2023-05-14 02:55:27 <veltas      > My IRC client is written in awk 2.0 but I've never bothered learning it properly
2023-05-14 08:29:20 <KipIngram   > Well, the point of builds> does> yesterday was so that I can use them to write up vocabularly words.
2023-05-14 08:29:49 <KipIngram   > Should be fairly straightforward - I remember how I went about it in the past.
2023-05-14 08:30:46 <KipIngram   > I think FIND is already compatible, but I haven't actually tested it in multi-vocabularly conditions.  But it has code that will walk a list of them.
2023-05-14 08:30:59 <KipIngram   > It's just only walked a list of one to date.
2023-05-14 08:31:17 <thrig       > I Walk the Line (The Best Of The Johnny Cash TV Show)
2023-05-14 08:31:38 <KipIngram   > Heh.  Theme music for Forth words.
2023-05-14 08:32:14 <KipIngram   > My wife's dad was a particularly big Johnny Cash fan.
2023-05-14 08:33:14 <KipIngram   > He was a mechanical engineer for Dow Chemical, but it a lot of ways never stopped being an Iowa farm boy.
2023-05-14 08:34:07 <vms14       > supposedly is good if you make your metacompiler able to bootstrap as much as possible
2023-05-14 08:34:15 <vms14       > a common thing in lisp and forth
2023-05-14 08:34:38 <vms14       > define some few primitives and the rest is defined in the language itself
2023-05-14 08:34:43 <KipIngram   > Cash was a Morse code operator while he was in the service and monitored Soviet radio traffic.  He was like the first person to receive information about Stalin's death.
2023-05-14 08:34:52 <vms14       > it adds overhead, but also portability
2023-05-14 08:35:34 <KipIngram   > vms14: I'm not sure how much overhead it would HAVE to add.  What makes you say that?
2023-05-14 08:35:51 <vms14       > it depends on your compiling method
2023-05-14 08:36:05 <vms14       > in forth colon words usually have very little cost
2023-05-14 08:36:12 <KipIngram   > My test of it is to have a system with a particular md5 hash that I can re-meta-compile and get a system with the same hash.
2023-05-14 08:36:23 <KipIngram   > I want it to be able to recreate a perfect image of itself.
2023-05-14 08:36:39 <vms14       > :0
2023-05-14 08:36:58 <vms14       > I keep playing with awk xD
2023-05-14 08:37:28 <vms14       > I always liked the unix philosophy and awk resonates with it
2023-05-14 08:37:56 <vms14       > I always wanted a language able to control other processes
2023-05-14 08:38:20 <KipIngram   > I know fairly little about awk.  I've only used it lightly and when I do I just search online for a magic awk spell to do what I need.
2023-05-14 08:38:30 <KipIngram   > I don't really "know" it at all.
2023-05-14 08:38:36 <vms14       > it's a performance nightmare, but executing commands decouples you from almost everything
2023-05-14 08:38:42 <KipIngram   > I use sed more often.
2023-05-14 08:38:51 <KipIngram   > sed and grep cover most of what I need to do.
2023-05-14 08:38:57 <vms14       > the command can be written in any language and by any person
2023-05-14 08:39:12 <vms14       > you just need to use messages
2023-05-14 08:39:34 <vms14       > you can have daemons that start yourself and pipe them to send commands and have a response
2023-05-14 08:40:28 <vms14       > for example I can make a c program that reads from stdin and has a bunch of strcmp
2023-05-14 08:40:42 <vms14       > the results if any just need to be printed on stdout
2023-05-14 08:41:10 <vms14       > it can be a one time command or a daemon
2023-05-14 08:41:16 <KipIngram   > :-)  Yes, that's how it works.
2023-05-14 08:41:28 <vms14       > KipIngram: it's powerful
2023-05-14 08:41:48 <vms14       > but the exec is expensive and some people might not like that
2023-05-14 08:42:06 <KipIngram   > It is -  my whole test harness set up for this new test stand I got last fall is written primarily in bash and along these lines.
2023-05-14 08:42:25 <KipIngram   > Doesn't matter - it's managing tests that take hours or days to run.
2023-05-14 08:42:45 <vms14       > I discovered netbsd has its own test suited named atf
2023-05-14 08:42:57 <KipIngram   > So for that kind of task it's irrelevant how it performs.
2023-05-14 08:43:32 <vms14       > https://man.netbsd.org/NetBSD-9.0/atf.7
2023-05-14 08:43:39 <KipIngram   > I have 12 drives in this stand; each one has its own directory on the stand and I can run as many instances of the test engine as I want.
2023-05-14 08:44:10 <vms14       > I've cd /usr/tests/ and run some tests, but got bored after a million  and stoped them
2023-05-14 08:44:11 <KipIngram   > And it's set up so that I actually run it from a different machine - a "controller," so to speak - I could have as many stands active as I liked.
2023-05-14 08:44:51 <KipIngram   > The directory names match the PCI address of each drive.
2023-05-14 08:44:54 <vms14       > I have to add a test suite on my lang
2023-05-14 08:45:14 <vms14       > usually is just to exec a program and look at the output
2023-05-14 08:45:17 <KipIngram   > Yeah, a test framework for something like that is good.
2023-05-14 08:45:27 <KipIngram   > I don't have one at the moment, but eventually want one.
2023-05-14 08:45:34 <vms14       > but in the same lang you just test function results
2023-05-14 08:46:24 <vms14       > KipIngram: what about an error system?
2023-05-14 08:46:37 <vms14       > you made something with signals I remember
2023-05-14 08:46:50 <vms14       > you turned a segfault into an error handler
2023-05-14 08:46:51 <vms14       > xD
2023-05-14 08:47:06 <KipIngram   > That's mostly how I've tested this drive test system - I just run it, several drives at a time, and see that it produces the expected results.
2023-05-14 08:47:40 <KipIngram   > Well, in Forth the traditional error response is just to restart the interpreter, usually with a clean stack.
2023-05-14 08:48:06 <KipIngram   > All I've added to that is the snapshotting, so that it restores a prior condition when you get back to the interpreter.
2023-05-14 08:48:14 <KipIngram   > And yes, it has limited signal support.
2023-05-14 08:48:25 <vms14       > ah right, you save the image every line
2023-05-14 08:48:26 <KipIngram   > Catches memory addressing errors and divide by zero type things.
2023-05-14 08:48:52 <KipIngram   > Yes, but that's really kind of an "add on" and isn't part of the "nitty gritty" of error recoevery.
2023-05-14 08:49:09 <KipIngram   > That signal stuff was hard to write, because it's all not documented very well.
2023-05-14 08:49:18 <KipIngram   > I really had to dig to find information on it.
2023-05-14 08:49:24 <vms14       > can you come back to more than one line?
2023-05-14 08:49:40 <KipIngram   > And it's got some very processor specific code in it.
2023-05-14 08:50:01 <KipIngram   > I don't understand - if you finished the previous line successfully then there's no reason to roll back further than that.
2023-05-14 08:50:08 <KipIngram   > No - it can't go back multiple lines.
2023-05-14 08:50:18 <KipIngram   > It overwrites the snapshot each time it finishes a line without error.
2023-05-14 08:50:35 <vms14       > does it save the image every line inside a colon definition that spans several lines?
2023-05-14 08:50:48 <vms14       > or it's turned off
2023-05-14 08:51:13 <KipIngram   > Forth interprets one "line" at a time, but what it regards as a "line" might be an entire block of code.
2023-05-14 08:51:25 <vms14       > so a colon definition is just a line
2023-05-14 08:51:28 <KipIngram   > Or many blocks of code.  Let's say I say "<block #> LOAD."
2023-05-14 08:51:40 <KipIngram   > That block might load many other blocks, recursively.
2023-05-14 08:51:52 <KipIngram   > But the snapshot is made at the beginning of the line I TYPED.
2023-05-14 08:52:05 <KipIngram   > It doesn't matter how many blocks it loads - it doesn't update the snapshot while doing that.
2023-05-14 08:52:16 <KipIngram   > So if there is an error anywhere in the code I loaded, it undoes all of it.
2023-05-14 08:52:22 <vms14       > ah, as you use blocks to save colon definitions there and then load them, you can roll back to the previous line after a load block
2023-05-14 08:52:27 <KipIngram   > Otherwise I'd have to issue a different command when I re-tried.
2023-05-14 08:52:41 <KipIngram   > It's all about being in a position to re-issue exactly the same typed line.
2023-05-14 08:53:10 <vms14       > KipIngram: do you have the ability to set "checkpoints"?
2023-05-14 08:53:19 <KipIngram   > I want to be able to just hit the cursor-up key to scroll back to the line and fix it and hit enter again.
2023-05-14 08:53:36 <KipIngram   > No, I don't have any other snapshotting type feature.
2023-05-14 08:53:44 <vms14       > well, what's different from using forget and recovering from your snapshot?
2023-05-14 08:53:53 <vms14       > I assume also the call stack is restored
2023-05-14 08:54:07 <KipIngram   > I do think that with the new one I'll be able to write an image of a running system off to disk, though.
2023-05-14 08:54:31 <KipIngram   > It would just kind of "reverse the image load process."
2023-05-14 08:54:42 <KipIngram   > Turn the addresses back into offsets, etc.
2023-05-14 08:54:52 <KipIngram   > And create a new image in blocks that I could later load and run.
2023-05-14 08:55:21 <KipIngram   > The way I'm structuring this one will make it a lot easier to identify which items need to be address/offset converted.
2023-05-14 08:55:34 <KipIngram   > In my prior systems they get all mixed together and it's unclear how to do that.
2023-05-14 08:56:00 <KipIngram   > That is, the value of some cells depends on where in RAM the code is loaded.
2023-05-14 08:56:26 <KipIngram   > In the new system, those cells will be well-segregated from the others.
2023-05-14 08:56:33 <KipIngram   > Mostly will just be the table entries.
2023-05-14 08:57:11 <KipIngram   > Trying to save current stacks would be difficult, so I probably won't try.
2023-05-14 08:57:17 <KipIngram   > You've got all these numbers on the stack.
2023-05-14 08:57:22 <KipIngram   > Which ones are addresses?
2023-05-14 08:57:31 <KipIngram   > In a typeless Forth system there's no way to tell.
2023-05-14 08:58:00 <KipIngram   > So if I save an image off to disk and reload it later, it will load with empty stacks.
2023-05-14 08:58:39 <KipIngram   > Anyway, I guess that is close to what you would call a "checkpoint," so no I don't support that now but yes, I figure I will eventually.
2023-05-14 08:59:44 <vms14       > you need a way to serialize everything you want
2023-05-14 08:59:54 <vms14       > the common way would be some sort of bytecode
2023-05-14 09:00:33 <vms14       > it is expensive, but you said you wanted to output to a disk
2023-05-14 09:00:49 <vms14       > you need to deserialize it later
2023-05-14 09:01:09 <vms14       > how you represent this data is the most important part of the whole process
2023-05-14 09:01:27 <vms14       > as some book told me: Data is more important than algorithms
2023-05-14 09:01:40 <vms14       > and mainly data representation affects the whole algorithm
2023-05-14 09:02:47 <vms14       > crc has done this already I guess
2023-05-14 09:03:02 <thrig       > there is Lore that Data ...
2023-05-14 09:03:14 <vms14       > he can make a vm image and load it from a file or whatever
2023-05-14 09:05:20 <KipIngram   > Well, vms14: the main idea is to use Forth source code to store things, and then you just load and compile it.  I'm not going to "serialize" for this image save - I've got a plan for how to write the RAM regions out to blocks.
2023-05-14 09:05:36 <KipIngram   > It will be basically a binary format, though.
2023-05-14 09:05:46 <vms14       > if you don't need to output to a file, it's easier
2023-05-14 09:06:14 <vms14       > you already have that
2023-05-14 09:07:53 <vms14       > I think you can chain restore snapshots if you also save the current restore snapshot
2023-05-14 09:08:22 <vms14       > say every line you save the state, you also save what was saved in the previous line, kind of a linked list
2023-05-14 09:08:59 <vms14       > well won't work unless you save all this into memory and mantain it, so meh
2023-05-14 09:09:23 <vms14       > a single checkpoint is already too expensive for a forthwright
2023-05-14 09:10:52 <vms14       > and there's the rule that you don't have to implement what you don't already need
2023-05-14 09:11:11 <vms14       > yagni, for short
2023-05-14 09:12:02 <vms14       > but if it's all about research, then no rules exist :D
2023-05-14 09:12:09 <vms14       > no rules for me
2023-05-14 09:13:20 <KipIngram   > Yeah, the snapshotting dovetails with command history.
2023-05-14 09:13:30 <KipIngram   > But I'm not *interested* in saving a whole history of snapshots.
2023-05-14 09:13:51 <KipIngram   > I just keep my last proven good state so I can restore it if my attempt to move forward further fails.
2023-05-14 09:13:53 <vms14       > you could limit it to 10, but you said already you don't have the need for more than one
2023-05-14 09:13:56 <KipIngram   > It's enough for my needs.
2023-05-14 09:14:09 <vms14       > what I would be interested in is to be able to set a checkpoint at any time
2023-05-14 09:14:14 <KipIngram   > It's already an expensive feature in terms of RAM - I don't want dozens of these snapshots hanging around.
2023-05-14 09:14:37 <KipIngram   > That's a totally different thing, and as I said, I think on this next system I will be able to write a system out to a set of disk blocks.
2023-05-14 09:14:50 <KipIngram   > But that will be something I do manually when I choose to.
2023-05-14 09:15:25 <KipIngram   > It's really fairly simmple - I'll have four or five RAM regions that serve different purposes in my system.
2023-05-14 09:15:38 <KipIngram   > I'll just write each one out to a block / some blocks.
2023-05-14 09:15:51 <KipIngram   > "de-converting" addresses where necessary.
2023-05-14 09:16:20 <vms14       > oh, simple multitasking could be derived from that
2023-05-14 09:16:23 <KipIngram   > The only difference between an image in blocks and one in RAM will be that the block image has all offsets and no addresses.
2023-05-14 09:16:56 <KipIngram   > Well, the way I plan this next system it should be entirely  possible to have multiple console sessions running.
2023-05-14 09:17:09 <KipIngram   > Something that operates kind of like screen.
2023-05-14 09:17:37 <KipIngram   > It's going to support multi-threading, and any thread could have a console if it wanted.
2023-05-14 09:18:11 <KipIngram   > I don't intend to try to support having a bunch of threads all compiling at the same time, though.
2023-05-14 09:19:23 <vms14       > I have increasing interest in a language based in executing commands
2023-05-14 09:19:36 <vms14       > any command would be a separated thread xD
2023-05-14 09:19:42 <KipIngram   > Zarutian_iPad: those builds> does> definitions we wound up with at the end last night are 38 characters long each.  Now that's the kind of definition I like.
2023-05-14 09:20:15 <vms14       > but the thing that interests me more is that no matters the language the command is written
2023-05-14 09:20:40 <KipIngram   > :-)  Then your interests are somewhat remote from Forth at the moment.
2023-05-14 09:20:41 <vms14       > a real programmer knows every programming language has better use cases than others
2023-05-14 09:20:43 <KipIngram   > But that's fine.
2023-05-14 09:21:27 <KipIngram   > https://madhadron.com/programming/seven_ur_languages.html
2023-05-14 09:21:43 <KipIngram   > That guy contends that knowing one of each major "type" is adequate, and that makes sense to me.
2023-05-14 09:21:58 <KipIngram   > He identifies seven major "families."
2023-05-14 09:22:15 <KipIngram   > Each with a "prototypical" language.
2023-05-14 09:23:17 <KipIngram   > I'm interested in the "capabilities" of the various languages, but what I'd like most if to have those capabilities (or as many as feasible) unified under a one-language hood.
2023-05-14 09:23:49 <KipIngram   > For example, I'd like to be able to "extend" my Forth to support types and objects and so on, but it would still "feel" a lot like the original Forth.
2023-05-14 09:24:15 <KipIngram   > maybe I load up a custom interpreter for some particular situation, like numerical analysis, etc.
2023-05-14 09:24:29 <KipIngram   > And when I'm done quit out of that interpreter back to the base system.
2023-05-14 09:24:57 <KipIngram   > So the basic Forth would be like a "manager" that could launch more sophisticated language processing facilities.
2023-05-14 09:25:24 <vms14       > I mainly meant for example you'd use perl awk or sed for a text related task rather than maybe C Java php etc
2023-05-14 09:26:18 <vms14       > you use sql for mysql, etc
2023-05-14 09:26:56 <vms14       > so if you can build a program using a combination of several programming languages, you can use the best suited language for every part of the program
2023-05-14 09:27:12 <vms14       > it's just that this implies fork and exec and it's a bit annoying
2023-05-14 09:27:13 <KipIngram   > I know, and that makes total sense.  But I'd prefer to use a common foundation.
2023-05-14 09:27:37 <KipIngram   > One of the things I love about Forth is that it maintains a consistent "syntax" regardless of how far you've extended it.
2023-05-14 09:28:03 <vms14       > KipIngram: that's why, a word can be a whole different word
2023-05-14 09:28:16 <vms14       > when you enter a word, you start a journey xD
2023-05-14 09:28:26 <KipIngram   > Yes.  You don't have the functions you add "looking different" from the native words.
2023-05-14 09:28:49 <vms14       > it will unify all those languages
2023-05-14 09:28:57 <KipIngram   > In  C your interface to a self-written function looks VERY DIFFERENT from a basic expression, for example.
2023-05-14 09:29:29 <KipIngram   > But in Forth your "new function" is just another word.
2023-05-14 09:31:27 <KipIngram   > All languages give you a way to add functionality.  Forth and Lisp are the only ones I know that let you "change the language" in a real way.
2023-05-14 09:31:52 <vms14       > a colon word would combine several languages (programs) in its definition
2023-05-14 09:32:20 <KipIngram   > I guess there's no reason you couldn't do that, but I don't have much interest in it.
2023-05-14 09:32:40 <KipIngram   > That feels to me a lot like writing new tools for  Linux.
2023-05-14 09:32:48 <vms14       > yeah, I'm wondering about whether is worth
2023-05-14 09:32:48 <KipIngram   > You can run any of them from the Linux command line.
2023-05-14 09:32:58 <vms14       > I like the decoupling
2023-05-14 09:33:16 <vms14       > and everyone could extend it using its preferred lang
2023-05-14 09:33:26 <KipIngram   > My interest is almost entirely at the bare metal end of the spectrum - I'm interested in a "slick" way to manage the hardware resources.
2023-05-14 09:33:27 <vms14       > the only requirement is read from stdin, write to stdout
2023-05-14 09:34:19 <vms14       > words could be files XD
2023-05-14 09:34:37 <KipIngram   > If I were really doing this right I'd be writing code to manage all the processor's memory management hardware (descriptor tables and so on).  But that's difficult when you're coexisting with an operating system, so... later.
2023-05-14 09:35:21 <KipIngram   > For now I'll just give the system a huge block of RAM and manage it with software.
2023-05-14 09:35:46 <vms14       > 
ACTION wonders about a folder being the dictionary

2023-05-14 09:36:11 <KipIngram   > Forth has vocabularies, which let you manage the structure of your namespace.
2023-05-14 09:36:24 <KipIngram   > It's still one "dictionary," but the word lists get a "tree structure."
2023-05-14 09:36:32 <vms14       > vocabularies/main/+
2023-05-14 09:37:18 <KipIngram   > But in Forth that has little to do with any kind of file system, since typically you don't have one.
2023-05-14 09:37:36 <vms14       > the find command would actually be used to find a word
2023-05-14 09:37:52 <vms14       > find vocabularies '+'
2023-05-14 09:37:57 <KipIngram   > :-)  Yes, that's what it does.
2023-05-14 09:38:10 <KipIngram   > In my system I have a variable PATH that points to some vocabularly.
2023-05-14 09:38:29 <KipIngram   > That's the first voc of a linked list of vocs - FIND searches them in order and only fails after it's tried all of them.
2023-05-14 09:38:43 <KipIngram   > And there's another variable CURR which designates which vocabulary to add new definitions into.
2023-05-14 09:38:51 <vms14       > ln `which find` vocabularies/core/find
2023-05-14 09:39:02 <KipIngram   > Executing vocabularly words lets you modify that search order.
2023-05-14 09:39:12 <KipIngram   > Take vocabularies out of it, add new ones in, etc.
2023-05-14 09:39:25 <vms14       > oh and forget can use the timestamp
2023-05-14 09:39:38 <vms14       > ahahaha
2023-05-14 09:39:46 <KipIngram   > I think you're drifting into various aspects of file systems here.
2023-05-14 09:39:53 <KipIngram   > Completely different thing.
2023-05-14 09:40:16 <vms14       > forget needs to remove all words defined after the word you remove
2023-05-14 09:40:27 <KipIngram   > But I think if you intend to use a system for a wide variety of work, you likely must have one.
2023-05-14 09:40:30 <vms14       > so you can look at their timestamps
2023-05-14 09:40:41 <KipIngram   > Well, at least it would make things a lot easier.
2023-05-14 09:41:15 <KipIngram   > I'm of the opinion that b-trees are good for organizing a file system.
2023-05-14 09:42:24 <KipIngram   > A big reason I intend to ultimately have one is that I want to integrate how I comment and document in with that.
2023-05-14 09:42:40 <KipIngram   > I want to be able to have hyperlinks in my source to comments and documentation.
2023-05-14 09:42:44 <vms14       > KipIngram: are you going to add literate programming in your forth?
2023-05-14 09:42:55 <KipIngram   > You'll have to tell me what you mean by that.
2023-05-14 09:43:07 <KipIngram   > I know what "literate" means in general,m but I sense you have something very specific in mind.
2023-05-14 09:43:13 <vms14       > literate programming is just document and code at the same time
2023-05-14 09:43:24 <KipIngram   > Oh, yes - I very much want to do that.
2023-05-14 09:43:26 <vms14       > usually using some tool that separates code from documentation
2023-05-14 09:43:34 <vms14       > and generates html, pdf etc
2023-05-14 09:43:41 <KipIngram   > I want to be able to be coding and just issue a key sequence and have a new window open that I can put documentation in.
2023-05-14 09:43:44 <vms14       > example: cweb
2023-05-14 09:43:55 <KipIngram   > Maybe on that last point.
2023-05-14 09:44:05 <KipIngram   > It would certainly be something the basic idea could support.
2023-05-14 09:44:28 <vms14       > https://en.wikipedia.org/wiki/Web_(programming_system)
2023-05-14 09:44:33 <KipIngram   > It'll just support entering text - what kind of functional format that text had will be flexible.
2023-05-14 09:45:14 <vms14       > it consists of two primary programs: CTANGLE, which produces compilable C code from the source texts, and CWEAVE, which produces nicely-formatted printable documentation using TeX.
2023-05-14 09:45:25 <KipIngram   > It'll really just be founded on the ability to create links between arbitrary fields in disk blocks.
2023-05-14 09:45:32 <KipIngram   > Stored in a b-tree.
2023-05-14 09:46:13 <KipIngram   > That is, there won't actually be any physical link content in the blocks - a separate b-tree will just say (location, location) and all those pairs will be stored in a way I can quickly search.
2023-05-14 09:46:20 <vms14       > javadoc would be another example
2023-05-14 09:46:28 <KipIngram   > So wherever I am in my code, I can look that location up and see if any links connect to it.
2023-05-14 09:46:37 <vms14       > I kind of like javadoc, except for the stupid amount of files it generates
2023-05-14 09:46:51 <vms14       > but it's java after all, it has to be like that
2023-05-14 09:46:51 <KipIngram   > If one does, I'll be able to follow it.
2023-05-14 09:47:36 <vms14       > oh yeah you told me about that
2023-05-14 09:47:47 <vms14       > you wanted an editor that can show documentation
2023-05-14 09:48:01 <vms14       > it was a cool idea
2023-05-14 09:48:02 <KipIngram   > Yes, and also with an integrated profiler / debugger.
2023-05-14 09:49:21 <vms14       > I need to add some way to document stuff
2023-05-14 09:49:27 <vms14       > then use it to document my lang
2023-05-14 09:49:43 <vms14       > I only had a database which I liked
2023-05-14 09:50:19 <vms14       > I described the arguments a word takes, what returns, example usage, related words...
2023-05-14 09:50:35 <vms14       > that went to a database that the language will open at start
2023-05-14 09:50:51 <vms14       > 'word doc   would read the documentation of that word in the database
2023-05-14 09:50:57 <vms14       > it was cool
2023-05-14 09:51:46 <vms14       > I tend to forget the order of arguments in some words
2023-05-14 09:52:03 <vms14       > 'word doc  was nice to remember them
2023-05-14 09:52:49 <KipIngram   > You're slightly all over the map this morning.  :-)
2023-05-14 09:53:15 <vms14       > what do you mean?
2023-05-14 09:53:24 <KipIngram   > Lots of areas of interest on your mind.
2023-05-14 09:53:33 <vms14       > ah, it's always like that
2023-05-14 09:53:43 <vms14       > I enjoy learning
2023-05-14 09:53:44 <KipIngram   > Not a criticism.
2023-05-14 09:53:53 <KipIngram   > I know what you mean.
2023-05-14 09:54:32 <vms14       > I have a wild imagination, I'd say it's my best feature xD
2023-05-14 09:55:13 <vms14       > it just can't stop, but it helps me a lot to imagine some circumstances and the effects they could give
2023-05-14 09:55:35 <vms14       > I use it to make analisis xD
2023-05-14 09:56:46 <vms14       > analysis*
2023-05-14 10:09:15 <KipIngram   > So, now that I have builds> does>, vocabularies will go with
2023-05-14 10:09:34 <KipIngram   > : voc  builds> 0 , 0 , 0 , does>> ...something... ;
2023-05-14 10:09:39 <KipIngram   > Ooops.  does>
2023-05-14 10:10:09 <KipIngram   > ...something... will basically take the vocabularly and remove it from the doubly linked list search path, if it's in there already, and then add it to the beginning of that path.
2023-05-14 10:10:45 <KipIngram   > If the link fields are already zero, it's not in the path currently and will just get added.
2023-05-14 10:11:29 <KipIngram   > Then I'll need a couple of other words for other operations, like just dropping a vocabulary from the path, making the first voc on the path the "current" vocabulary, and so on.
2023-05-14 10:11:40 <KipIngram   > Won't be much code overall - just a few lines.
2023-05-14 10:12:24 <KipIngram   > I'll have to either go dig up or "re-remember" the code for managing a doubly linked list.  It can be done rather elegantly, but only if you get it written just right.
2023-05-14 10:24:25 <KipIngram   > vms14:; in typical forth the word "definitions" takes the first vocabulary in the search path and makes it "current" so it will receive definitions.
2023-05-14 10:24:30 <KipIngram   > I usually call it defs.
2023-05-14 10:24:53 <KipIngram   > So you'd say "<voc name> defs" and now new words go into that vocabulary.
2023-05-14 10:25:22 <KipIngram   > And you're also searching that vocabulary first, since <voc name> made it that way, but if you don't want that you can pop it off of the list.
2023-05-14 10:25:39 <KipIngram   > <voc name> defs vdrop
2023-05-14 10:25:40 <KipIngram   > or something.
2023-05-14 10:27:03 <vms14       > KipIngram: I think maybe you're right in that I have a lot of thoughts today
2023-05-14 10:27:04 <KipIngram   > I need to tidy up a little - a few days ago I decided I wanted to change my vocabulary layout from <word pointer> <forward link> <backward link> to <forward link> <backward link> <word pointer>.  I may unwind that change or I may keep it but adjust things in a different way from how I did it.  I kind of "cheated" a little to accommodate that change, and I don't much like it.
2023-05-14 10:27:24 <KipIngram   > It seems just a little more than usual, though as you say you're always a person of wide ranging interests.
2023-05-14 10:27:37 <KipIngram   > May just mean you're well-rested so have a lot of mental energy at the moment.
2023-05-14 10:28:06 <vms14       > that posix journey I had when I reinstalled netbsd while the router was dying and got me stuck with a system with only posix stuff and ctwm was kind of illuminating
2023-05-14 10:28:10 <vms14       > :D
2023-05-14 10:28:19 <KipIngram   > The reason I made that change was that I felt like if I always have links at the beginning of structures, it would be easier to use a universal set of list management words.
2023-05-14 10:28:30 <KipIngram   > But I'm questioning now whether that's really "necessary."
2023-05-14 10:28:46 <KipIngram   > Really as long as the links are always next to each other things ought to work out fine.
2023-05-14 10:31:17 <KipIngram   > I also think it's high time I write ."
2023-05-14 10:31:23 <KipIngram   > I've been stumbling along without it.
2023-05-14 10:31:48 <KipIngram   > Just using . and emit.
2023-05-14 10:32:00 <vms14       > I like emit
2023-05-14 10:32:14 <vms14       > it shown me what the real answer of everything is
2023-05-14 10:32:17 <vms14       > 42 emit
2023-05-14 10:32:25 <KipIngram   > :-)
2023-05-14 10:32:33 <KipIngram   > My key and emit will deal with utf8.
2023-05-14 10:32:52 <KipIngram   > The items they work with might be multi-byte, but all utf8 chars will still fit in a stack cell, so...
2023-05-14 10:32:56 <vms14       > I can't implement key :/
2023-05-14 10:33:07 <KipIngram   > It's tedious to get setup for that.
2023-05-14 10:33:09 <vms14       > well I can, but I usually don't
2023-05-14 10:33:15 <KipIngram   > You have to dick with the termios.
2023-05-14 10:33:37 <KipIngram   > And doing that in a way that gets things properly restored on exit is a bit of a pain.
2023-05-14 10:34:49 <KipIngram   > I got the vocabulary layout changed back to its original order.
2023-05-14 10:35:23 <KipIngram   > The list management words will just have to be given pointers to the forward link field.
2023-05-14 10:35:54 <KipIngram   > Or perhaps the offset of the forward link field in the structure - that seems more likely.
2023-05-14 10:36:09 <KipIngram   > That might vary from case to case.
2023-05-14 10:42:04 <unjust      > vms14: https://ibb.co/WBzdYkx this is what's become of trying to use <canvas> with RPN
2023-05-14 10:46:03 <vms14       > unjust: I had some basic canvas stuff time ago in a js implementation, but I guess it was lost
2023-05-14 10:46:44 <unjust      > it seems like canvas's API borrowed from postscript
2023-05-14 10:46:53 <vms14       > https://vms.neocities.org/xd
2023-05-14 10:46:56 <vms14       > lol
2023-05-14 10:47:00 <vms14       > it still works
2023-05-14 10:47:01 <vms14       > hahaha
2023-05-14 10:47:43 <vms14       > ohh it's cause it's the whole lang embedded in a <script> tag
2023-05-14 10:48:19 <unjust      > i can see the script type="bro"
2023-05-14 10:48:49 <unjust      > cool effect
2023-05-14 10:49:18 <unjust      > mine is @ https://www.risingedge.co.za/eczema/
2023-05-14 10:49:47 <KipIngram   > Apple has some neat documentation around their font processing that's interesting.
2023-05-14 10:50:01 <KipIngram   > Describes a very Forth-like vm for manipulating font data.
2023-05-14 10:50:13 <unjust      > do you have some links for that?
2023-05-14 10:50:39 <unjust      > or some keywords to search for?
2023-05-14 10:50:46 <vms14       > https://termbin.com/kb2u that's the code, which is also embedded on that xd.html
2023-05-14 10:51:47 <vms14       > the syntax is different from what I use now
2023-05-14 10:52:16 <KipIngram   > I'll see if I can find it, unjust.  It wasn't more than a few weeks ago I saw it - probably still in my browser history.
2023-05-14 10:52:29 <unjust      > thanks KipIngram, sounds like a good read
2023-05-14 10:53:21 <KipIngram   > https://developer.apple.com/fonts/TrueType-Reference-Manual/
2023-05-14 10:53:29 <vms14       > unjust: TypeError: this.cells.at is not a function
2023-05-14 10:53:38 <vms14       > on every thing I type
2023-05-14 10:53:40 <vms14       > 1
2023-05-14 10:53:46 <vms14       > or : oh 1 2 3 ;
2023-05-14 10:53:59 <unjust      > which browser are you using?
2023-05-14 10:54:02 <vms14       > seamonkey
2023-05-14 10:54:10 <unjust      > i've only tried it in firefox so far
2023-05-14 10:54:12 <vms14       > it fails on some "new" js stuff
2023-05-14 10:54:17 <unjust      > ah
2023-05-14 10:54:24 <unjust      > maybe it doesn't like the classes
2023-05-14 10:54:28 <KipIngram   > One nuisance of linked list management is dealing with the first item, which is generally pointed to by a variable.
2023-05-14 10:54:28 <vms14       > yeah don't worry, it's seamonkey's fault
2023-05-14 10:54:54 <vms14       > you should just focus on ff, chrome and safari
2023-05-14 10:55:22 <vms14       > I miss chrome :/
2023-05-14 10:55:35 <unjust      > could refactor it to use object = { "methodName": function(...) { ... } }; instead of class { method() { ... } }
2023-05-14 11:00:50 <vms14       > nah, don't worry about seamonkey
2023-05-14 11:01:04 <vms14       > it does not support a lot of js stuff
2023-05-14 11:01:19 <vms14       > use js as you think you should
2023-05-14 11:01:39 <vms14       > and only care about chrome safari and ff
2023-05-14 11:02:02 <vms14       > also I'd recommend you to look at pwa if you want to keep going with this lang
2023-05-14 11:02:13 <unjust      > what is pwa?
2023-05-14 11:02:20 <vms14       > it will allow to almost have native apps in a smartphone written on your lang
2023-05-14 11:02:25 <vms14       > progressive web applications
2023-05-14 11:02:43 <unjust      > i'll check that out
2023-05-14 11:02:55 <vms14       > they fake a native app, also linux, windows, and somehow mac allow to install them as applications on your computer
2023-05-14 11:03:15 <vms14       > in a smartphone for example you can push notifications
2023-05-14 11:03:21 <unjust      > i was happy with how it functioned on the phone in a browser already, but that's something i hadn't considered
2023-05-14 11:03:31 <vms14       > and they can be on google store, windows store, etc
2023-05-14 11:03:49 <unjust      > thanks for the tip
2023-05-14 11:03:54 <vms14       > unjust: the browser is a multimedia api
2023-05-14 11:03:57 <vms14       > a very huge one
2023-05-14 11:04:29 <unjust      > i'm learning that as i go along, i don't know enough GLSL yet to make use of the webgl canvas context yet but it's available at least
2023-05-14 11:04:36 <vms14       > https://developer.mozilla.org/en-US/docs/Web/API
2023-05-14 11:04:58 <unjust      > i noticed the <audio> tag recently and thought a sampler or basic mod tracker might be a good future project
2023-05-14 11:05:08 <vms14       > you have a whole sound api
2023-05-14 11:05:12 <vms14       > and a midi one
2023-05-14 11:05:29 <vms14       > and the sound api is huge as it was made with musicians in mind
2023-05-14 11:05:49 <vms14       > there's a lot of stuff
2023-05-14 11:06:28 <vms14       > https://web.dev/progressive-web-apps/
2023-05-14 11:06:42 <unjust      > have you made use of pwa yet for your own projects?
2023-05-14 11:08:19 <vms14       > no, cause I did not keep improving the js implementation
2023-05-14 11:08:50 <vms14       > but it's something I'd have to do sooner or later if I were
2023-05-14 11:09:15 <vms14       > it allows you to write applications for your phone
2023-05-14 11:09:23 <vms14       > a thing I always wanted secretly
2023-05-14 11:09:57 <vms14       > then you have web workers and the cache api, to provide online services
2023-05-14 11:10:09 <vms14       > offline*
2023-05-14 11:10:32 <vms14       > and the indexddb which crc is working with
2023-05-14 11:10:42 <vms14       > he is going to implement blocks using that
2023-05-14 11:11:33 <unjust      > what are the size constraints on indexdb usage?
2023-05-14 11:11:50 <vms14       > 60% of total device storage for domain
2023-05-14 11:11:56 <vms14       > althought depends on browser
2023-05-14 11:12:15 <thrig       > some folks put the offline storage on /tmp and nuke it a lot
2023-05-14 11:12:17 <vms14       > safari will keep asking the user ever x data
2023-05-14 11:12:21 <unjust      > so if i can serialize the state of my system into a json object, can i store it there and restore it later as long as no one clears the history/cache?
2023-05-14 11:12:51 <vms14       > yes
2023-05-14 11:13:01 <vms14       > you have also the cache api, which is easy to use
2023-05-14 11:13:16 <vms14       > intended mainly to be used inside a web worker
2023-05-14 11:13:24 <vms14       > a web worker is kind of a proxy for fetch
2023-05-14 11:13:48 <vms14       > when the browser makes a request it will call the web worker instead
2023-05-14 11:13:56 <vms14       > including js fetch
2023-05-14 11:14:13 <vms14       > and you can use fetch inside the web worker, or the cache api
2023-05-14 11:15:16 <vms14       > there's a lot of stuff to learn
2023-05-14 11:46:31 <KipIngram   > Zarutian_iPad: That 8+ I have near the end of does> is a header layout dependency.  I guess a lot of the code in the two words is - there is an assumption there that pfa fields and cfa fields are four bytes, and that pfa2 pfa1 cfa should be the first fields in the header.
2023-05-14 11:46:42 <KipIngram   > Probably  not any "lovely" way around that.
2023-05-14 11:47:05 <KipIngram   > create  obviously already has some of those dependencies in it.
2023-05-14 11:47:33 <KipIngram   > For the most part I have a small word set that navigates me around headers.  But it's not a  "perfect" sequestering of those details by any means.
2023-05-14 11:48:08 <KipIngram   > I do think my current layout made yesterday's definitions particularly clean.
2023-05-14 11:48:52 <KipIngram   > It was one of you guys - either you or MrMobius, I think - that suggested the current layout (or the core of it, at least - I tweaked the suggestion a little).
2023-05-14 11:49:14 <KipIngram   > I previously had some stuff before the name string, and some stuff after - ideas I'd just inherited from FIG years and years ago.
2023-05-14 11:49:41 <KipIngram   > That made it necessary to traverse the name string in both directions for various purposes.  It's eays to traverse it forward, but not as easy to traverse it backwards.
2023-05-14 11:49:53 <KipIngram   > The suggestion was that the name string should go last, and that was a very good idea.
2023-05-14 11:50:15 <KipIngram   > I then modified hte idea by moving the PFA fields to before the CFA instead of after.
2023-05-14 11:50:32 <KipIngram   > I did that because primitives don't have a CFA field, so having it at one end made it easy to make it optional.
2023-05-14 11:50:45 <KipIngram   > And THAT made it easy to add a second one for builds> does>.
2023-05-14 11:51:35 <KipIngram   > I'll use the same layout for the next one, except I will need to add an index field so that I know where the CFA and PFAs are in that table.
2023-05-14 11:52:22 <KipIngram   > For builds> does> in that system, that will mean allocating an additional index for the final words, because I need two pfas.  Unfortunately that will also allocate a second CFA, which isn't needed.
2023-05-14 11:52:54 <KipIngram   > It did occur to me, though, that I could save a list of those "wasted" CFAs - if I defined a primitive later I could use them for that purpose, since primitives need ONLY a CFA.
2023-05-14 11:59:39 <KipIngram   > Zarutian_iPad: What threw me onto the wrong track yesterday is that I forgot to consider the return stack altogether.  It was clear that I couldn't use the data stack to pipe this information along the pipeline.
2023-05-14 12:00:02 <KipIngram   > Didn't think of the return stack, and that left only the defining word definition itself as a "medium."
2023-05-14 12:00:59 <KipIngram   > So I actually figured a fair bit of the final code out myself - it was you pointing out that the very address I needed was sitting there on the return stack while does> executes that made it all click.
2023-05-14 12:01:04 <KipIngram   > I'd totally overlooked that.
2023-05-14 12:01:55 <KipIngram   > I'm not 100% sure how I got onto the "builds> and does> are immediate" path, but once I was on it it gave me an entirely different set of constraints.
2023-05-14 12:02:21 <KipIngram   > The origin of that may have been years and years ago, when I wrote an earlier system.
2023-05-14 12:15:12 <KipIngram   > I really, really love mosh.  :-)
2023-05-14 12:15:52 <KipIngram   > I installed it on the raspberry pi, and now I have a console window open into the pi that just stays there, across network issues, sleeping my notebook, etc.
2023-05-14 12:16:00 <KipIngram   > Behaves totally like a local console session.
2023-05-14 12:22:28 <KipIngram   > I'm not sure what went foul with the previous Pi - it just stopped working one day.  It may be fine - may have just been the wall wart.
2023-05-14 12:22:51 <KipIngram   > I should investigate - if it can be revived I could use it for hobby stuff now.
2023-05-14 12:23:21 <KipIngram   > The new one came with a little fan, so maybe that will help it have a better time of things.
2023-05-14 12:53:07 <KipIngram   > Looks like on Linux "getconf -a | grep -i cache" will give you information about your cache setup.
2023-05-14 14:20:19 <MrMobius    > KipIngram: could it be the SD card?
2023-05-14 14:36:13 <KipIngram   > Yeah, I guess that's possible too.  I've not done any troubleshooting at all.
2023-05-14 14:36:36 <KipIngram   > I've got a Pi imager installed on my notebook - I'll make a new card and check that out.
2023-05-14 20:38:23 <KipIngram   > Hmmm.  I see a potential problem in my system.  Not *likely* to be something that will bite me, but now that I've seen it it will bug me.
2023-05-14 20:39:12 <vms14       > a broken window
2023-05-14 20:39:25 <KipIngram   > My link fields have limited range.  Not "tiny," but in theory if I switched to a vocabulary and defined a butt load of words, and then switched back to Forth, I might not be able to reach over all the stuff I defined in the other vocabulary.
2023-05-14 20:39:30 <vms14       > it's a term from the art of programming book
2023-05-14 20:39:50 <KipIngram   > It's unlikely I'll ever define such a large vocabulary.
2023-05-14 20:39:59 <vms14       > it compares little bugs like that to a broken window in a home
2023-05-14 20:40:13 <KipIngram   > I don't know, I think a broken window sounds worse.
2023-05-14 20:40:21 <vms14       > it says this escalates fast into a much more
2023-05-14 20:40:25 <KipIngram   > That's an actual "right now" problem.
2023-05-14 20:40:37 <vms14       > a broken window in a house makes it feel like its an abandoned house
2023-05-14 20:40:51 <vms14       > a kid will come and add a grafitti on it
2023-05-14 20:41:06 <vms14       > and so on
2023-05-14 23:20:15 <Zarutian_iPad> 
ACTION can not sleep because some idiot left an idling anemic diesel engine running and only the bass of it carries

2023-05-14 23:22:03 <Zarutian_iPad> the annoying thing is that it is not a steady sound but sounds like an combustion engine that is being chocked off as it is going to shut off
2023-05-14 23:23:24 <Zarutian_iPad> but in this case it gets a bit of fuel and air to repeat this death rattle
2023-05-14 23:24:25 <thrig       > focus on every forth breath or something like that?
2023-05-14 23:31:56 <Zarutian_iPad> the most annoying part I can not locate the damn thing to turn it off
2023-05-14 23:38:40 <KipIngram   > We don't really seem to have much grafitti around here.
2023-05-14 23:38:51 <KipIngram   > I see it down in the area my office is in.