Compilation error while flatmapping JSValue

[kamilkloch]

The following code

import org.json4s._
import org.json4s.jackson.JsonMethods._

val json = parse(...)

val bidsAndAsks1 = for {
    JObject(child) <- json
    JField("symbol", JString(name)) <- child
    ...
} yield ...

produces

[error] CustomResources.scala:108:20: value flatMap is not a member of org.json4s.JValue
[error]     JObject(child) <- json
[error]                    ^
[error] one error found

Compilation succeeds without the plugin. Plugin version 0.3.1.

[oleg-py]

Can't be fixed. This DSL is directly at odds against what better-monadic-for is supposed to do.

json4s heavily relies on the fact that you'd have withFilter clause. It provides an implicit conversion to MonadicJValue, which does not have flatMap, and then withFilter creates a JValueWithFilter which does. Because bm4 whole purpose is to remove that withFilter call, the conversion never happens. See here. map/flatMap/withFilter both eventually delegate to filter of MonadicJValue which essentially puts the value into a List and filters that.

That won't be super helpful, however. By removing withFilter clause, you don't get to actually filter out values that don't match, and you'll end up crashing with MatchError. So, this DSL is incompatible with bm4, period.

In dotty, it would require case keyword to be usable as patterns are irrefutable by default (scala/scala3#2578). So you are going to have to write:

for {
    case JObject(child) <- json
    case JField("symbol", JString(name)) <- child
    ...
} yield ...

I looked into adding something similar to bm4, but it is quite a bit of effort, and you're the first to need it in over a year, so no promises it'll ever land here, esp. given that dotty is coming reasonably soon. You'd have to either abandon bm4 or this DSL of json4s.

[kamilkloch]

Thanks for the comprehensive and super-fast response!!