TrailingBlankLines should not trigger on empty file with single blank line

[JeanMertz]

Files with a single (empty) line should not trigger TrailingBlankLines: Final newline missing..

[jonas054]

True.

[jonas054]

I get the following results when I run rubocop -D (0.20.1 or latest master) on an entirely empty file and one with just a newline under Ruby 1.9.3:

/tmp/993/newline.rb:1:1: C: Encoding: Missing utf-8 encoding comment.
/tmp/993/empty.rb:1:1: C: Encoding: Missing utf-8 encoding comment.
/tmp/993/empty.rb:1:1: C: TrailingBlankLines: Final newline missing.

I think it's wrong to report any offense for an empty file, while the report about encoding comment for newline.rb is ok.

[JeanMertz]

@jonas054 is there any reason to add the encoding comment when there is no content (besides the \n) in the file? It seems weird to me to force an encoding comment on an empty file.

(I'm using Ruby 2.1 myself, so didn't run into the warning you described)

[bbatsov]

I guess it doesn't make sense to run cops on blank files.

[JeanMertz]

@bbatsov except for the ones that act on something other than the content (for example, the snake case naming one).

[bbatsov]

Yep, you're totally right.

[jonas054]

@JeanMertz You have to draw the line somewhere. I'd like to draw it between an empty file and a non-empty file. If there is at least one character, then the file has an encoding.

[JeanMertz]

@jonas054 seems reasonable. Although what about @bbatsov's (slightly modified) proposal of not running any content cops on empty files? I'm not sure if there's already such a classification (ie. if the cop checks the file, or the contents of the file), but that would completely solve this issue.

[jonas054]

No there is no such classification, so we would need to introduce a special case for one cop in generic code. That's not so good. I prefer to fix the problem in Encoding and TrailingBlankLines, and then add a high level spec example checking that no offenses are reported for an empty file.