Davy Landman

Yes, I agree, only add default layout in the end, if non of the modules ended up contributing layout. That way you get less surprises, but still enable the case...

A lot of things are happening in this PR, and I'm missing: - new tests for the functionality added. - discussion on effect of existing json serialization (which we also...

Is there a way to increase the test coverage? Only 35% of the added lines for the JSONReader are covered according to codecov.

Cool stuff :+1:

``` rascal>lexical D = "d"+; ok rascal>Tree d1 = [D]"dddd"; Tree: appl( prod( lex("D"), [iter(lit("d"))], {}), [appl( regular(iter(lit("d"))), [appl( prod( lit("d"), [\char-class([range(100,100)])], {}), [char(100)]),appl( prod( lit("d"), [\char-class([range(100,100)])], {}), [char(100)]),appl( prod(...

ah, that would make more sense indeed.

``` rascal>int x = 0; int: 0 rascal>for (/t:x := [C]"ccc") println(t); list[void]: [] rascal>for (/t:d1 := [C]"ccc") println(t); c c c list[void]: [] ``` It does seem related to...

The behavior with true fresh variables is different: ``` rascal>for (/Tree t := [C]"ccc") println(t); c c c c c c ccc ccc list[void]: [] rascal>for (/t:d1 := [C]"ccc") println(t);...

@jurgenvinju since #2255 is merged, can we close this one?

@jurgenvinju I think we can close this one right?