Empty arrays (as passed by her/faraday) are parsed as [nil]

[PragTob]

👋

And first thanks for all the work on the project! 💚

We have a grape API and applications talking to each other, for which we use her which uses faraday.

At some point the faraday behavior changed for sending empty arrays lostisland/faraday#801

The problem occurs when we use her to query resoruces Resources::Something.where(param_name: []). It used to be that they were omitted (at least as URL params, unsure) but now they are sent as param_name[] in the URL, which is a bugfix. Grape parses this as [nil] which is not what I'd expect (I'd expect []) and also then fails the parameter validation (we have a type of Array[String]). allow_blank: true doesn't help because an array with nil isn't blank (and we'd still have the wrong value to deal with).

We're running the latest grape release (1.3.3).

I'll work around the issue with coerce_with or defining my own type or something. Still thought it was worth opening an issue about it.

[dblock]

There was a bunch of work related to this in 1.4.0 (HEAD). I suggest writing a spec and seeing if master has fixed this, first?

[PragTob]

@dblock thanks for letting me know :) I have a bunch of specs for this. Let me see if it's fixed on main.

[PragTob]

@dblock just tried it out with 197d5b4 - still exactly the same behavior for us.

[dnesteryuk]

@PragTob Could you add a spec for this bug, please?

I've tried this in tests

get '/required?cats[]='

declared(params) gave:

{"cats"=>[""]}

[PragTob]

Morning @dnesteryuk !

I believe the difference is = vs. no = aka in our case we don't have the =/the way faraday/grape queries.

I'll add a PR with a spec (at least) just have to check whether I can do it on company time or if it's gotta be my own :)

[PragTob]

Failing spec etc. sits at #2080 now

[PragTob]

Moved to rack/rack#1696 as it seems to me it's rather in rack.