D3212: patch: implement a new worddiff algorithm

spectral (Kyle Lippincott) phabricator at mercurial-scm.org
Tue Apr 10 15:39:01 EDT 2018


spectral added subscribers: dhduvall, indygreg.
spectral added inline comments.

INLINE COMMENTS

> quark wrote in color.py:94
> As I mentioned in the summary, I believe `diff.inserted` and `diff.inserted.changed` should have a same color. And `diff.inserted` probably shouldn't be bold.
> 
> Looking at this review page, you will notice the `diff.inserted` and `diff.inserted.changed` are using a same color, where `diff.inserted.unchanged` is using a different (lighter) background color.
> 
> `dim` works fine where it is supported. For terminals that do not support it, people can override the settings. For `dim` feature detection, that's an issue in the color code which is unrelated to this change. Since worddiff is experimental and off by default, I don't think dim detection should block this patch.

I understand the reasoning for wanting `diff.inserted` and `diff.inserted.changed` to be the same color, but unfortunately I think it might not be super feasible in upstream.

'dim' support in my terminfo database (and via testing) is actually pretty common among the biggest terminals; for oddities, I'm seeing:

- the linux console (very few people use it) has weird support
- screen simulates it with underline, tmux passes it through (so I don't see it)
- rxvt-unicode doesn't support it
- the terminfo profile for cygwin doesn't indicate support for it
- Apple's terminal advertises itself as xterm-compatible and seems to support it
- iTerm2 similarly

I'm willing to be convinced it's OK, especially since this mode isn't the default.  @dhduvall originally wrote the "not supported by many terminals" bit, I wonder if they have any suggestions.  @indygreg has also dealt with terminal issues (and deals with windows more than me, so might know more there).

REPOSITORY
  rHG Mercurial

REVISION DETAIL
  https://phab.mercurial-scm.org/D3212

To: quark, #hg-reviewers, durin42, yuja
Cc: indygreg, dhduvall, yuja, spectral, mercurial-devel


More information about the Mercurial-devel mailing list