Add .validate method for refined companions #382

howyp · 2017-12-20T20:10:51Z

Now we have 'companion' objects for refined types, I thought it would be nice to have a neat way to refine to cat's Validated type.

It always uses NonEmptyList on the error side at the moment. I guessed that this is what people will use 90% time but if we think there will also be need for using List or even List Refined Not[Empty] then I we could support that too?

codecov · 2017-12-20T20:35:20Z

Codecov Report

Merging #382 into master will increase coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master    #382      +/-   ##
=========================================
+ Coverage   96.58%   96.6%   +0.01%     
=========================================
  Files          41      42       +1     
  Lines         498     500       +2     
  Branches       11      10       -1     
=========================================
+ Hits          481     483       +2     
  Misses         17      17

Impacted Files	Coverage Δ
...ain/scala/eu/timepit/refined/cats/validation.scala	`100% <100%> (ø)`

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update bdf2d0c...bb93a9d. Read the comment docs.

fthomas · 2017-12-20T21:12:08Z

Looks great, thanks @howyp!

This would also resolve #75 although that one is about adding a method to RefType. My guess is that companions of refined types will replace the RefType.applyRef, refineV and similar methods in the long run so it is okay to have this functionality only on the companion objects.

fthomas · 2017-12-20T21:15:15Z

modules/cats/shared/src/test/scala/eu/timepit/refined/cats/CatsSpec.scala

@@ -11,16 +12,26 @@ class CatsSpec extends Properties("cats") {
    val refTypeOrder: Unit = () // shadow the `Order` instance so the `Eq` instance is tested
    locally(refTypeOrder) // prevent a "unused" warning

-    PosInt.unsafeFrom(5) === PosInt.unsafeFrom(5)


I'm using unsafeFrom in the tests to avoid the overhead of the RefineMacro so that test code compiles faster.

Ah, ok, I'll revert

fthomas · 2017-12-20T21:22:42Z

modules/cats/shared/src/main/scala/eu/timepit/refined/cats/CatsValidated.scala

+object validation {
+  implicit class CatsValidateOps[FTP, T](companion: RefinedTypeOps[FTP, T]) {
+    def validate(value: T): ValidatedNel[String, FTP] =
+      Validated.fromEither(companion.from(value)).leftMap(NonEmptyList.one)


There is a toValidatedNel on Cats' EitherSyntax which would allow us to just write companion.from(value).toValidatedNel here.

fthomas · 2017-12-20T21:26:49Z

I'm totally fine with using ValidatedNel here. It is probably the predominant data type for error accumulation in the Cats ecosystem, so I don't see the need to deviate from that.

howyp · 2017-12-20T23:09:20Z

Not sure about packaging here - do we need a .syntax as well as a =.validation one? Might be overkill

fthomas · 2017-12-20T23:41:16Z

I'm not sure if I like d59f2d8. It adds a validate extension to any type and I find such extensions always strange. I would be okay if we only supply the validate method for companions and that is the only way to refine values with Validated.
I know that the library currently has more than one way to do things and RefType is one of the worst contender in this regard. I think in the future I'd like to reduce the surface area of refined's API to make the library simpler.

Regarding syntax vs. validation. I think I like syntax more than validation. It would allow us to maybe add more extension methods that are not related to Validated to the companions.

kusamakura · 2017-12-21T07:48:43Z

How about some scalaz love? ;) I can take it up if you want, @howyp.

fthomas · 2017-12-21T08:25:55Z

Adding this functionality to the Scalaz module would be nice too. I would prefer a separate PR for that.

kusamakura · 2017-12-21T08:43:57Z

@fthomas Alright, I'll graft the code from this one into the scalaz module.

howyp · 2017-12-21T09:08:53Z

I can understand why you might feel hesitant about having an operator on anything (though it would only work on types which have a matching refined predicate). It seemed to me to be quite a 'cats' way of doing things - in the style of say .valid[B]. The alternative is to support something like:

import validation._
val myInt = 5
validate[Positive](myInt)

Would you prefer that for the moment?

My motivation was to provide symmetrical support for cases where you haven't defined a full refined type/don't want to create companion objects, but just want to be able to apply a predicate to a raw type. I agree it would be good to distill the API surface of Refined, and that standardising on companion objects for refined types is a great way to go. But only providing second-class support for ad-hoc refinements via a predicate seems a shame to me.

Perhaps how to reduce the surface of the API would be best discussed in a new issue?

kusamakura · 2017-12-21T09:15:49Z

@howyp I think validate[Positive](myInt) would just be refineV[Positive](myInt).toValidatedNel. On the other hand, the companion method is nice.

howyp · 2017-12-21T19:17:22Z

Ok, I've dropped the 10.validate[Positive] syntax

fthomas · 2017-12-21T20:19:38Z

Ok, merging now. Thanks again @howyp!

Add .validate method for refined companions

f57b3e0

fthomas reviewed Dec 20, 2017

View reviewed changes

howyp added 3 commits December 20, 2017 21:56

Use .unsafeFrom in CatsSpec for test compile speed

58faa4b

Use cat's .toValidatedNel in CatsValidated

562b718

Fix name of validation file

bb93a9d

howyp force-pushed the cats-validation branch from d59f2d8 to bb93a9d Compare December 21, 2017 19:12

fthomas merged commit a6c014f into fthomas:master Dec 21, 2017

howyp deleted the cats-validation branch December 21, 2017 20:35

This was referenced Dec 21, 2017

Add error accumulating function to RefType #75

Closed

Add RefinedTypeOps for companion objects of refined types #369

Merged

kusamakura mentioned this pull request Dec 22, 2017

Scalaz validate syntax for RefinedTypeOps #385

Merged

fthomas mentioned this pull request Dec 23, 2017

Reduce API surface #386

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add .validate method for refined companions #382

Add .validate method for refined companions #382

howyp commented Dec 20, 2017

codecov bot commented Dec 20, 2017 •

edited

Loading

fthomas commented Dec 20, 2017

fthomas Dec 20, 2017

howyp Dec 20, 2017

fthomas Dec 20, 2017

fthomas commented Dec 20, 2017 •

edited

Loading

howyp commented Dec 20, 2017

fthomas commented Dec 20, 2017

kusamakura commented Dec 21, 2017

fthomas commented Dec 21, 2017

kusamakura commented Dec 21, 2017 •

edited

Loading

howyp commented Dec 21, 2017 •

edited

Loading

kusamakura commented Dec 21, 2017

howyp commented Dec 21, 2017

fthomas commented Dec 21, 2017

Add .validate method for refined companions #382

Add .validate method for refined companions #382

Conversation

howyp commented Dec 20, 2017

codecov bot commented Dec 20, 2017 • edited Loading

Codecov Report

fthomas commented Dec 20, 2017

fthomas Dec 20, 2017

Choose a reason for hiding this comment

howyp Dec 20, 2017

Choose a reason for hiding this comment

fthomas Dec 20, 2017

Choose a reason for hiding this comment

fthomas commented Dec 20, 2017 • edited Loading

howyp commented Dec 20, 2017

fthomas commented Dec 20, 2017

kusamakura commented Dec 21, 2017

fthomas commented Dec 21, 2017

kusamakura commented Dec 21, 2017 • edited Loading

howyp commented Dec 21, 2017 • edited Loading

kusamakura commented Dec 21, 2017

howyp commented Dec 21, 2017

fthomas commented Dec 21, 2017

codecov bot commented Dec 20, 2017 •

edited

Loading

fthomas commented Dec 20, 2017 •

edited

Loading

kusamakura commented Dec 21, 2017 •

edited

Loading

howyp commented Dec 21, 2017 •

edited

Loading