Support Box

[latot]

Hi, I'm using lintr in code and github, I think would be great support the box module, because actually lintr don't detect the functions contained in the modules.

https://github.com/klmr/box

Thx.

[AshesITR]

Hi, can you describe what kind of support you'd like?
Note that the object_usage_linter can see functions defined in your globalenv, so it may help to import your boxes before calling lint() if you see too many false positives from that particular linter.

[latot]

mm, I have a code, but seems the problem is in other place, i'll check it deeper, for now, I think would be great a fix for this:

b.R:1:11: style: Put spaces around all infix operators.
box::use(./a)

Thx.

[MichaelChirico]

seems like an unfortunate syntax choice by box to save 2 characters IMO

[AshesITR]

If you are using the development version of lintr, you could wrap your box::use() statements in a nolint block like this:

# nolint start: infix_spaces_linter.
box::use(./a)
# nolint end

[latot]

I'm not using the dev version, I was checking the options, I think woukd be great have a little more doc about how to use the options. For now, I put an issue in box, this can works as a workaround thx! still think this need a proper solution.

[AshesITR]

In lintr 2.0.1, you can use # nolint start instead of # nolint start: infix_spaces_linter. as a workaround (though this will suppress all lints in the surrounded region, so use with care).

[klmr]

seems like an unfortunate syntax choice by box to save 2 characters IMO

Just to be clear, the reason why ‘box’ uses unquoted expressions is not “to save 2 characters”. For the curious, klmr/box#68 has an explanation of the rationale (tl;dr: type correctness).

— Do ‘lintr’ rules support scoping? I.e. could I write a ‘lintr’ rule that would enforce space around binary operators except inside a box::use declaration?

[AshesITR]

Yes, @klmr, the infix_spaces_linter uses the XML parse tree, so can do arbitrary XPath based rules to lint or not lint something.
To avoid false negatives, it's probably best to exactly match box::use(<expr>/<expr>) as a negative in this case.

Feel free to create a PR adding this exception to infix_spaces_linter.