For-comp with no yield synthesises a .map call

[japgolly]

reproduction steps

Desugar the following:

for {x <- List(1); y = 1} ()

problem

It effectively desugars to:

List(1).map { x => val y = 1; (x, y) }.foreach { case (x, y) => () }

expectation

Because there's no yield, I expect all .flatMap and .map calls to be .foreach instead. It should desugar to:

List(1).foreach { x => val y = 1; () }

[Jasper-M]

Isn't that more or less the general issue that assignments (y=1) in for comprehensions incur an extra map call?

[joroKr21]

You need https://github.com/oleg-py/better-monadic-for

[martijnhoekstra]

working as specced at https://scala-lang.org/files/archive/spec/2.13/06-expressions.html#for-comprehensions-and-for-loops to everyones chagrin as

A generator p <- e followed by a value definition p′ = e′ is translated to the following generator of pairs of values, where x and x′ are fresh names:

(p, p′) <- for (x@p <- e) yield { val x′@p′ = e′; (x, x′) }

https://contributors.scala-lang.org/t/pre-sip-improve-for-comprehensions-functionality/3509/52 is the canonical contributors thread.

[dwijnand]

Perhaps scala/scala3#2573 is the canonical epic.

[SethTisue]

unlikely to progress in Scala 2. Scala 3 is a different story

[SethTisue]

(if it ever were to progress in Scala 2, it would be because it happened in Scala 3 first)