Talk:Find words which contain the most consonants

From Rosetta Code

but each consonant should appear only once in a word[edit]

Currently, the task has a requirement that consonants not be repeated.

This has been implemented in two different ways. (Not repeated during the counting process vs. not repeated at all.) --Rdm (talk) 12:48, 25 July 2022 (UTC)

That does not make any sense to me. How does a word such as "crystallography" not have repeated r,y,l? --Pete Lomax (talk) 21:52, 26 July 2022 (UTC)
A couple of ambiguities I think. "the most consonants" – are we listing only words which contain the largest observed set of unique consonants ? The maximum observed cardinality of such a set is 9. How then, does this description lead us to listing words with consonant sets of cardinality 8 ? A bit unclear, and submissions diverge.
Part of the problem, I think, is that over-familiarity with the language and style of task descriptions in the rest of the Calmosoft task corpus tends to undermine confidence about the intention of "should appear only once in a word". That phrase is the only indication that "most consonants" does not mean "maximum string length after vowel deletion". Are we sure that it was intended to carry double weight ?
Largest set of consonants (no repetitions contribute to the cardinality of the set) + additionally no words containing any consonant repetitions are of interest, regardless of the size of their set of consonants ? Probably, but it could do with a bit of tidying, to resolve uncertainty about whether "in a word" refers to the set of consonants used in that word, or to the list of characters in that word.
The Phix output prefix captures it well – "Most consonant words" could label a number of quite different sets. Hout (talk) 03:50, 27 July 2022 (UTC)
If the preferred argument is "Don't do as Calmosoft says – do as Calmosoft does", then perhaps the task description should include a phrase like "(see Ring entry)" ?.
(A little burdensome – and also costs redundant server load – to expect everyone to figure out who the author of an unclear description is, and which submission(s) is/are theirs) Hout (talk) 04:07, 27 July 2022 (UTC)

No need to delete versions on Rosetta Code[edit]

I've restored a Python version which had been deleted on the grounds that it made a particular interpretation (in fact two complementary interpretations) of the ambiguous task description.

There's no need to delete versions on Rosetta Code – the first step is to comment on the Discussion page, and the general pattern is to add and preserve, perhaps after discussion, with updates. Hout (talk) 17:14, 26 July 2022 (UTC)

Often tasks are disambiguated by referring to the authors original submission making the disgarding of words with repeated constants correct.--Nigel Galloway (talk) 18:15, 26 July 2022 (UTC)
Indeed, and that's why the final output of that variant excludes words with repeated consonants.
I find the preceding (unfiltered) listing interesting though, and I'm inclined to leave it, while the task formulation remains ambiguous, as a reminder that a bit of reformulation still seems desirable. Hout (talk) 19:57, 26 July 2022 (UTC)