2021-12-02 01:22:16 <veltas      > joe9: I am being a bit sarcastic when I say that, but I think either way is legit, I just want the one with the least coding
2021-12-02 01:22:32 <veltas      > Sometimes you do actually need to work with data in a file, or in blocks
2021-12-02 01:22:52 <veltas      > But today is not that day
2021-12-02 01:29:25 <veltas      > crc: Very nice demonstration of reorder there, although I don't 100% follow at first glance
2021-12-02 01:29:32 <veltas      > Anyway got to get ready for work
2021-12-02 01:41:50 <siraben     > not Forth but I too am getting on the AOC train https://github.com/siraben/haoc-2021/blob/master/day2.hs
2021-12-02 04:03:02 <veltas      > heresy!
2021-12-02 04:04:53 <veltas      > crc's solution is better because it doesn't store the input in memory
2021-12-02 06:19:54 <crc         > http://forth.works/share/53483323fdbc4d0d1d455a4a09224ef4 is my quick solution for day 2
2021-12-02 06:24:48 <f-a         > what does tokenise do?
2021-12-02 06:34:48 <crc         > s:tokenize splits a string on each occurrence of a character (removing the character) and constructs an array from the pieces
2021-12-02 06:37:43 <f-a         > useful
2021-12-02 06:47:57 <boru        > I was just saying in ##asm how day two seemed to lend itself to being solved with Forth quite nicely.
2021-12-02 06:48:16 <eris[m]     > yea
2021-12-02 06:50:06 <crc         > if my forth had parsing words, I'd probably just define forward, up, and down as words and then just included the input file
2021-12-02 06:54:04 <boru        > Yeah, that's what I was thinking.
2021-12-02 07:40:17 <veltas      > Haven't looked at it yet but looking forward to it based on what I'm hearing here
2021-12-02 07:40:32 <veltas      > Looks like the dumb 'paste the input in my source' approach is going to pay off
2021-12-02 08:33:25 <veltas      > Day 2 https://pastebin.com/raw/qQi9zyFt (done in lunch break)
2021-12-02 08:59:39 <dave0       > what is the correct word to use here instead of LIT ? (nop is the dummy word)     : nop ; : IF LIT ?branch , here @ LIT nop , ; immediate
2021-12-02 09:00:26 <dave0       > LIT happens to work but it's only cos i know the internals
2021-12-02 09:52:05 <crc         > here's an ans version of today's first half: http://forth.works/share/1e329b6655fab8ea2dd1aa484c314898
2021-12-02 10:00:58 <boru        > Stop being so productive!
2021-12-02 10:03:29 <crc         > no :)
2021-12-02 10:03:41 <boru        > Pft!
2021-12-02 10:47:16 <veltas      > crc: Very nice (Patrick Bateman face)
2021-12-02 10:47:41 <veltas      > Much better than mine, in my opinion
2021-12-02 10:55:39 <veltas      > Elizabeth Rather would approve
2021-12-02 12:05:14 <joe9        > crc, good one.
2021-12-02 14:24:41 <hodbogi     > Has anyone ever thought about using HAMT for a Forth dictionary?
2021-12-02 14:56:54 <imode       > hodbogi: fancy seein' you here.
2021-12-02 14:59:13 <imode       > I don't think the HAMT would apply because a forth dictionary isn't just a lookup table, it's also what you'd call the heap, where you can allocate spans of memory.
2021-12-02 16:51:02 <veltas      > "Are you okay, veltas? You're sweating."
2021-12-02 17:47:03 <veltas      > I'm using a linked list for my dictionaries ;)
2021-12-02 17:50:49 <f-a         > seems sensible
2021-12-02 17:50:56 <f-a         > are there other ways to implement forth dicts?
2021-12-02 17:57:47 <hodbogi     > imode, get out
2021-12-02 17:57:48 <hodbogi     > :P
2021-12-02 17:57:55 <imode       > never.
2021-12-02 17:58:01 <hodbogi     > heh
2021-12-02 17:58:57 <hodbogi     > I thought maybe a HAMT would be kinda nice if you have a lot of them. there's been a couple different dictionary layouts I've thought of in the past
2021-12-02 17:59:45 <hodbogi     > imode, Wanna help me finish my operating system?
2021-12-02 18:02:00 <imode       > only if you help me finish mine.
2021-12-02 18:05:30 <veltas      > Linked lists were fast enough in the 70s, they're fast enough now
2021-12-02 18:06:13 <veltas      > f-a: Lots of forths use hash tables for faster compiling, like gforth and mark4's x4
2021-12-02 18:07:39 <veltas      > I don't know about HAMT specifically
2021-12-02 18:11:40 <dave0       > i thought about it and a stack for dictionary would work too
2021-12-02 18:12:41 <dave0       > when you create a new word, push it onto the stack.. when you want to look up a word, search from most-recently-pushed to least-recently-pushed
2021-12-02 18:13:42 <dave0       > then everything is a stack! data stack! return stack! dictionary stack!
2021-12-02 18:34:55 <f-a         > thanks
2021-12-02 18:39:52 <eris[m]     > but how do you allocate..?
2021-12-02 18:40:08 <eris[m]     > the dict is already a stack allocator
2021-12-02 18:40:20 <eris[m]     > word defs r just a linked list inside of that
2021-12-02 21:13:33 <tabemann    > I use a simple linked list because hash tables are hard to work with when compiling incrementally to flash
2021-12-02 21:19:55 <dave0       > tabemann: my msdos forth is coming along... it has 15 linked lists, one for each word length
2021-12-02 21:20:13 <dave0       > but i made it an array, so it only costs 15 cells
2021-12-02 21:21:13 <tabemann    > I have two linked lists, one in flash and one in RAM
2021-12-02 21:21:26 <dave0       > ah
2021-12-02 21:22:33 <tabemann    > however, on the RP2040 I on bootup compile an array of dictionary item hashes and pointers to dictionary items in RAM based on the flash dictionary
2021-12-02 21:22:46 <dave0       > if you put all the 1 char words on list1 and all the 2 char words on list2 and 3 char words on list3 etc... you don't have to separately store the length of each word
2021-12-02 21:22:46 <tabemann    > because compiling gets very slow without it
2021-12-02 21:23:03 <dave0       > in a way i hashed by length
2021-12-02 21:23:32 <tabemann    > to me the issue is that when I boot I need to discover the flash dictionary's end
2021-12-02 21:23:59 <tabemann    > if I made separate linked lists for, say, every wordlist, I wouldn't know where to discover each of their ends
2021-12-02 21:24:12 <dave0       > ah
2021-12-02 21:24:52 <dave0       > yeah with 1 list you only need to follow the links to the last one
2021-12-02 21:26:51 <tabemann    > I go the opposite direction
2021-12-02 21:27:26 <tabemann    > I scan from high in flash down to where the $FFFFFFFF ends and where the value $DEADBEEF can be found
2021-12-02 21:28:30 <dave0       > oh is that where you will create the next word?
2021-12-02 21:29:40 <tabemann    > words in each dictionary are in a linked list from the last word to the first word