User talk:Blashyrkh

Crazy J

Hi! Can you please put the CC expressions of the combinators in Crazy J? For example, T=JII instead of ``jii. I put a few of the simple examples in the table. Your notation can be hard to read for the more difficult expressions. Please put the CC expressions for the more complicated combinators. Thanks. Bobby Jacobs (talk) 12:58, 5 June 2026 (UTC)

I see you've done it yourself. You're welcome ;) --Blashyrkh (talk) 14:19, 5 June 2026 (UTC)

Between IJ and SK

Please put the regular combinator formulas of the cancellative combinators in the table in User:Blashyrkh/Between IJ and SK. For example, Xxy=y in lieu of 000010. They are easier to understand than the BLC syntax. Thanks. Bobby Jacobs (talk) 18:43, 19 August 2026 (UTC)

In fact, I abandoned that research because User:PkmnQ proved that there are actually bases more potent than IJ and still incomplete (e.g. IJ(YK)). Are you sure it's worth continuing? --Blashyrkh (talk) 19:00, 19 August 2026 (UTC)

I'm looking at it again and now I'm wondering what the smallest counterxample would be, and if there's one that doesn't require the usage of some Y or ω-like trick. –PkmnQ (talk) 03:31, 23 August 2026 (UTC)

w/o Y and ω-family you could just apply it to a bunch of linear combinators and put the variable in the place where the argument is ignored to get an affine combinator, so I conjecture there's nothing as a counterexample without ysing Y and ω.Cleverxia (talk) 04:13, 23 August 2026 (UTC)

I almost proved that any proper cancellative combinator is enough for IJX to be complete (stroked out text at User:Blashyrkh/Between_IJ_and_SK). But some arbitrarily taken non-proper ones were ok too. So, I wonder how you guys would define a "Y or ω-like trick". --Blashyrkh (talk) 07:10, 23 August 2026 (UTC)

I know that it has been solved, but I just wanted to see the combinators in a way that I could understand. Bobby Jacobs (talk) 01:54, 28 August 2026 (UTC)

iota

Maybe we should add iota to the combinatory logic tables. by the way, where is your brute forcer located at? Cleverxia (talk) 02:05, 23 August 2026 (UTC)

iota is there, at the last row of the table. I was thinking about adding Tromp's alpha and my beta and gamma there, but not sure, they are not very known. Which brute forcer do you mean, I made a lot of them? --Blashyrkh (talk) 07:10, 23 August 2026 (UTC)

well I meant the columns, not the rows. that is, list the shortest iota reperesentation for each of the combinators. For the bruteforcer, anyone should be OK Cleverxia (talk) 08:16, 23 August 2026 (UTC)

We discussed it already at Talk:Combinatory_logic#Table_of_common_well-studied_combinators_-_other_bases?. The table would look ugly with more columns, especially on mobile devices. Ok, we'd add iota, then we'd add SK (without I), S'KI (Bobby Jacobs's "converse starling", see Talk:Combinatory_logic#The_bird_S') and S'K, then "converse iota", then Tromp's Alpha, then my Beta and Gamma... And the table would become unreadable.

My bruteforcer for "between IJ and SK" problem is at github, the link is at the dedicated page. Here is it. --Blashyrkh (talk) 08:28, 23 August 2026 (UTC)

14's Quine

How did you make the quine? And the other language quines? --the nit is typing... 08:32, 20 September 2026 (UTC)

You certainly can read a 14 code, try to read it :) I intentionally added the # instruction to my interpreter for debug dump of the stack, it helps much.

The 14 quine consists of 3 sections:

• initialization - just pushes numbers into the stack
• the first non-destructive loop that prints those numbers as numbers (in exactly the same format as the "initialization" section)
• the second loop (which is allowed to be destructive) which prints numbers as characters (according to some number-to-character mapping).

All we need to do now is to make numbers to encode instructions of both loops.

My other quines work similarly. --Blashyrkh (talk) 09:02, 20 September 2026 (UTC)

Also take a look at Crazy J/QUINE. Despite the fact that the language is completely different, the page contains a detailed explanation which demonstrates the similar idea. --Blashyrkh (talk) 09:31, 20 September 2026 (UTC)

Quaternions

This might be a good golfing candidate. Cleverxia (talk) 05:25, 26 September 2026 (UTC)

I need a closer look at it, its computational capabilities are still unclear to me. Unfortunately, some essential (for chosen program forms) language features are implementation-defined, and there's no reference implementation yet. It makes golfing a bit vague. --Blashyrkh (talk) 05:40, 26 September 2026 (UTC)

fixed. the reference implementation outputs the real part as a character, and takes character input as a real number; also it represents the quaternion as [[1,i],[j,k]] internally. Cleverxia (talk) 07:11, 26 September 2026 (UTC)

Added to my table. But don't expect me to make a quine in half an hour :) --Blashyrkh (talk) 07:23, 26 September 2026 (UTC)

Moved the further discussion to Talk:Quaternions#Quine discussion

I'd rather prefer someone to join this golfing contest. 2644 bytes Quine in Lazy K - is it "great" or "poor" result? How can I be sure if no one else try to do it. At least I know that Kaveh Yousefi's Truth-machine in CQfuck is a masterpiece unreachable to my skills. --Blashyrkh (talk) 05:49, 26 September 2026 (UTC)

I will try doing. 2644 bytes should be quite good... Cleverxia (talk) 07:11, 26 September 2026 (UTC)

Put your results to your page (say, User:Cleverxia/Golf). When you beat me, I'll put easy-to-notice mark "Beaten by Cleverxia" into the corresponding cell of my table. --Blashyrkh (talk) 07:23, 26 September 2026 (UTC)

I'll have a go at some of the quines, I think in some sense quines are the "most free" kind of program out of the seven given (as in, it's the one where programs can vary the most). –PkmnQ (talk) 07:30, 26 September 2026 (UTC)

I agree. As for me, if the language is new, I don't start immediately with a quine (probably, Котята-арбузята were an exception). Truth-machine is good to play with IO, conditionals and infinite loops. Kolakoski sequence - lists and nested loops (hello, CQfuck, I hate you and I'll be back for you). If you have an idea for other golfing program forms, feel free to suggest. --Blashyrkh (talk) 07:39, 26 September 2026 (UTC)

I had put A+B program form to my table because of Котята-арбузята. The language has no arithmetics, and making A+B is a challenge by itself (without emphasis on golfing). --Blashyrkh (talk) 07:49, 26 September 2026 (UTC)

maybe counting up in decimals (or unary) should be a good challenge. counting in decimals is a big challenge for languages w/o number output. Cleverxia (talk) 08:09, 26 September 2026 (UTC)

Yeah, good idea. Added to my table. --Blashyrkh (talk) 08:29, 26 September 2026 (UTC)

Also the Meta cat program which is fairly difficult to implement effeciently (but easy to implement shortly), so this may need to have a different scoring criterion like scoring the output length on a set of cases like 0123456789, The quick brown fox jumps over the lazy dog. Hello, World! 3.14159263589793238!@#$%^&*()-_=+`~ and such. This makes the "+1" solutions unoptimal. Cleverxia (talk) 13:47, 30 September 2026 (UTC)

or maybe score the program on itself and take the output length as score. (for example current 1+ program scores 11246 bytes. output here) 13:54, 30 September 2026 (UTC)