Burndown list for consolidating base lib about motoko-base HOT 13 CLOSED

Working on this today. I'll try to tackle it all, unless I hear otherwise.

from motoko-base.

@matthewhammer:

• In TrieMap renaming fromIter -> fromEntries for consistency with iter -> entries

Ah, this actually was on purpose. I thought that for uniformity, toX/fromX should always denote type conversions, where X names the resp type (I suppose that could go into the design guide as well). So either fromIter, or perhaps better in this case, some other name not using from (not sure which one, though).

from motoko-base.

Let's go with foreach then.

from motoko-base.

I think we can close this now.

from motoko-base.

Notes:

• Skipping the compiler primitives in the current draft PR (#44 ); anyone want to help there?
• In TrieMap renaming fromIter -> fromEntries for consistency with iter -> entries

from motoko-base.

Update #44 now addresses nearly all of the items above (not compiler prims).

from motoko-base.

Some folks expressed their dislike of changing map to transform, which is understandable. My main motivation was to avoid confusion and domain conflicts like with a map function on maps. But if you all think this is preferable, I'm happy to switch back.

The other naming that may be overly obscure is apply. In most cases, iter would be the more obvious and recognisable name, except that (1) it is unrelated to the type Iter, and (2) it is an odd name in the case of singleton containers like options (even though OCaml happily provides Option.iter).

Thoughts?

from motoko-base.

I find apply not great; it sounds as if I am applying the container to a value. forEach might be a more proletarian alternative. But I would be happy with iter or iterate.

Also in favor of reinstanting map, but of course I am heavily primed to understand map. transform sounds too generic – anything could be a transformation.

Maybe pick someone in the org who is close to our target audience and ask them? Andew maybe, being a front-end developer?

from motoko-base.

@nomeata, agreed. Anybody else has thoughts? Is there enough appetite for another (smaller) name change like this?

from motoko-base.

Alternatives for apply might be do or act or every, though I'm not sure we want to burn do as a keyword. Scala uses foreach, FWIW.

from motoko-base.

Or just each.

from motoko-base.

@nomeata. @rossberg, @kritzcreek wants to know what to use for apply

from motoko-base.

FWIW, I also prefer foreach to replace apply, to name side-effect-based iteration that accepts each item with the function arg.

@rossberg -- In fact, I do think it's directly related to the type Iter (foreach is the "computation form" of imperative iteration for a structure, rather than the "object/value form" for that given structure), but I do not see any benefit from reusing the name "iter" for this operation, since it may just be too confusing, given that we also want an operation that produces iterators, and we want to call that operation iter. Given this situation, it doesn't make sense to call them the same thing.

I'm happy with transform. I'm also happy with map, since I think people are used to concepts being manifested separately as both nouns (values) and verbs (computations) without those concepts being either identical nor confused. map is a good example of this.

from motoko-base.