D5139: VCS commit message width feedback

[René J.V. Bertin]

rjvbb added a comment.


  Please read through the original review on ReviewBoard, it contains a number of exchanges with Aleix that I haven't copied here (and I opened this ticket only because RB is being phased out). I will see about linking to or copying the screenshots attached to the RB RR.
  
  In short, I see 2 related issues with the current implementation that uses underlining:
  
  1. There is no indication whatsoever that the underlining refers to something other than spelling mistakes.
  2. The signal modality is completely unrelated to the situation being flagged and the code actually calls for the kind of underlining used by the spellchecker, which doesn't help with point 1.
  
  > why do you change the spacing? what has this to do with width?
  
  But that's exactly the point: spacing HAS to do with width. When you change the letter spacing the immediate effect is that the line becomes wider, and that should give a subtle but clear enough reminder that you are typing a line that is getting too long.
  The whole proposition is based on the principle that text should contain hints that are easy enough to understand (because relevant) but also easy enough to ignore.
  Seeing letter spacing increase while typing immediately gives me the hint that I should insert a newline if I want to heed the standard commit message guidelines. Of course the change to Italics when you exceed the "hard" limit isn't a width-related modality change but it shouldn't be too hard to understand either when you're already aware you've typing a long line. I considered using bold, but the effect of that depends too much on typeface and point size in use.
  
  Underlining is also distracting esp. when it uses different colours and may cause clipping (it did with certain versions of KTextEditor, IIRC). The "before" screenshot also confirms my recollection that the whole line was underlined, not just the part exceeding the limit. While that isn't completely wrong it also doesn't help understanding the warning. Finally, red suggests an error while here we are talking about mere guidelines that may or may not correspond to more or less severe errors depending on the repository.

REPOSITORY
  R33 KDevPlatform

REVISION DETAIL
  https://phabricator.kde.org/D5139

To: rjvbb, #kdevelop, mwolff
Cc: mwolff, kdevelop-devel, #kdevelop, geetamc, Pilzschaf, akshaydeo, surgenight, arrowdodger
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kdevelop-devel/attachments/20170328/d6e11c58/attachment.html>