I'm working on modernizing Rosetta Code's infrastructure. Starting with communications. Please accept this time-limited open invite to RC's Slack.. --Michael Mol (talk) 20:59, 30 May 2020 (UTC)

Talk:Pancake numbers

From Rosetta Code

no examples[edit]

Guilty as charged for introducing the cheeky "gap" algorithm, but the task says "and for each show an example". Only five (F#, Go, Julia, Phix, and Wren) do so, the other 11 (ten of which were added after that clause appeared) do not and should really be marked as incorrect or incomplete. --Pete Lomax (talk) 00:52, 18 January 2021 (UTC)

Yet another no-example added, so I've marked now 12 entries as incomplete. The trivial gap algorithm does not help in comparing different programming languages, which is the primary purpose of this site, whereas something a bit more substantial should. --Pete Lomax (talk) 01:24, 20 January 2021 (UTC)

Everything is wrong[edit]

Every solution on this page was branched from an initial Phix algorithm by User:Petelomax. The page states that the algorithm was written by Pete, and that it was “was freshly made up”. It also states that it “It agrees with https://oeis.org/A058986/b058986.txt”. The issue is that it doesn’t. That lists p(19) as 22, whereas the algorithm yields 21.

Even if it’s right, I question the point of such providing such a solution. It’s mostly reverse-engineered guess-work from a solution generated by actual exhaustive search. Could we purge the page to only keeping solutions that *actually* find the solutions through exhaustive search? Monarchdodra (talk) 12:57, 9 March 2021 (UTC)

As per https://oeis.org/history?seq=A058986 the sequence was changed Dec 5th 2020. I've added a modified version. If everyone agrees and someone marks my entry along with everyone else's as incorrect, I'll replace the original.
The suggestion there is no point in a fast estimate method is frankly outrageously ridiculous. Sure it needs a caveat (as I tried to put on the Phix entry), maybe in this case it should go in the task description itself. Besides, since the whole point of RC is to compare languages, a task that allows comparison of both fast and exhaustive approaches has more merit than just exhaustive. --Pete Lomax (talk) 16:38, 9 March 2021 (UTC)
I’m fine with a “fast estimate” approximation if it’s somewhat clearly marked as such. As it stands, you could write a pretty interesting translation DAG from this page. It actually took me a while to go C++ -> C -> Go -> Phix. It also took me a while to realize that the algorithm was just that. It took reading through the Phix description to realize that the Julia solution (translated from Go), was actually not a formal solution.
I’d be fine if the description allowed a “2 problems” kind of question, where it asks for “a fast approximation”, and then “an exhaustive search” as extra credit. This would also encourage more creative algorithms that for finding an approximation.
As as it stands, it feels to me that every solution here is just parroting your original approximation, without really understanding the how or why.
PS: Thanks for this history page: That’s what I had supposed had happened, but I thought the change would have been listed in the “EXTENSIONS” part Monarchdodra (talk) 18:42, 9 March 2021 (UTC)