Re: [PATCH] About the :distant-foreground face attribute

[Daniel Colascione]

On 01/14/2014 02:05 PM, Jan Djärv wrote:
> 14 jan 2014 kl. 21:06 skrev Daniel Colascione <address@hidden>:
>
> > On 01/14/2014 12:01 PM, Jan Djärv wrote:
> > > No.  If Emacs generates a color, Emacs desides what looks good.  If
> > > the system defines a color, the system (or the user if customized)
> > > desides what looks good.  I don't think it matters what I think about
> > > colors generated by your patch, I might even think they look better
> > > than many system defined colors.  But as a principle I think the
> > > desision is not Emacs to make *by default*.  Users may of course
> > > apply customizations to Emacs and change it.
> >
> > In 24.4, Emacs has already been changed to override the system selection 
> > foreground color with various font-lock faces.
>
> Which font lock faces are you talking about?  No system I know of defines
> system colors for things like comment face, function face etc.
>
> > Why is it okay to do that when there's no contrast problem, but suddenly, when 
> > there's a contrast problem, we can say that the principle of following system 
> > colors is important?
>
> The principle of following system foreground is only important if system
> background is used.

Then why don't we always use the system *selection* foreground color 
when we use the system *selection* background? When we apply font-lock 
foreground colors to a region you've highlighted, we're overriding your 
system's selection foreground color. Why is it okay to override that 
color in some situations but not others?

> This is currently for NS/Gtk+ only.  For Lucid/Motif/No toolkit, we don't use
> system colors at all, because they are not known and can not be known,
> because the API to get them is not available.
>
> > You're applying this principle very selectively.
>
> System background + contrast problem => system foreground.
> How is that selectively, it is a clear rule.

It's clear, but arbitrary. Another clear and arbitrary rule is this:

  System background + contrast problem => adjust foreground.

You haven't presented any justification for your arbitrary rule being 
better than my arbitrary rule.