Re: mode-line-inverse-video

[Noah Friedman]

Miles wrote:
>So can someone tell me what deal is with the `mode-line-inverse-video'
>variable?  It suffices to force mode-lines (and header-lines) into
>inverse-video if their face doesn't do so, but setting it to nil doesn't
>do the opposite.  Since I would guess the primary use of this variable by
>users is to *turn off* inverse-video mode-lines (as the default is t),
>this behavior only confuses things.
>
>I would suggest that it's simply too blunt a stick, and should be removed
>(with some documentation referring to the appropriate faces to modify),
>or the current behavior reversed (i.e., if it's nil, then force all
>mode-lines to be non-inverse-video, otherwise do nothing).

One of the things that really annoyed me about early versions of Lucid
Emacs was the fact that obsolete, and sometimes completely vestigial
variables were kept around and their docstrings not even updated to reflect
the fact that they were no longer used.

I think that mode-line-inverse-video ought to be eliminated entirely in
favor of changing faces instead.  It might or might not break someone's
.emacs (it depends on whether they are setting it or using its existing
value), but that could be looked at as a good thing: it brings attention to
the fact that the variable no longer exists.

Or perhaps the variable should remain declared but have a docstring only
saying "this variable is obsolete; here's what it used to do..."  and it
should be marked as obsolete for the byte compiler.  IMO, it should not
have any effect even on ttys; let's kill ad-hoc APIs that have been
replaced with more systematic ones.  Major version number releases are the
right time to do it.