In general, rubyfmt respects (within limits) user choices about multilining expressions, but we don't currently do this for multiline expressions. This has a few poor side effects: (1) users often want to multiline to make e.g. adding comments for specific sections of a call chain easier, and (2) there end up being some (IMO) awkward formatting of chains like the following:
MyClass.call(
[],
""
).do_some_other_stuff!
which is inconsistent with rubyfmt's normal call chain styling.
This PR implements multilining for user-multilined expressions. It does this by checking the line location of the DotTypeOrOp items in the call chain and multilining if they're on multiple lines. This might be a slightly imperfect if the user writes something like this
# both format as `foo.bar.baz`
foo
.bar.baz
foo.bar.
baz
but IMO this is a reasonable implementation tradeoff. We could alternatively check every expression's start/end, but this ends up requiring a lot more trickery and is more prone to inconsistency.
Related to #359
In general,
rubyfmtrespects (within limits) user choices about multilining expressions, but we don't currently do this for multiline expressions. This has a few poor side effects: (1) users often want to multiline to make e.g. adding comments for specific sections of a call chain easier, and (2) there end up being some (IMO) awkward formatting of chains like the following:which is inconsistent with
rubyfmt's normal call chain styling.This PR implements multilining for user-multilined expressions. It does this by checking the line location of the
DotTypeOrOpitems in the call chain and multilining if they're on multiple lines. This might be a slightly imperfect if the user writes something like thisbut IMO this is a reasonable implementation tradeoff. We could alternatively check every expression's start/end, but this ends up requiring a lot more trickery and is more prone to inconsistency.