Open josh-degraw opened 10 months ago
Thanks for bringing this up!
Bikeshedding question: should this be the minimal change necessary (1 space)? Or should it be a multiplier of your indent_size (if you use 2 spaces, use 2 spaces)?
Using the indent_size
will make it seem more deliberate I think. A single space will indeed strike more as a bug.
From a style guide perspective, the above proposal seems fine.
This code is invalid. The compiler sees these as two separate statements. So short of updating the compiler, code in this style will need to have at least one additional space on the newline before the closing angle bracket.
I agree that ideally this should be valid for the compiler.
As discussed in https://github.com/fsprojects/fantomas/issues/2706 and https://github.com/fsprojects/fantomas/pull/2852, with the recent changes regarding multiline bracket formatting, multiline generic type parameters don't necessarily always follow the same guidelines, so we should formalize how they can/should be formatted.
For example, if using
Stroustrup
bracket style, you might expectto be formatted like this:
However, the placement of the angle brackets can be problematic.
For example, if you have an application expression:
And try to directly apply
Stroustrup
style to it:This code is invalid. The compiler sees these as two separate statements. So short of updating the compiler, code in this style will need to have at least one additional space on the newline before the closing angle bracket.
Bikeshedding question: should this be the minimal change necessary (1 space)? Or should it be a multiplier of your
indent_size
(if you use 2 spaces, use 2 spaces)?Also, it probably should be noted that truly "Aligned" angle bracket style doesn't work:
Some users may expect that to work but it's invalid code. We should maybe update the style to mention this (if it's not there already).