2024-08-23 23:23:33 <zelgomer    > : later  r> r> swap >r >r ;
2024-08-23 23:24:26 <zelgomer    > like most forth things, i can't decide whether this is clever or terrible
2024-08-23 23:42:48 <xentrac     > usually those are the same
2024-08-23 23:43:07 <xentrac     > it seems fairly clear though
2024-08-23 23:43:37 <xentrac     > it would probably be faster on an interpreted Forth as : later 2r> swap 2>r ;
2024-08-23 23:45:01 <xentrac     > it seems highly likely to cause debugging difficulties when you can't figure out why its caller's caller is returning into the middle of a callee
2024-08-23 23:45:50 <xentrac     > but it seems like a nice way to do Golang-style `defer` cleanup on return
2024-08-23 23:46:22 <xentrac     > or I guess Python-style resource managers
2024-08-23 23:46:43 <xentrac     > I'd probably call it something like `finally` or `eventually` if I dared to use it
2024-08-23 23:47:43 <xentrac     > I was going to object that if its caller's caller is in the middle of a loop, it will probably crash Forth, but calling `defer` in a loop is problematic in Golang too
2024-08-23 23:52:50 <veltas      > zelgomer doesn't like the 2-words
2024-08-23 23:53:00 <veltas      > I think it's clever
2024-08-23 23:53:56 <zelgomer    > yeah true. i actually don't like do-loop in forth and never implemented it, though. i'm using it for my meta-] so that you can activate the target compiler from a host word. otherwise, ] nests and doesn't return until the corresponding [, which isn't how you normally expect ] to work
2024-08-23 23:54:28 <amby        > i still need to write myself some better looping words
2024-08-23 23:55:04 <xentrac     > I see, that isn't at all the use I was thinking of
2024-08-23 23:55:28 <amby        > currently i only have DFOR which is a shorthand for BEGIN ?DUP WHILE 1-
2024-08-23 23:56:02 <xentrac     > I figured out how to write better looping words today, looking at Henry Baker's COMFY-65, the LISP 1.5 manual, McCarthy's original LISP paper, and Dijkstra's guarded command language
2024-08-23 23:56:26 <xentrac     > but now I've fallen into a pile of Pascal papers and can't get out
2024-08-23 23:56:48 <veltas      > C > Pascal
2024-08-23 23:56:51 <veltas      > fite me
2024-08-23 23:57:35 <xentrac     > well, specifically what I was trying to figure out is when people started using {} in grammars to indicate repeated elements, and Wirth's 01970 paper on Pascal is the earliest I've found so far
2024-08-23 23:58:19 <veltas      > C didn't use {} in its grammar and doesn't today, even
2024-08-23 23:58:29 <xentrac     > I thought it might be COBOL in 01960 or ALGOL in 01960 but neither of them use it.  The Algol-60 report doesn't even use [] for optional items
2024-08-23 23:58:40 <veltas      > Neither does C
2024-08-23 23:59:00 <xentrac     > that's true!  and I think that's partly due to the influence of yacc
2024-08-23 23:59:19 <veltas      > Must be, I never knew what BNF etc was until it came up in a compiler course
2024-08-23 23:59:34 <xentrac     > I didn't actually implement repeating items in the bootstrapping part of https://github.com/kragen/peg-bootstrap/blob/master/peg.md either