fmt: line-wrapping differs from gnu implementation in some situations

[ChrisDryden]

Description

When trying to come across edge cases in the non utf-8 fmt PR I came across that some outputs to fmt were not matching due to the line breaks being inserted in different places

Examples:

  $ printf 'aa bb cc dd ee' | /usr/bin/fmt -w 8
GNU
  aa
  bb cc
  dd ee

uutils
  aa bb cc
  dd ee


  $ printf '漢 字 test 日 本 語' | /usr/bin/fmt -w 10
GNU
  漢 字
  test 日
  本 語

uutils
  漢 字 test
  日 本 語


  $ printf '漢字 test 日本語' | /usr/bin/fmt -w 15
GNU
  漢字 test
  日本語

uutils
  $ printf '漢字 test 日本語' | uutils fmt -w 15
  漢字 test 日本語

Where to fix:

src/uu/fmt/src/linebreak.rs is the file that has the logic

[ChrisDryden]

This might be the same issue as this: #5162 but its not limited to the goal option, it also applied to -w

[pixelb]

GNU fmt currently counts the \n as part of the count. That's confusing and inconsistent. I'll change that in the next couple of days. See GNU discussion

[elskow]

is this still up for grabs? I can take a take look at it.

Also, re: @pixelb's note. I'm assuming we want to match the current GNU output for now despite the upcoming changes?

[ChrisDryden]

Yes I believe the goal is to target 1-1 compatibility with latest GNU