[TODO] add items(table) to allow using algorithms on a table

[timotheecour]

without this, we can't use algorithms on tables and are forced to use explicit for loops

[skilchen]

isn't this a perfect opportunity to use a template?

template items[K,V](t: Table[K, V]): tuple[key: K, val: V] =
     pairs(t)

[timotheecour]

template items*[K,V](t: Table[K, V]): tuple[key: K, val: V] =
  ## EDIT: some doc comment (without this, no error)
  pairs(t)

when isMainModule:
  let a = {10:"foo", 12:"bar"}.toTable
  for ai in a:
    echo ai

Error: ignored invalid for loop

EDIT seem above line marked EDIT

[skilchen]

Thats not true for me. It only happens, if you add additional code to the items template.

import tables

template items*[K,V](t: Table[K, V]): tuple[key: K, val: V] =
  pairs(t)

when isMainModule:
  let a = {10:"foo", 12:"bar"}.toTable
  for ai in a:
    echo ai

==>

(key: 10, val: "foo")
(key: 12, val: "bar")

[timotheecour]

It only happens, if you add additional code to the items template.

indeed, it's hitting this bug #9230 which I just filed.
I can do another PR to change it to your suggested way after #9230 gets fixed; but otherwise it' no good because we can't doc-comment items

[Araq]

Seems silly to me.

without this, we can't use algorithms on tables and are forced to use explicit for loops

YES! Exactly! A for loop is one single construct to learn that supports ALL of these epicycles of functional programming with its never ending map, fold, foldr, sum, groupby...

[krux02]

I don't disagree on an items on a table. But I doubt it should be an alias to pairs. items normally yields the values and pairs keys and values. Since it is unclear what the best option for items is I prefer to not provide an items. Then local implementations may vary without conflics in the name.