adjoin seems too similar to join ...

That's kind of why I liked it.

... which is unrelated.

join intersperses adjacent elements of a list with something.

This (in the simple 2 at a time variant) is often used to add adjacent elements of a list together -- ad-join.

A vote for overlaps (BTW, wouldn't "overlaps" be enough?).

Um. I thought about that, and still haven't reached a conclusion which is better.

In the singular, it sort of implies something only happens once.

But if you read 'overlaps' as a plural as in "there are many overlaps"; rather than the (*)???: "it overlaps", then it makes more sense I think.

Not sure what grammatical term applies here?

The nice thing about overlap(s) is it works for the variable width (depth?) of overlap also which adjoin doesn't.

I think I'm beginning to favour overlap(s) also.

But now I'm also thinking about overmap and mapover?

---

With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'

Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.

"Science is about questioning the status quo. Questioning authority".

In the absence of evidence, opinion is indistinguishable from prejudice.

> (BTW, wouldn't "overlaps" be enough?).

Well, mentioning the plural brings a new perspective...

For me designing and naming a good API is far more complicated than coding, since it involves human psychology.

Especially since sometimes good names are already used or even only reserved.

This brings a new concern, wasting an otherwise better used word.

overlaps could mean much more things and especially in plural it could indicate tuples bigger than just a pair of values.

The English grammar has no dualis alongside singular and plural like for instance Arabic, so only constructs like pair or couple indicate 2 tuples, and couples are normally disjoint. Maybe a construct with prefix "bi-" could solve this (think bigamie ;-)

Other than describing the tuple as a whole one could try describing the relation between two elements.

So my best guess ATM is "neighbor" but in singular(!). The neighbor can't mean anything else it describes a relation between 2 objects.

Words like next or link already have different meaning and bichain sounds artificial in my ears.¹

Otherwise I would consider successor or one of the synonyms like follower or one of it's synonyms like companion.

As soon as the definition is clear seeking through a thesaurus is of help.

Cheers Rolf

( addicted to the Perl Programming Language)

updates

¹) bimap ?

One reasonable metric for rating them might be: Imagine you're seeing the function name for the first time, together with a definition of its signature (parameter/return types). Would you have guessed its correct meaning? How sure would you have been?

Of course that's still subjective, so let me present my intuitive reactions so that you can compare them to your own:

Simple version of the function: Map a list to all its subsequences of 2 adjacent items

For the "pairwise" version, the function would only take a code block and a list. Given that function signature, I think the only ones for which I would have guessed the correct meaning and would have been reasonably sure about it, are:

RESULT_LIST = map_adjacent { CODE } LIST
[download]

RESULT_LIST = neighbors { CODE } LIST
[download]

A few comments on some of the others:

• RESULT_LIST = adjoin { CODE } LIST
  [download]

  Doesn't sound like it operates on adjoining items, but rather like it causes things to adjoin (in some unspecified way).

• RESULT_LIST = overlaps { CODE } LIST
  [download]

  Sounds like it does something less generic and more complicated, although I wouldn't be sure what. Maybe LIST accepts a list of ranges rather then numbers, and the function finds (and processes) overlapping ranges? Or something.

• RESULT_LIST = mapAdj { CODE } LIST
  [download]

  Map "adjacent"? Map "adjusted"? Map "adjunct"? ...

• RESULT_LIST = between { CODE } LIST
  RESULT_LIST = between_pairs { CODE } LIST
  [download]

  I might have guessed this one correctly, but would have felt unsure about it.

• RESULT_LIST = pipeline { CODE } LIST
  [download]

  Sounds like the return value of CODE becomes the first input parameter for the next iteration, like in List::Util's reduce. At least that would match my understanding of pipelines in computing.

• RESULT_LIST = map_right_neighbour { CODE } LIST
  [download]

  Sounds like it does one iteration strictly for each element of LIST, with $b set to the element's right neighbor, and on the last iteration, $b set to undef.

Generalized version of the function: Map a list to all its subsequences of n adjacent items

In this case, the function would likely take a code block, a number, and a list. Unfortunately, I don't think I would be very sure about any of my guesses in this scenario, for any of the suggested names. One that hasn't been suggested yet (probably because it's too long), but would be pretty self-explanatory for me, is:

RESULT_LIST = natatime_sliding { CODE } NUMBER, LIST
[download]

Those where there's at least a decent chance that I would have guessed the correct meaning, are, in descending order:

RESULT_LIST = map_adjacent { CODE } NUMBER, LIST
[download]

RESULT_LIST = moving { CODE } NUMBER, LIST
[download]

RESULT_LIST = slide { CODE } NUMBER, LIST
[download]

moving is nice because it's effectively a generalization of "moving average" to "moving <custom calculation>"...

@moving_averages = moving { sum(@_)/@_ } $n, @data
[download]

...but of course if the context in which you're using it has nothing to do with statistics, that recognition value will be lost.

between and map_right_neighbour wouldn't work at all anymore for n > 2.
adjoin, overlaps, mapAdj and pipeline would still have the same problems as above.

Thanks for your thoughtful review.

natatime_sliding

Sounds like OAPs complaining that their tea'n'chat session are getting later and later .... :) But yes, it's too long.

mapAdj

Map adjacent; map adjoining; map adjunct; all of those work.

Since salva pointed out the scala version, I keep looking again at slide/sliding; but they make me think of sliding scales & sliding checksum, neither of which are appropriate here.

As I said elsewhere, many function names (and keywords) do not mean much unless you know what they mean in the context of a computer program.

Take reduce or join or grep or splice. Write the functionality of any of those out in full:

take_a_list_of_values_convert_them_to_their_string_representations_if_
+necessary_intersperse_them_with_this_other_string_and_return_a_single
+_string_formed_from_them_all( "\t", @nums);
[download]

And it would be ridiculous.

You only know what grep does, because you learnt it, not from the name, so the key things are to be vaguely mnemonic and memorable, and preferably short.

I'm more and more persuaded by overlap & overmap.

---

With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'

Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.

"Science is about questioning the status quo. Questioning authority".

In the absence of evidence, opinion is indistinguishable from prejudice.

why no map_slice or map_splice? cause like slice/splice it returns a list ... and you're slicing the list you're iterating over