2023-05-18 00:02:09 <vms14       > " oh my cat is nice" [ fields line .s ] wk
2023-05-18 00:02:38 <vms14       > fields => ['oh','my','cat','is','nice']  line => 'oh my cat is nice'
2023-05-18 00:03:07 <vms14       > now a different word to specify a separator
2023-05-18 00:03:54 <vms14       > it works by lines, it could end being useful
2023-05-18 00:04:04 <vms14       > awk is a very nice tool
2023-05-18 00:04:38 <vms14       > not sure if binding $1 $2 $3...
2023-05-18 00:04:51 <vms14       > I'll create a temp environment for it
2023-05-18 00:19:14 <vms14       > " oh my cat is nice" [ $2 . cr ] wk   => cat
2023-05-18 00:19:25 <vms14       > that's cool
2023-05-18 03:03:54 <veltas      > When vms14 gets back, have a look at the FOR command on Windows as well, it's another "mini awk"
2023-05-18 03:04:20 <veltas      > I think something like that or your miniawk idea would be really useful for spreadsheet style and simple database work
2023-05-18 03:04:25 <veltas      > And configuration files
2023-05-18 03:04:35 <veltas      > And maybe IPC
2023-05-18 04:59:03 <veltas      > FUZxxl: By the way, it turns out for FreeBSD to work we have to read argc from 8(%rsp), and argv thereafter
2023-05-18 04:59:20 <veltas      > And it doesn't save the same registers Linux saves for syscalls
2023-05-18 04:59:51 <veltas      > The latter is fine, because SysV spec only claims Linux support for its syscall appendix, but the former is a violation of the SysV ABI
2023-05-18 05:00:17 <veltas      > FreeBSD seems to actually call its ABI "freebsd:x86:64" or something, but they don't seem to document it
2023-05-18 05:00:28 <veltas      > It's all documented by source
2023-05-18 05:05:33 <FUZxxl      > hm interesting
2023-05-18 05:05:46 <FUZxxl      > which registers does FreeBSD save?
2023-05-18 06:02:23 <crc         > Looks like rdi, rsi, rip, rsp
2023-05-18 06:02:42 <crc         > It doesn't preserve r8, r9, r10, or rdx
2023-05-18 06:03:46 <crc         > I've not dug further into this yet, though am planning to do so once I'm able get gdb working
2023-05-18 06:04:53 <FUZxxl      > interesting
2023-05-18 07:19:59 <KipIngram   > Documentation on all of this is rather unreliable in all cases.  Just this weekend I ran into a case where an online resource clearly called out the list of registers that syscalls did not preserve "in Linux," but I found that if I saved only those registers my system crashed on the syscall.  I.e., I saved those, used the "sys" instruction, and restored them - crash.
2023-05-18 07:20:11 <KipIngram   > Sorry, the "syscall" instruction.
2023-05-18 07:20:26 <KipIngram   > But if instead I used a macro I wrote called "sys" that saves most everything, it worked fine.
2023-05-18 07:20:45 <KipIngram   > I didn't chase down the details, but SOMETHING beyond that list given in the website needed to be saved.
2023-05-18 07:21:23 <KipIngram   > I suppose that the system the person who wrote that was using did work that way, but there does appear to be system-to-system variation.
2023-05-18 07:22:39 <KipIngram   > That was when I was working up that iocotl syscall to get screen dimensions.
2023-05-18 07:26:26 <KipIngram   > This is interesting.  If I use the defuse online assembler, in 64-bit mode, and assemble "call qword [r12]", it assembles to
2023-05-18 07:26:43 <KipIngram   > 0:  41 ff 54 24 08          call   QWORD PTR [r12+0x8]
2023-05-18 07:26:52 <KipIngram   > where the heck does that +0x8 come from?
2023-05-18 07:28:23 <KipIngram   > If I change the 0x08 to 0x00 and dissassemble, then it disassembles to what I started with.  Looks like a bug in the site, to me.
2023-05-18 07:29:39 <KipIngram   > If I specify [rsi-8], then THAT assembles to just [rsi]
2023-05-18 07:31:20 <veltas      > Don't know that assembler
2023-05-18 07:31:39 <veltas      > Maybe qword is the constant 8, and it sees 8[r12] like gas sees 8(%r12)
2023-05-18 07:31:52 <KipIngram   > https://defuse.ca/online-x86-assembler.htm#disassembly
2023-05-18 07:31:57 <veltas      > You might need qword ptr
2023-05-18 07:32:28 <KipIngram   > Interesting.  Yeah, qword ptr seems to fix it.
2023-05-18 07:32:38 <veltas      > qword ptr is Intel syntax
2023-05-18 07:32:41 <veltas      > qword is nasm
2023-05-18 07:32:52 <KipIngram   > Aha.
2023-05-18 07:33:05 <KipIngram   > Ok - well, that explains why I wrote it that way to start with.
2023-05-18 07:33:47 <veltas      > I don't know what gas expects in Intel syntax mode
2023-05-18 07:40:45 <KipIngram   > It's still interesting that it just genned that 8 up out of nothing, though.
2023-05-18 07:41:50 <veltas      > As I said, I assume qword is the constant 8
2023-05-18 07:42:17 <veltas      > So you can do e.g. "sub rsp, qword"
2023-05-18 07:43:17 <KipIngram   > Oh, now I see what you mean.
2023-05-18 07:43:51 <KipIngram   > Yes, that seems to be what it's doing.
2023-05-18 07:43:53 <KipIngram   > Hah.
2023-05-18 07:43:58 <KipIngram   > Fair enough.
2023-05-18 07:48:15 <veltas      > You might be able to do call [r12]
2023-05-18 07:48:35 <veltas      > As it's sort of implied you want a qword by default
2023-05-18 07:48:56 <KipIngram   > Yeah, that works.
2023-05-18 07:49:08 <KipIngram   > I'm just tinkering to get ready to do that call into a new image.
2023-05-18 07:49:37 <KipIngram   > But now that I think about it I'll probably arrange that so that I just say call rrTOS - have the address to call in my top of stack reg.
2023-05-18 07:50:30 <KipIngram   > Push all my regs, call rrTOS, pop all the regs and then drop TOS.
2023-05-18 07:51:01 <KipIngram   > I can test that by just sticking a return instruction somewhere.
2023-05-18 07:53:13 <KipIngram   > So the idea is to metacompile to block buffers using "offsets."  Then have a word that will "load" the image from those, which will involve allocating buffers to put them in and converting the offsets to addresses appropriately, and then call into that.
2023-05-18 07:55:12 <KipIngram   > Guess I'm defining my own "executable format" here.
2023-05-18 08:03:35 <veltas      > Yeah I too have many ideas
2023-05-18 08:03:46 <veltas      > And then applying them is another thing lol
2023-05-18 08:04:30 <KipIngram   > :-)
2023-05-18 08:04:34 <KipIngram   > Same.
2023-05-18 08:04:59 <KipIngram   > I think I'm actually going to at least start on this one, though.
2023-05-18 08:05:28 <KipIngram   > Getting that editor to print just a range of lines was really the last thing I needed to "get ready."
2023-05-18 08:06:24 <KipIngram   > Well, I may add a copy/paste buffer to it first.
2023-05-18 08:07:37 <KipIngram   > Which will be more like a copy/paste stack.
2023-05-18 08:07:44 <KipIngram   > Little string stack.
2023-05-18 08:29:56 <veltas      > Well the current thing I'm doing is writing a forth word that dumps a series of blocks to a file, by outputting the starting forth editor commands necessary to load the blocks
2023-05-18 08:30:47 <KipIngram   > Ah - that's interesting.  Sort of "serializing" it as an edit stream.
2023-05-18 08:30:49 <veltas      > So it converts a block to something like "100 SCR !" "0 CHR !" "P <first line contents>" "U <second line contents>" etc
2023-05-18 08:31:07 <veltas      > And then further ones it can use N to get the next block
2023-05-18 08:31:17 <veltas      > Yeah basically
2023-05-18 08:31:33 <veltas      > And the point of this is so I can work with blocks, but keep this equivalent form in git
2023-05-18 08:32:15 <veltas      > So if someone wants to, they can work on the file in git, and at the very least see diff's nicely in git
2023-05-18 08:32:24 <veltas      > But I can work in blocks
2023-05-18 08:33:09 <KipIngram   > On one of my prior systems I wrote a couple of small utilities to take blocks.dat into a text file and vice versa.  Just so I could do editing with vim if I needed to.
2023-05-18 08:33:31 <KipIngram   > Those took care of the 64-byte line structure in that system.
2023-05-18 08:34:58 <KipIngram   > In this one, since it uses vim compatible line feeds, if I vim blocks.dat, there's block 1 right there.  I can't actually edit it, though, because that would move everything around, but it wouldn't be that hard to write something that "re-4096'd" everything.
2023-05-18 08:35:46 <veltas      > I'm not sure how to handle writing ^, other than print a dummy character and then do CHAR ^ xxx BUFFER yyy + C! on the next line for each occurrence
2023-05-18 08:36:08 <veltas      > Because in starting forth editor ^ is the optional command delimiter
2023-05-18 08:41:53 <veltas      > Or maybe U<space><space> followed by S" <linecontents>" xxx BUFFER yyy + SWAP MOVE
2023-05-18 08:44:14 <veltas      > Or add a word to insert ^ at CHR, like ^^, and do U <pre>^^^ I <post>
2023-05-18 08:44:45 <KipIngram   > I'm thinking right now about how to store these lines on this string stack.
2023-05-18 08:44:59 <KipIngram   > I'm considering bounding each line's data on both ends with a count byte.
2023-05-18 08:45:10 <KipIngram   > Just so I can traverse items without scanning in either direction.
2023-05-18 08:45:30 <KipIngram   > So <n><n bytes><n>
2023-05-18 08:46:12 <KipIngram   > I'm looking at using a particular disk buffer for that, with the first 16 bytes marking the first unused offset.
2023-05-18 08:47:03 <veltas      > Use 0 as delimiter
2023-05-18 08:47:14 <veltas      > Can scan both directions, less memory usage
2023-05-18 08:47:23 <KipIngram   > That doesn't help me get to the other end of the string.
2023-05-18 08:47:35 <veltas      > Put a hidden 0 at start
2023-05-18 08:47:40 <KipIngram   > In my source blocks the only 0 is at the end of the block's data.
2023-05-18 08:47:49 <veltas      > Okay
2023-05-18 08:47:59 <veltas      > Does it need to be a source block?
2023-05-18 08:48:02 <KipIngram   > Yes, I am thinking about that - to know when there are no more items on the string stack.
2023-05-18 08:48:17 <veltas      > Surely it's not a source block?
2023-05-18 08:48:20 <KipIngram   > No, but I'm just thinking about where the items that will go onto this come from.
2023-05-18 08:48:37 <KipIngram   > Generally speaking they will be a series of printable chars with a 0x0A at the end.
2023-05-18 08:48:48 <KipIngram   > That's what one line in a source buffer looks like.
2023-05-18 08:49:39 <KipIngram   > So push will add the currently targeted editor line to the stack; pop will open a space at the target location and move the item back in.
2023-05-18 08:52:43 <KipIngram   > Some scanning is unavoidable, but I'm trying to not add "more" - trying to use the words I've already written that scan to take care of things.
2023-05-18 08:53:41 <KipIngram   > ++ scans the target point forward one line.  So it's easy enough to bound the current line by making a temporary copy of part of the state and ++-ing it.
2023-05-18 08:53:49 <KipIngram   > Then I've got pointers to the current line and the next line.
2023-05-18 08:56:07 <KipIngram   > You know, maybe the way to do this is to make the string stack look just like an editable block.  Then I could use the words I've already written on it.
2023-05-18 08:56:17 <KipIngram   > That's probably the first avenue to consider.
2023-05-18 08:56:55 <KipIngram   > Write a -- partner for ++ that scans BACK one line.  Then I have what I need to bracket the latest line "pushed."
2023-05-18 09:01:21 <KipIngram   > Oh wow.  The -- word is quite easy.
2023-05-18 09:01:30 <KipIngram   > .: out  2dup =;; ;
2023-05-18 09:01:32 <KipIngram   > .: (--)  1- .c@ 10 !=me 1+ ;
2023-05-18 09:01:34 <KipIngram   > : --  out 1- (--) ;
2023-05-18 09:02:26 <KipIngram   > out just checks to see if I'm at the start of the block already.
2023-05-18 09:02:38 <KipIngram   > Kicks me out with no changes if I am.
2023-05-18 09:03:16 <KipIngram   > Oh, wait, that's not quite right.
2023-05-18 09:03:27 <KipIngram   > Won't stop properly if -- takes me to the first line.
2023-05-18 09:04:03 <KipIngram   > But I think I just need to drop a call to out into (--)
2023-05-18 09:04:44 <KipIngram   > Yeah, at the very beginning.
2023-05-18 09:06:50 <KipIngram   > That works perfectly.
2023-05-18 09:07:08 <KipIngram   > ++ checks for the null byte at the other end and won't go past it.
2023-05-18 10:11:07 <KipIngram   > Ok, adding -- raised my byte count on the editor source to 822.
2023-05-18 10:11:28 <KipIngram   > I'm hoping to get this string stack integrated and still be under 1kB.  :-)
2023-05-18 10:12:05 <KipIngram   > That leaves me around 5-6 ish lines of code - don't know if that will work out or not.
2023-05-18 10:42:48 <KipIngram   > You  know, this work on the editor is "selling" me more than ever on my stack frames.
2023-05-18 10:43:20 <KipIngram   > They're just wildly useful.  And that little timing exercise vms14 and I did last night - that showed their benefits strongly as well.
2023-05-18 10:44:36 <thrig       > it's gone too far if it's offering to throw in a salad spinner, free!
2023-05-18 10:44:51 <KipIngram   > Haven't had that offer come my way yet - I'll keep you posted.
2023-05-18 10:45:02 <KipIngram   > Those commercials of that sort are a hoot.
2023-05-18 10:45:22 <KipIngram   > One I remember from decades ago was "Gunsu knives."
2023-05-18 10:45:26 <KipIngram   > Ginsu
2023-05-18 10:45:38 <KipIngram   > Or maybe it was Ginzu - can't remember for sure.
2023-05-18 11:17:19 <Zarutian_iPad> Ginzu
2023-05-18 11:17:52 <Zarutian_iPad> thats how I have seen it written
2023-05-18 11:20:23 <KipIngram   > Looks a bit more  "Asian" that way.
2023-05-18 11:21:25 <Zarutian_iPad> eh, it is how it is pronounced but I do not have the most detailed/precise IPA for it.
2023-05-18 11:43:46 <KipIngram   > It's just been decades since I actually heard it pronounced.
2023-05-18 12:30:21 <KipIngram   > In case anyone's interested, to swap a pair of "2 cell records" on the stack one can do this:
2023-05-18 12:30:32 <KipIngram   > : 2swap  >r -rot r> -rot ;
2023-05-18 12:32:36 <MrMobius    > gross
2023-05-18 12:32:38 <MrMobius    > :P
2023-05-18 12:32:44 <KipIngram   > Yeah.
2023-05-18 12:32:54 <KipIngram   > Definitely not something I use with any frequency.
2023-05-18 12:36:16 <drakonis    > https://ratfactor.com/forth/the_programming_language_that_writes_itself.html this is a good post that keeps getting longer and longer
2023-05-18 13:13:42 <veltas      > KipIngram: I must confess
2023-05-18 13:13:58 <KipIngram   > NEVER CONFESS!!!!!
2023-05-18 13:14:04 <veltas      > I don't understand your ed sauce you posted yesterday, but it's okay, I'm okay not knowing
2023-05-18 13:14:22 <veltas      > I don't know your Forth dialect well and there's no comments or stack signatures
2023-05-18 13:14:23 <KipIngram   > It just takes a little getting used to.  I may be crazy.
2023-05-18 13:14:32 <thrig       > reduce ed in butter and onions until ...
2023-05-18 13:14:46 <KipIngram   > Yeah, I know.  I have plans for how to incorporate comments long term.
2023-05-18 13:15:13 <KipIngram   > I want to put some of them "elsewhere" and have the ones that are local such that I can turn on / off displaying them.
2023-05-18 13:15:21 <KipIngram   > I hate the way they make my source look.
2023-05-18 13:15:36 <KipIngram   > I do plan to solve that problem, though.
2023-05-18 13:15:44 <KipIngram   > My current practice is poor.
2023-05-18 13:16:12 <veltas      > I'm not sure what the right approach is, but sparse and decisive comments, with basic stack comments, seem to be necessary for me to have a reasonable shot
2023-05-18 13:16:20 <veltas      > Or really dead simple code with familar vocab
2023-05-18 13:16:24 <veltas      > familiar*
2023-05-18 13:16:48 <KipIngram   > I will say, though, that I've set this all aside for extended periods and have found that I can still follow it fairly well when I come back to it, so "it seems to work for me."
2023-05-18 13:17:02 <veltas      > Honestly also I don't have a lot of energy right now, been sick the last couple days (I don't feel sick though, just tired)
2023-05-18 13:17:08 <KipIngram   > Hopefully that will stay true until I get the point I can add commentary and documentation.
2023-05-18 13:17:20 <veltas      > You obviously have strong conventions etc but I don't know them
2023-05-18 13:17:23 <KipIngram   > Oh, I'm sorry man.  Those times suck.
2023-05-18 13:17:30 <KipIngram   > Right.
2023-05-18 13:17:43 <veltas      > It doesn't suck at all actually, but I'm just really tired and have a slight headache
2023-05-18 13:17:51 <veltas      > I'm not sure what it is, minor food poisoning maybe
2023-05-18 13:18:09 <KipIngram   > That is what "saves me," I think - in terms of my own comprehension.  I've established particular practices, and I remember THOSE.  So when I see certain code idioms, it says something to me that it wouldn't be saying to you.
2023-05-18 13:19:01 <veltas      > Yeah idioms are useful
2023-05-18 13:19:10 <veltas      > Forth idioms are harder
2023-05-18 13:22:16 <KipIngram   > Ok, I've got a draft of the code to add yy and pp.
2023-05-18 13:22:20 <KipIngram   > Yank and paste.
2023-05-18 13:22:49 <KipIngram   > Almost certain to be something wrong with it, but I think it's in the ballpark - I think "something near this" will work.
2023-05-18 13:23:06 <KipIngram   > 11 more lines of code.
2023-05-18 13:23:31 <KipIngram   > Probably will not stay under 1kB, though.  I haven't counted but it looks suspiciously large.
2023-05-18 13:24:19 <KipIngram   > It uses a specified block as the string stack and treats it exactly like source blocks, so I can leverage words already written heavily.
2023-05-18 13:25:05 <vms14       > string stack?
2023-05-18 13:25:11 <vms14       > :0
2023-05-18 13:25:36 <KipIngram   > Yes - in my editor I'll be able to say yy to pull out the current line and save it on a string stack.
2023-05-18 13:25:49 <vms14       > oh, like a clipboard
2023-05-18 13:25:50 <KipIngram   > Then I can move the target position and say pp and it will get pasted in at the new location.
2023-05-18 13:25:54 <KipIngram   > Yes.
2023-05-18 13:26:01 <KipIngram   > Except it will hold more than one item.
2023-05-18 13:26:16 <KipIngram   > Probably could have made it smaller if I'd just had one buffer.
2023-05-18 13:26:34 <vms14       > I don't know if I have to change something in my lang that will make me rewrite it from scratch again
2023-05-18 13:26:36 <KipIngram   > But this is how I did it in an earlier implementation, and I liked how it works.
2023-05-18 13:26:49 <vms14       > I think the next change will be start generating code
2023-05-18 13:27:12 <veltas      > String stack is powerful
2023-05-18 13:27:38 <veltas      > emacs and readline have it (and vim but the controls suck in comparison)
2023-05-18 13:27:54 <vms14       > as long I don't find any other flaw I can start adding words and making programs with it
2023-05-18 13:28:26 <vms14       > I always wanted to make a window manager
2023-05-18 13:28:39 <vms14       > I need to add ffi to C for that
2023-05-18 13:34:33 <KipIngram   > You know, using the >r / r@ / r> completely screws ability to factor.
2023-05-18 13:34:53 <KipIngram   > Ah, you know, I could use an address register instead.
2023-05-18 13:34:58 <KipIngram   > I'm only saving one thing.
2023-05-18 13:35:08 <KipIngram   > And that would be more efficient anyway.
2023-05-18 13:54:34 <Zarutian_iPad> I did not know that Redis was originally written in Tcl. One thing that puzzles me is the hate Tcl gets from some peeps.
2023-05-18 13:54:53 <vms14       > Zarutian_iPad: what hate?
2023-05-18 13:54:59 <vms14       > I always see everyone loving tcl
2023-05-18 13:55:31 <vms14       > everyone talks me wonders about tcl so much that I went to see how tcl is xD
2023-05-18 13:56:20 <Zarutian_iPad> “the syntax is not intutive” yeah because you only learned algol68 syntax derived programming languages and not one based on spefic parsing rules
2023-05-18 13:56:38 <vms14       > how it is that forthwrights have no relation with postscript?
2023-05-18 13:56:45 <vms14       > I've never seen one talking about ps
2023-05-18 13:57:10 <veltas      > I think many here are aware of it but not too interested
2023-05-18 13:57:14 <Zarutian_iPad> ps is okay ish but damn the gfx model is topnotch!
2023-05-18 13:58:25 <vms14       > I guess tcl's ability to treat code as data is what people seem to love so much
2023-05-18 14:00:08 <Zarutian_iPad> I put Tcl, Scheme, Lua, and eForth together in category of neat languages to know
2023-05-18 14:00:33 <vms14       > Zarutian_iPad: I had to learn tex to create a pdf, but it turns that pdf is a subset of ps and that tex ends writing that
2023-05-18 14:00:44 <Zarutian_iPad> (javascript is in there too but more of a Lua Scheme hybrid)
2023-05-18 14:00:57 <vms14       > which means there's nothing ps can't do that tex can
2023-05-18 14:01:21 <vms14       > but the whole thing is different, tex is a markup language where you specify the layout of a document, etc
2023-05-18 14:01:22 <Zarutian_iPad> pdf is just interpreted ps iirc
2023-05-18 14:01:40 <vms14       > in ps you draw and the pdf is the result of that drawing
2023-05-18 14:02:06 <vms14       > I need to create a tool to document stuff and I have to decide which one to choose for a pdf
2023-05-18 14:02:23 <vms14       > tex might be easier to automate, but ps more fun
2023-05-18 14:02:42 <Zarutian_iPad> a bit like QuickDraw PICT format as you draw to such output and instead of putting a pixmap on screen it saves the drawcalls in a file.
2023-05-18 14:04:06 <vms14       > I'll start with some html generator
2023-05-18 14:04:17 <vms14       > which I have to implement again xD
2023-05-18 14:10:40 <KipIngram   > Well, that raises the source byte count for the editor to 1230 bytes.  No way I'll be able to optimize that down to under a kB.  :-(
2023-05-18 14:10:42 <KipIngram   > Oh well.
2023-05-18 14:11:18 <thrig       > bloat
2023-05-18 14:11:42 <KipIngram   > Heh heh.
2023-05-18 14:12:07 <KipIngram   > Well, that was me just being greedy.  I still feel pretty good about it overall.  Once I get this last part actually working, that is.
2023-05-18 14:46:53 <vms14       > [ oh text oh... #oh .meh :color red :background blue ] html 'oh.html to.file
2023-05-18 14:47:16 <vms14       > <oh style="background:blue;color:red;" class="meh" id="oh">oh...</oh>
2023-05-18 14:48:53 <vms14       > I turn everything into a list and it's done
2023-05-18 14:49:01 <vms14       > :D
2023-05-18 14:49:12 <vms14       > also it's recursive
2023-05-18 14:52:15 <vms14       > [ p [ span :color blue [ p text oh :color red ] ] .meh  ] html .cr
2023-05-18 14:52:15 <vms14       > <p class="meh"><span style="color:blue;"><p style="color:red;">oh</p></span></p>
2023-05-18 15:23:24 <unjust      > vms14: https://staff.science.uva.nl/a.j.p.heck/Courses/Mastercourse2005/tutorial.pdf
2023-05-18 15:23:51 <Riviera     > it's a bit of a shame
2023-05-18 15:23:53 <unjust      > look at that those words in the first example, you can see where html5 canvas borrows heavily from
2023-05-18 15:24:00 <Riviera     > there are so many great docs about PostScript out there
2023-05-18 15:24:12 <Riviera     > i kind of find it sad that it's not really a thign anymore only for these books alone :)
2023-05-18 15:24:34 <unjust      > tons of printers still run postscript interpreters, you can access them if you try hard enough
2023-05-18 15:24:52 <gordonjcp   > you could just run ghostscript
2023-05-18 15:24:54 <Riviera     > yeah well
2023-05-18 15:25:00 <unjust      > i'm surprised by the lack of interest in display postscript
2023-05-18 15:25:03 <gordonjcp   > somewhere I have a Mandelbrot set generator in postscript
2023-05-18 15:25:04 <Riviera     > i don't mean that it's not possible anymore
2023-05-18 15:25:11 <gordonjcp   > unjust: VHS vs. Betamax
2023-05-18 15:25:12 <Riviera     > of course i can do these things
2023-05-18 15:25:27 <gordonjcp   > unjust: X was good, DPS was good
2023-05-18 15:25:37 <Riviera     > i always wanted to see DPS
2023-05-18 15:25:49 <Riviera     > like, would you have scalable vector screenshots?
2023-05-18 15:25:52 <gordonjcp   > ultimately "network effects" tipped it towards X, at least partly because it was easier to license
2023-05-18 15:25:53 <Riviera     > or how should i imagine that?
2023-05-18 15:26:46 <unjust      > depends on whether the graphics rendered were defined using absolute or relative units i guess
2023-05-18 15:27:13 <Riviera     > fair enough, but typically, say text and window decorations? or you don't know?
2023-05-18 15:27:31 <unjust      > i don't know, but i'd be interested in finding out
2023-05-18 15:27:47 <Riviera     > yeah me too :)
2023-05-18 15:31:45 <unjust      > that question hit home a little. i've been messing around with doing A4 page layout (for my invoices, because i'm tired of word processors) in an html canvas lately using a forth-like lang in javascript - and yesterday i had implemented a FONT word that would scale my sizing based on the DPI I had set up for my PAGE. i'm trying to keep all my sizes in MM.
2023-05-18 15:31:52 <unjust      > 150  VALUE DPI
2023-05-18 15:31:53 <unjust      > 25.4 VALUE MM/INCH
2023-05-18 15:32:02 <unjust      > : MM ( millimetres -- pixels ) DPI * MM/INCH / CEILING ;
2023-05-18 15:35:33 <unjust      > : A4  210 MM   297 MM ;
2023-05-18 15:35:39 <unjust      > A4 CANVAS VALUE canvas
2023-05-18 15:36:05 <unjust      > ( CANVAS sets create a "canvas" element with the supplied width and height
2023-05-18 15:37:39 <vms14       > unjust: so you're kind of implementing ps
2023-05-18 15:37:43 <Riviera     > unjust: would be curious to see an example :)
2023-05-18 15:37:46 <Riviera     > but
2023-05-18 15:38:04 <Riviera     > pizza delivery just came, so enough enough with postfix, i need my pizza fix
2023-05-18 15:38:13 <Riviera     > k, pretty lame joke, laters :)
2023-05-18 15:39:00 <vms14       > you draw on the canvas and then get an image from the canvas context?
2023-05-18 15:39:16 <unjust      > vms14: yes
2023-05-18 15:39:21 <vms14       > it's cool
2023-05-18 15:39:36 <vms14       > I had some words to automatically put a canvas on max height and with
2023-05-18 15:39:37 <unjust      > Riviera: i'll post a screenshot in a bit
2023-05-18 15:39:57 <vms14       > and the drawing functions used a "global" ctx
2023-05-18 15:40:20 <vms14       > it's fine if most of your programs will use only one canvas primarily
2023-05-18 15:40:37 <vms14       > you can set the current ctx from any canvas at any moment
2023-05-18 15:41:23 <vms14       > also a hook on the canvas on .onresize that executed a 'resize' word if existed in the dictionary
2023-05-18 15:41:46 <vms14       > you could make it a bit like lua
2023-05-18 15:41:58 <vms14       > and use requestAnimationFrame
2023-05-18 15:42:07 <vms14       > lua/ love2d
2023-05-18 15:42:35 <vms14       > love2d is actually very cool and easy to work with
2023-05-18 15:43:00 <vms14       > it assumes functions and calls them for events, drawing, update, etc
2023-05-18 15:43:11 <vms14       > you could add hooks manually if you wanted
2023-05-18 15:46:13 <unjust      > i wrote the words that wrap the canvas methods to take the context reference off the stack before the method arguments
2023-05-18 15:54:22 <unjust      > hopefully this clears up the ginsu spelling confusion: https://imgur.com/a/Vh08fK9
2023-05-18 16:45:14 <KipIngram   > Oh, it is s.
2023-05-18 16:45:29 <KipIngram   > Hah!   Should have just trusted my first guess.
2023-05-18 16:45:57 <thrig       > A Boy Named Ginsu
2023-05-18 16:52:27 <unjust      > those knives are the only as-seen-on-tv/informercial products i've had that actually lasted a decent period of time
2023-05-18 16:52:53 <KipIngram   > Yeah, once in a while one of those things might actually be good.
2023-05-18 17:29:57 <MrMobius    > unjust: that's neat. I've been using SVG since you can scale to arbitrary DPI and it works pretty well.
2023-05-18 17:41:59 <veltas      > Whenever I want to talk to vms14 they're out of the channel
2023-05-18 17:46:50 <veltas      > He who controls the ps, controls the universe
2023-05-18 17:55:41 <unjust      > MrMobius: what do you with the output? generate tool paths?
2023-05-18 17:56:02 <unjust      > s/you/you do/
2023-05-18 18:09:23 <KipIngram   > How bizarre.
2023-05-18 18:09:53 <KipIngram   > I put in primitives for loading and recovering the address registers (rsi and rdi).
2023-05-18 18:10:20 <KipIngram   > They weren't keeping their values at first - I discovered that my TYPE primitive was using those registers but wasn't saving and restoring them.
2023-05-18 18:10:37 <KipIngram   > Ok, fine.  I put in push si at the beginning and pop si at the end - A then worked fine.
2023-05-18 18:10:54 <KipIngram   > rsi, that is.
2023-05-18 18:11:03 <KipIngram   > Put in push / pop rdi, and that made B work.
2023-05-18 18:11:06 <KipIngram   > BUT...
2023-05-18 18:11:33 <KipIngram   > When I added the rdi push/pop, suddenly my system works fine, but when I say bye and return to the os my typing doesn't show up.
2023-05-18 18:11:41 <KipIngram   > I have to stty sane to get it back.
2023-05-18 18:12:08 <KipIngram   > This is repeatable - if I remove the push/pop rdi, everything is fine when I exit.  If I put it in, it's not.
2023-05-18 18:12:14 <KipIngram   > That makes NO SENSE AT ALL.
2023-05-18 18:15:09 <KipIngram   > Ah, actually it does.
2023-05-18 18:15:25 <KipIngram   > I was failing to set rdi to 1 in my exit ioctl call to restore the termios.
2023-05-18 18:15:36 <KipIngram   > It's the syscall parameter that specifies the target device.
2023-05-18 18:15:47 <KipIngram   > I just added a mov rdi, 1 there and now it's fine.
2023-05-18 18:15:48 <KipIngram   > Wow.
2023-05-18 18:15:56 <thrig       > some shells (zsh) do a better job at putting the terminal back the way it was. others, not so much
2023-05-18 18:27:48 <MrMobius    > unjust: working on different ways to make a keypad for electronics projects
2023-05-18 18:28:13 <MrMobius    > so various forms or printing and 3D printing
2023-05-18 18:28:33 <unjust      > numeric pads specifically or custom layouts?
2023-05-18 18:30:11 <unjust      > i'm curious about input devices that aren't rigid flat keyboards, but i haven't tried any chorded typing methods yet
2023-05-18 18:47:31 <KipIngram   > Hey, it's working.
2023-05-18 19:39:14 <MrMobius    > unjust: custom but flat keypad
2023-05-18 19:40:07 <MrMobius    > unjust: https://ibb.co/MPJNQF6
2023-05-18 19:54:42 <unjust      > very cool, what material will those buttons be made of?
2023-05-18 19:57:04 <KipIngram   > No math functions?
2023-05-18 19:57:25 <KipIngram   > (i.e., sin, cos, exp, ln, etc.)
2023-05-18 19:58:03 <KipIngram   > Maybe it's not for a calculator.
2023-05-18 20:23:18 <MrMobius    > unjust: I've tried a lot of different stuff. PLA with printed plastic card material glued on, PLA with dyed acrylic, dyed PETG, something called HJF, 3 color PLA
2023-05-18 20:23:36 <MrMobius    > ya Forth-based calculator so youll have to type sin and cos in manually
2023-05-18 20:30:56 <unjust      > which filament gave the best result so far?
2023-05-18 21:35:20 <MrMobius    > PETG is harder to print with but takes the dye pretty well
2023-05-18 21:35:36 <MrMobius    > PLA is easy but melts before you can get the dye to infuse