Re: [RFC/PATCH] builtin/blame: darken redundant line information

From: Stefan Beller <sbeller@google.com>
Date: Jun 13, 2017, 18:00 UTC
Message-ID: <CAGZ79kakSt8bYr7E9gnNPSVFafyD9G+VUAU+hBBJnX+r_TH=Ag@mail.gmail.com>
In-Reply-To: <xmqq37b37p9o.fsf@gitster.mtv.corp.google.com>

On Tue, Jun 13, 2017 at 10:48 AM, Junio C Hamano <gitster@pobox.com> wrote:

Show 15 quoted lines

>
> I never said "start and end" (you did).  I just wanted the boundary
> of A and B and C clear, so I'd be perfectly happy with:
>
>          context
>         +A      dim
>         +A      dim
>         +A      highlight #1
>         +C      highlight #2
>         +B      highlight #1
>         +B      dim
>         +B      dim
>          context
>
> You can do that still with only two highlight colors, no?

So to put it into an algorithm:

1) detect blocks
2) if blocks are adjacent, their bounds are eligible for highlighting
3) the highlighting is implemented using the "alternate" strategy
  in that any line highlighted belonging to a different block flips
  the highlighting, such that:

          context
         +A      dim
         +A      dim
         +A      highlight #1
         +B      highlight #2
         +B      dim
         +B      dim
          context

So if we go this way, we would need indeed 6 colors:

  Dimmed, Highlighted, HighlightedAlternative

color-moved modes:
nobounds::
  uses dimmed only
allbounds::
adjacentbounds::
  See algorithm above, using dimmed for inside the block and
  both highlights for bounds, making sure adjacent block bounds
  alternate the highlighting color.
alternate::
  Uses only highlighting colors, complete block is colored with
  one of the highlights

I think that is reasonable to implement. But I do still wonder if we really want to add so many new colors. I'll give it a try after my next submodule series.

Thanks, Stefan