RE: frame-set-background-mode changes frame backgroundforspecial-display frames

[Drew Adams]

        Why is it correct to use the default background of
        `default-frame-alist' for a special-display frame
        that I explicitly assigned a different background
        color?

    You did not say that you assigned a different background color.

I thought I did:

         I have a set of frame parameters for
         special-display frames, and other sets for frames
         dedicated to buffers *Help* and *Completions* (using
         `special-display-buffer-names' with explicit frame
         parameters). What happens is that the background colors
         of these frames are not what they should be.

Sorry if that wasn't clear enough.

    Please provide a _complete, self-contained test case_ for every bug
    report.

I see this problem systematically in my own setup - but, as I indicated,
only with CVS snapshots newer than GNU Emacs 22.0.50.2
(i386-mingw-nt5.1.2600) of 2005-04-16 on LAPTOP Distributor `Microsoft
Corp.', version 5.1.2600
configured using `configure --with-gcc (3.4)'.

I have spent a *lot* of time trying to track this down in my own setup, as I
did not succeed in starting from emacs -Q and building up with pieces of my
code to reproduce it. That's why I started my bug report with the caveat: "I
don't have a simple recipe starting from emacs -Q".

I would have thought that putting the following code in a file foo.el and
then starting emacs with emacs -Q -l "foo.el" might be enough to reproduce
the bug (e.g. trying `C-h i'), but it is not enough. I must be doing
something more in my setup that manifests the bug.

(remove-hook 'same-window-regexps "\\*info\\*\\(\\|<[0-9]+>\\)")
(setq special-display-regexps '("[ ]?[*][^*]+[*]"))
(setq minibuffer-frame-alist
      (append (list (cons 'minibuffer 'only)
                    (cons 'background-color "PaleGoldenrod"))
              minibuffer-frame-alist))
(setq default-frame-alist
      (append (list '(minibuffer)
                    (cons 'background-color "LightBlue"))
              default-frame-alist))
(setq special-display-frame-alist
      (append (list (cons 'background-color "LightSteelBlue"))
              special-display-frame-alist))
(setq pop-up-frames t
      pop-up-frame-alist (append default-frame-alist pop-up-frame-alist))
(setq my-minibuffer-frame (make-frame minibuffer-frame-alist))

Sorry, but I cannot do better than the description I gave. Please take a
look at the code that I found was causing the frames to change background
color. Hopefully it will provide some clue; in my setup, it is the culprit.
What I can't tell you is what there is in my setup that works together with
that code to cause the problem. I can say that my code works fine with Emacs
versions from 20.7 through the April 2005 CVS Emacs 22 snapshot. It's
possible that my code has always done something wrong that was always
tolerated before and is no longer, correctly. I suspect, instead, that a bug
was introduced into Emacs after April 2005.

I doubt that you want me to send multiple libraries to reproduce part of my
setup, and I don't have the time needed to narrow the problem down further
and provide a self-contained test case. I think that's the best I can do
now. If you could recognize a problem in the code in question, then that
would be good. If not, too bad for me. Thanks for taking a look.