 00:44:24 messes them up 00:44:33 I'll keep that in mind 00:44:46 bash and c++ Fucked me up 00:45:04 c++ was awful and I hated it and bash was even worse. It was like stepping out of a bad home into a hyper abusive psychotic home 00:45:04 I was semi-lucky. I started out with some basic, then assembly, then forth, then C then lisp 00:45:12 lol 00:45:34 and so after 2 bad languages in a row the problem I wanted to solve was how to choose a language were due to design choices I would never logically be forced to switch. There would not be a thing that that language should be able to do that it could not. 00:45:37 And here I am. 00:45:46 by the time I got to C I realized it was bullshit to have to jam all my programs into an infix algebraic formula 00:46:38 forth is the most satisfying language I've used, but lisp is the 900 lb gorilla that can't really be ignored 00:46:56 C would be so much nicer if it had a decent macro system 00:47:05 well, there's always m4 00:47:13 but that's YET another language to learn 00:47:29 yes, there's always that and then it doesn't know about C's syntax at all 00:48:15 yeah, I used it to craft my 4th in c headers. It did it in about 1/2 a page of code that mostly looked like line noise 00:51:14 I spent 4 years looking for forth essentially 00:51:18 and lisp and forth both do what I want 00:51:24 but forth is simpler 00:51:26 John[Lisbeth]: I can't really think of a language like that. I mean you can do just about anything in forth pretty elegantly, but you might have to build a lot of the utilities yourself 00:51:43 with lisp you can get most everything prebuilt 00:51:51 Have you tried clojure? 00:51:58 so lisp can rewrite it's self completely from the core up and so for that you get everything forth has because you can turn lisp directly into ans and remove all the lisp parts 00:52:12 so lisp and forth are equivelant in their abilities 00:52:24 except that nothing crafts tight code like forth 00:52:32 lisp just happens to require some pieces to it which are much more complex and add bulk 00:52:37 and forth does not have those pieces 00:52:40 not by default 00:52:53 FreeForth is the tightest code I've seen so far 00:53:09 but sadly its only for x86 right now 00:53:13 and not metacompiled 00:53:59 lbforth looks pretty cool. Bootstraps with a metacompiler written in lisp to get you a c-forth and then that is used to metacompile the forth to several different common cpu targets 00:54:26 freeforth is probably pretty tight for a couple reasons 00:54:50 first of all if it is indeed x86 then that makes it tighter than gforth at the least because gforth is written in c to my knowledge and x86 produces smaller forths than c 00:54:54 it saves a lot of stack popping. For example it renames registers rather than swapping 00:55:33 plus it uses the cpu's on status register and compiles jumps based on those flags so that has got to be screaming fast 00:55:42 furthermore ans forth is meant to be a robust forth that could meet the needs of everything a programmer could logically want from their computer to do. That adds bulk 00:55:44 s/on/own/ 00:56:02 Not only are there things not every programmer will need in ans forth but ans forth makes some speed sacrifices just to be portable and backwards compatible 00:56:09 what I'd like to see is a freeforth metacompiled with several target cpu's 00:56:32 freeforth on the other hand is made by a single person and is suited the best to their own needs. There is no telling how much effort was spent on backwards compatile or potability outside of x86 00:56:40 *portability 00:56:53 not that it would not be portbae outside of those environments but doing so would add to the bulk 00:57:02 it would be a challenge, but I don't see anything inherent to x86 in the language's basic design 00:57:23 the point being that a person's own personal forth is way faster than one they got from someone else because it works just for them 00:57:27 that is unless you were dealing in a very different type of cpu 00:57:47 so you really have to choose between portability and speed once you get to a certain point 00:57:57 and it is not black and white but a gradient between portability and speed 00:58:45 basically 00:58:59 I think freeforth could be pretty portable between most conventional cpu designs 00:58:59 if you had a near perfectly abstract machine it may be very slow and if you had a near perfectly fast machine it would only work on your computer 00:59:20 Or you can choose something which is portable to most machines and decently fast 00:59:39 its just that porting freeforth would be a moderately difficult task because there is a lot of (currently hand crafted) assembly 00:59:55 One of the issues with this though is that there are very few forth developers out there compared to other languages 01:00:19 there is one called ff by the author of chicken scheme that has very little assembly but it would probably suffer a bit in speed. Eforth is very similar 01:00:31 Compared to lisp we have very few developers, but lisp already is very spares of developers 01:00:38 yeah, that's why I've been forced to use lisp for most things 01:00:45 indeed 01:00:51 we forthers can not afford to be split 01:00:57 not in my opinion 01:01:04 I'm liking clojure currenty for the broadest coverage 01:01:26 ans is probably the closest thing we have to something that we can agree on. The previous forth standards by moore and his people should be just as good. 01:01:29 I save forth for those tasks where nothing else comes close, like embedded and drivers 01:01:40 I want to use forth for high level programming 01:01:50 then you should probably try factor 01:02:17 In my mind factor is not a good prescident for how it should be done. I think gforth is more appropriate 01:02:41 gforth is certainly a fine system. Isn't it gpl tho? 01:02:53 Yes. This seems like the most logical license. 01:03:00 Although because ans is not gpl this is solveable 01:03:08 kevinfish: why the lisp metacompiler for the C target and forth for all other targets? just historical reasons and inertia? 01:04:01 koisoke: yeah, idk, but I see on his todo list eliminating that 01:04:35 John[Lisbeth]: well, I convinced dirk zoller to use lgpl on pfe so it could be used for commercial projects 01:05:00 generating C is different enough from generating machine code to memory/an image, so two different metacompilers makes sense i guess 01:05:12 ans is what we are talking about not gforth 01:05:17 gforth is an implementation of the ans standard 01:05:39 --- quit: xek (Quit: Ex-Chat) 01:06:11 right, but before you can use it for a commercial project (where you don't want to release all your source code) you're going to have to find/build a non-gpl version of ans 01:06:26 --- join: dys (~dys@2003:5b:203b:100:6af7:28ff:fe06:801) joined #forth 01:08:13 Chicken scheme looks pretty nice for creating stand alone apps 01:09:15 IN order to get forth to be developed to the point where it is as fast and capable as lisp we need at least as many developers as there are lisp developers working on an implementation of ans 01:09:30 and a pretty liberal license: https://opensource.org/licenses/BSD-3-Clause 01:09:30 if it is proprietary then that will not happen because the source will be closed 01:09:41 with a open or free license it could occur 01:10:23 but part of the problem is that many languages are alot easier than forth by far 01:10:29 at least to a new programmer 01:10:32 John[Lisbeth]: do that and you will get something so bloated it is unsuitable for what most people want to do with forth 01:10:33 such as factor 01:10:39 exactly 01:10:43 and that is why you don't do it 01:10:43 well, forth is probably faster, but it could probably use something like clojars/boot for its libraries and build tools 01:10:52 you create a word which will do if you use that word 01:10:59 but if you don't use that word it doesn't happen 01:11:41 this has, in essence, been the same discussion in Forth since I got into it in the early 80's 01:12:05 which is? 01:12:31 reusable libraries, version control, bringing in new blood, etc... 01:12:41 so on the one hand 01:12:51 I think there is no doubt that the ans standard already has libraries built in 01:13:10 I think that people simply do not write and publish them in portable ans forth 01:13:31 there are some standard libraries out there. Nothing on the order of what most other languages have tho 01:13:34 for version control we have git. We do not need to build it into the language 01:13:51 git is great and robust version control 01:13:52 Well, for example, check out clojure/hoplon/boot 01:14:05 it can be done locally if you dont want to release your code to github 01:14:22 and those aren't the only thing we need to bring in new blood 01:14:42 clojure makes sure that you build only against certain versions of libraries so upgrades in them don't break your code 01:14:54 here is my reasoning 01:14:57 I don't know of anything like that in forth yet 01:15:07 but I've been out of it for a while, so maybe there is something now 01:15:14 clojure is immutable 01:15:19 forth is not 01:15:23 we must keep this in mind 01:15:28 John[Lisbeth]: immutability has its advantages 01:15:33 we can make immutable libraries in code but we do not have to 01:15:33 kevinfish: i'm not either, but it is hard to have package management like that when you don't have a package system 01:15:40 we can try to control state at the least 01:15:48 but we do not have to 01:16:04 it wouldn't be that hard to make an immutable forth IMO 01:16:09 imagine gforth but lets change the name to xforth because I mean any forth at all which claims to be ans compliant 01:16:16 lets say xforth is a project on git which does this 01:16:52 if the xforth developer and I are working separately and I use xforth as a base then I can put whatever kind of high level stuff on it I want and the forth below that he made will still be usable for whatever level he intended it for. 01:17:43 you see, I think early on we need to make up our mind: are we going to rely on pre-existing tools to do the machine level optimization, or are we going to let forth do it. If the former, we should probably just stick with a C version because the compilers that are out there are hard to beat. If the latter then something like freeforth is hard to beat 01:17:43 furthermore when you add to an ans forth you need not temper with the ans standard at all 01:17:58 both 01:18:08 not only can we depend on forth but we can use machine tools if we like 01:18:23 and we can also try to implement those machine tools in highly optimized forth as moore proposed 01:18:46 as moore more or less put it if you have written 2 + 2 - you have wasted your time 01:18:57 If we are going to do forth in C then we have the same issues that scheme has: http://home.pipeline.com/~hbaker1/CheneyMTA.html 01:19:09 And you can take a forth program and rip out every single thing that does not do what you want and it becomes one tiny thing that does what you want very well. 01:19:45 If people want forth to behave like a machine component of an operating system they need to remove the parts that are not helping the machine 01:20:08 but if they want to use it more like a shell like a lisp repl is then in that version they need to keep it all in one big heap or what have you 01:20:42 Forth should do everything and if you have been handed a forth that you assume does everything and then you make it so that forth doesn't do everything anymore then you fucked it up 01:22:04 and that includes making it so it can't be used as low level anymore 01:22:08 at least in my opinion 01:22:11 well, you could always do something like target compiling to eliminate all the words that weren't called 01:22:17 kevinfish: i don't see C's advantae there as peformance. naive subroutine threading + peephole optimization is only worse than a good compiler by a factor of ~2-3, so not really worth the overhead and longer compilation turnarounds for most uses. but cross-platform compatibility can be worth something 01:22:45 eliminating all the words that were not called is a very important step but there are more things that you can do 01:23:25 koisoke: yeah, I'm thinking targeting some other portable assembler might be more fruitful 01:24:10 First of all I think if you have something like ans forth or smaller that certainly means you can do low level with it 01:24:16 I think low levelers clearly have no problem using forth already 01:24:44 high levelers have trouble using forth 01:25:32 and I think I can take a forth and if that forth does everything then I can modify that forth to be easy for high levelers and still do everything, plus a few kilobytes 01:25:50 and I think the more experienced I get at forth the less those kilobytes will be 01:26:13 Just the bare minimum to make it as easy as something like javascript 01:27:04 in part it is just making alternate words for words that are cryptic looking. Making a version of : and ; called define and finish-define 01:27:13 or finish-definition 01:27:27 https://www.gnu.org/software/lightning/ 01:27:31 John[Lisbeth]: Why not 'def' and 'endef'? 01:27:38 sure why not indeed 01:27:40 equally phonetic 01:27:42 John[Lisbeth]: nobody wants cobol fingers 01:27:49 certainly not 01:27:58 but you see my point 01:28:06 : dereference @ ; 01:28:08 see what I mean? 01:28:08 : and ; take about a minute to learn and is not what puts people off of forth 01:28:13 and for example 01:28:23 teach new forth programmers to use the { word to get their arguments 01:28:47 the { word makes forth alot easier and comes with ans and makes it much more like lisp or c 01:28:53 it comes for free in gforth so why not 01:29:04 It also enforces comments which are important 01:29:32 It just a matter of adding a few words to forth and not running any of them 01:29:46 well, one thing non-forther's are going to find difficult is to properly factor their code so that the few stack operators can actually get to the arguments, but I've found that to be one of the BEST things about it. To be forced to do it so you are forced to think about (and name) every part of your program 01:30:05 kevinfish: I can vouch for that, 01:30:22 I dunno I think the { word is really great 01:30:37 It helped me solve a problem where I thought I'd need a ~10G database in ~2M of RAM. 01:30:42 that's why I think the BEST way to promote forth is to get to people BEFORE they learn coding the wrong way with one of these algol based langauges 01:30:42 and I think you cant just expect people to like : ($($*$ @ ; 01:30:45 look at it 01:30:55 : ($*($*&(#*$&%(*Q&$(%*$%98374598347598a9&H&h9s8dh98fdsg @ ; 01:31:05 see? 01:31:12 you can do it but give them an easier way to type it 01:31:34 define mything dereference findef 01:31:40 way nicer 01:31:44 there was a forth-like language called "Icon" that was supposed to be good for beginners. It used pictures/icons instead of words 01:31:59 there is a key concept I Think I am not getting accross though 01:32:13 How much space does it add to a forth to define a new word? 01:32:22 on average if you coudl guess 01:32:30 John[Lisbeth]: very little. 01:32:37 It used to be only about 6 bytes 01:32:37 for an empty def, a few dozen bytes 01:32:38 indeed 01:32:46 very very little 01:32:51 back in the days when it was a count byte and the first 3 letters :) 01:33:20 there was a back link, and a link to the code 01:33:22 by todays standards and especially by tomorrows standards too little for us to bother 01:33:25 a data field was optional 01:33:34 defining a word takes very little room at all most of the time 01:33:36 John[Lisbeth]: punctuation has the advantage of being visually distinctive when skimming code 01:33:44 now if I define a word that word may do alot of things, right? 01:33:48 if I use it 01:33:52 if I decide to use that word it may do alot 01:33:53 and now with seporate symbol tables you can toss the symbols when you are done compiling 01:34:07 but if I just define it it is mostly harmless right? 01:34:29 and proliferating synonyms to shave a few minutes off the learning curve has huge costs in readability 01:34:29 I can define a gigantic abstraction system which puts you in all kinds of crazy types and immutable and object oriented and all that jazz and never use any of it 01:34:31 yeah, generally in today's context, definitions are trivial space wize 01:34:31 am I wrong? 01:34:37 define the whole world 01:35:10 I can define a whole object oriented immutable purely-lazy purely-jazz-hands whatchyamacall it doodad system and never run it 01:35:14 just define it 01:35:15 and then 01:35:19 sure 01:35:21 I can define words which control the syntax of it 01:35:28 and then I can define a word which enters into that syntax as a macro 01:35:36 if you have a forth with a good dictionary, like BBL had, with one thread per word, so it still compile's fast, having a lot of definitions will barely be noticed 01:35:39 and then you type gforth thisfileiamtalkingabout.txt 01:35:42 and what happens? 01:35:55 what happensis I am greeted by the standards ans forth prompt 01:36:13 because I have not changed ans at all. Ans is still there. I added a few kilobytes 01:36:34 IDK what modern forth has the dictionary like BBL had, but its worth checking out. It literally gobbles code. No need to keep a compiled version. 01:36:58 this is true 01:36:59 gobbles code? 01:37:00 either way will do 01:37:10 I think they mean it will "interpret" fast 01:37:12 yeah, compiles code at an amazing rate 01:37:16 indeed 01:37:18 that being said 01:37:24 it compiles it so fast you think it didn't work! 01:37:33 we can indeed put an entire high level language on top of ans forth and have it be not very much larger than ans forth 01:38:04 and you would just type a woord liek iamanoob and it would put you in that other language 01:38:15 IIRC webassembly is going to be based on a concatenative/stack language very much like forth 01:38:27 he he he he he 01:38:31