Re: display problem

[Kenichi Handa]

In article <address@hidden>, Jason Rumney <address@hidden> writes:

> Miles Bader wrote:
> > The "19" should be "21" in the unicode branch.
> >
> >   
> I'm not sure I understand what this magic number 19/21 is. Is there a
> constant for it that could be substituted?

That number should be the same as this in src/lisp.h.

#define CHARACTERBITS 22

I've just installed changes for glyph-char and glyph-face
for the same reason.

> I'm surprised that that is the only change needed. The code points
> between the unicode branch and Emacs 22 should be completely different,
> shouldn't they?

Yes.  But, the ASCII character codes are the same.  So, the
following code doesn't have to be changed as far as it uses
the correct version of make-glyph-code.

(aset standard-display-table
      ?\014
      (vconcat "\n"
               (mapcar (lambda (c) (make-glyph-code c 'my-highlight))
               "          Section (Printable Page)          ")))

It should also work even if the string "Section (Printable
Page)" contains non-ASCII characters as far as the file
containing that code is decoded correctly at the time of
loading.

By the way, in emacs-unicode-2, make-glyph-code doesn't work
for a face of ID greater than 511 (in Emacs 22, maximum face
ID is 4095.  I think there are fairly easy two ways to
increase this limit.

(1) By contriving the usage of bits.

For instance, by using 1-bit as a flag, we can use 13-bit
faces for all Unicode BMP characters, and 8-bit faces for
the other less-frequenty-used characters.

(2) By using a table that maps 511 face codes to the actual
face IDs.

The table registers all faces given to make-glyph-code.  So,
at most 511 different faces can be used in make-glyph-code.

And the both methods can be combined.

Is it worth doing such a work?   Currently:
  (length (face-list)) => 140
So we won't reach the limit 511 for a while.

---
Kenichi Handa
address@hidden