2022-03-01 14:01:19 <remexre     > if i have quotations, and already removed loops (in favor of combinators to express common looping patterns), is there any enormously-strong reason to not do the same for conditionals?
2022-03-01 14:01:23 <remexre     > or is this haskellbrain slipping in
2022-03-01 14:05:20 <remexre     > eg instead of IF foo ELSE bar THEN, write { foo } { bar } COND
2022-03-01 14:05:35 <remexre     > DUP IF foo ELSE DROP bar THEN  =>  { foo } { bar } ?COND
2022-03-01 14:06:01 <remexre     > IF foo THEN  =>  { foo } WHEN
2022-03-01 14:06:22 <remexre     > DUP IF foo ELSE DROP THEN  =>  { foo } ?WHEN
2022-03-01 17:38:27 <KipIngram   > remexre: Does your Forth have the ?DUP word?
2022-03-01 17:39:02 <KipIngram   > Oh, never mind - that Forth was what you replaced.  I see now.
2022-03-01 17:40:02 <KipIngram   > I've thought about different styles of conditional handling.  These days my habit is to factor the conditional stuff into its own word and put a conditional return at the beginning of the word.
2022-03-01 17:40:35 <eris[m]     > what's that lok like?
2022-03-01 17:40:37 <KipIngram   > It inverts the condition - if I want to execute on true, then I return on false.
2022-03-01 17:40:42 <KipIngram   > It might something like
2022-03-01 17:40:56 <KipIngram   > : foo  ?0; ...stuff to do if true... ;
2022-03-01 17:41:05 <KipIngram   > Actually my word is 0=;
2022-03-01 17:41:19 <KipIngram   > : foo  0=;  ...stuff to do if true... ;
2022-03-01 17:41:37 <KipIngram   > I might also put a ? in the name of foo, to indicate it makes a decision.
2022-03-01 17:41:42 <KipIngram   > But that's just a style choice.
2022-03-01 17:41:54 <eris[m]     > ahhh
2022-03-01 17:41:55 <eris[m]     > makes sense
2022-03-01 17:42:12 <KipIngram   > I feel like my definitions are noticeably shorter with those in use.
2022-03-01 17:42:35 <KipIngram   > I also have conditional two-level returns, and I sort of build "vertical control structures" with all that.
2022-03-01 17:42:42 <eris[m]     > : 0=? r> drop ;
2022-03-01 17:42:43 <eris[m]     > ?
2022-03-01 17:43:05 <KipIngram   > I've started to evolve standard multi-level "patterns" that I use over and over.
2022-03-01 17:43:11 <KipIngram   > Yes, basically, except mine's a primitive.
2022-03-01 17:43:16 <eris[m]     > that's common i think
2022-03-01 17:43:16 <eris[m]     > yea
2022-03-01 17:43:44 <eris[m]     > i have something similar in that i have implicit conditional loops in the form of a word called +;; and -;;
2022-03-01 17:43:45 <KipIngram   > I have a lot of them, for a lot of conditions.  I wrote a python script to generate the assembly source for those, into an include file for nasm.
2022-03-01 17:44:15 <KipIngram   > :-)
2022-03-01 17:44:20 <KipIngram   > I like terse symbolic names.
2022-03-01 17:44:31 <KipIngram   > Deliver me from THIS_IS_MY_WORD_TO_DO_FOO
2022-03-01 17:44:47 <eris[m]     > yup
2022-03-01 17:44:58 <eris[m]     > +;; is basically just recurse if false
2022-03-01 17:45:30 <KipIngram   > Ah, nice.  I have a word called ME that jumps to the most recently defined word, so it's how I implement tail recursion.
2022-03-01 17:45:42 <KipIngram   > The advantage of having a special name is that I can then have "conditional ME" words.
2022-03-01 17:45:47 <eris[m]     > ye, mine just calculates it manually
2022-03-01 17:45:58 <eris[m]     > i don't use a forth impl i wrote
2022-03-01 17:46:02 <eris[m]     > so it's just using LATEST
2022-03-01 17:46:47 <KipIngram   > Some of my progrmamer friends complain about my word names - say they're unreadable.  But... they're readable to me, because I've learned them.
2022-03-01 17:46:53 <KipIngram   > That's who it's for anyway.
2022-03-01 17:47:26 <KipIngram   > I invented a "convention" under which I prefix existing word names with .
2022-03-01 17:47:38 <KipIngram   > Those are new words, separate dictionary entries.  Nothing automated.
2022-03-01 17:47:54 <KipIngram   > But what they always do is the same thing the base word does, except they remove one less item from the stack than the base word.
2022-03-01 17:48:20 <KipIngram   > So if =; is a return if equal that removes the two elements, .=; is a return if equal word that keeps the lower element.
2022-03-01 17:48:23 <eris[m]     > huh, neat
2022-03-01 17:48:30 <KipIngram   > Very nice for applying a sequence of comparisons to a data variable.
2022-03-01 17:48:36 <eris[m]     > uxn (a forth like project) has something like that
2022-03-01 17:48:42 <KipIngram   > You get rid of the comparison item, but keep the actual datum.
2022-03-01 17:49:03 <eris[m]     > you suffix a k to do that
2022-03-01 17:49:17 <KipIngram   > Was handy in my NUMBER word, where I had to compare the character to a variety of different things.
2022-03-01 17:51:09 <KipIngram   > Putting that all together (the single and double conditional returns and the . prefix), I was able to get my next string char on the stack and then say something like ?0-9 ?A-F ?a-f .? -?
2022-03-01 17:51:31 <KipIngram   > Each word checked the char, if it matched did what it was supposed to do and then DOUBLE returned, else just returned.
2022-03-01 17:51:39 <KipIngram   > Single return sent me to the next test - double return sent me out.
2022-03-01 17:51:55 <eris[m]     > that's neat!
2022-03-01 17:52:09 <KipIngram   > That's one of those patterns that sort of emerged.
2022-03-01 18:24:10 <joe9        > remexre KipIngram, eris[m]  , can you please paste some code using those ( COND +;; )  words? I would like to see how the forth code looks with them.
2022-03-01 18:24:50 <eris[m]     > i don't have a big body
2022-03-01 18:24:54 <eris[m]     > i had a funky little lists example
2022-03-01 18:25:58 <eris[m]     > : >end cdr dup cdr +;;
2022-03-01 18:26:04 <eris[m]     > cdr being cell+ @
2022-03-01 18:26:23 <eris[m]     > which gets the second value from a pair implemented as a pointer
2022-03-01 18:27:12 <eris[m]     > i dont really preserve a lot f code i work with
2022-03-01 18:27:56 <eris[m]     >  : length >r cdr dup cdr r> 1+ swap +;;
2022-03-01 18:28:00 <eris[m]     > loop body
2022-03-01 18:28:06 <eris[m]     > : length 0 length nip 1+ ;
2022-03-01 18:28:09 <eris[m]     > use of loop
2022-03-01 19:04:00 <remexre     > KipIngram: no, the optimizer I sketched out on paper used to have trouble with variable stack effects so I trained myself to always put a DROP in the ELSE
2022-03-01 19:04:36 <remexre     > (I have since fixed the optimizer's design, though it's not implemented yet)
2022-03-01 19:05:04 <remexre     > but yeah the ?COND and ?WHEN are basically ?DUP COND and ?DUP WHEN, respectively
2022-03-01 19:12:36 <remexre     > joe9: can send a proper example when I'm home, but the rewrites I wrote above are "pretty much it"
2022-03-01 20:34:40 <crc         > in my forth, COND is `choose` and WHEN is `if`; I don't have combinators for ?COND or ?WHEN. An equivalent to ?WHEN might be worth adding based on a quick review of my sources.