Support for `:list`

[helins]

I realized that there is currently no support for :list, unlike other collections (:vector, :set). Since this is pretty basic, I was wondering if it was somehow a deliberate choice. If not, a PR looks fairly straightforward to write.

[ikitommi]

I was dropped at some point of not being very useful - :sequential is a better alternative for most cases as it covers both lists and sequences and because of that, transformations don't have to coerce the mapped results back into a list.

(list? '())
(list? (conj '() 1))
(list? (reverse '()))
(list? (list (seq '())))
; => true

(list? (map identity '()))
(list? (cons 1 '()))
(list? (seq '()))
(list? (concat '() '()))
; => false

Do you find explicit list useful?

here's the commit: https://github.com/metosin/malli/commit/0acde2bc021e581aeb8ae341ec44b2d12e1200cd

[helins]

Granted, this has probably never caused much problems because lists are seldom used in the vast majority of projets. They are commonly used in metaprogramming but not much elsewhere. I was more surprised because of the fact that it is a core data structure rather than by unfulfilled obvious use cases.

The only reason I noticed it is, once again, due to generation. :sequential generates a vector, so do all the seqexp schemas but :cat (or when wrapped in :cat) which generates a lazy seq. So a :list or rather a :seq of ints passing both validation and generation must be written like this:

[:and
 [:cat [:* :int]]
 seq?]

 ;; Note, does not allow :ref

Or

[:and
 [:sequential
  {:gen/fmap #(or (list* %)
                            '())}
  :int]
 seq?]

Which looks a bit over-engineered for a simple list. Unless I am missing something?

[ikitommi]

Registering a custom -collection-schema would be the way to do that.

Sadly, the current impl of m/-collection-schema is not the best for this (one would also have to install new methods on various multimethods like malli.generator/-schema-generator etc.

But then again, we can make m/-collcection-schema more versatile: https://github.com/metosin/malli/pull/433.

[helins]

So it looks that there is now everything needed for having :list?

What would a good PR be like, anything besides those points? Adding:

• -collection-schema into core base-schemas
• :list method in malli.generator/-schema-generator
• :list method in malli.json-schema/accept that mimicks :vector
• :list method in malli.clj-kondo/accept that simply returns :list

Note on example in #433: predicate should be seq? instead of list? since seqs and lists and interchangeable in pratice. A bit suprisingly, list* produces a seq, not a list.

I would argue that it is a good idea having :list since it's a core data structure and it will make it easier using Malli for describing code-related things (macros, fns operating over code, ...).

[helins]

(and it is also consistent with having support for list?)