`set-frame-font' doesn't support font aliases

`set-frame-font' doesn't support font aliases

[Eduardo Ochs]

Part 1 - Problem
----------------

The current code for `set-frame-font' doesn't support font
aliases - for example, these sexps don't work:

  (set-frame-font "5x7")    ;; error: "Font `5x7' is not defined"
  (set-frame-font "fixed")  ;; uses some unexpected font

but these ones do work:

  ;; "5x7":
  (set-frame-font
    "-misc-fixed-medium-r-normal--7-70-75-75-c-50-iso8859-1")
  ;; "fixed" (same as "6x13" in my machine - see below):
  (set-frame-font
    "-misc-fixed-medium-r-semicondensed--13-120-75-75-c-60-iso8859-1")

As far as I can tell, the expansion of aliased fonts is done
using this file:

  (find-file "/usr/share/fonts/X11/misc/fonts.alias")

and pgas (Pierre Gaston) pointed to me that "xlsfonts -l" can be
used to show the real name of a font. Typical command lines are:

  xlsfonts -l  -fn 5x7
  xlsfonts -ll -fn 5x7

I am using a recent CVS Emacs on Debian 4.0 ("Etch", stable, xorg).
Running (emacs-version) here returns this:
  "GNU Emacs 23.0.60.2 (i686-pc-linux-gnu, X toolkit, Xaw3d scroll bars)
   of 2008-05-30 on dekooning"



Part 2 - Workaround
-------------------

(defun ee-aref (alist idx)
  "Like `aref', but for alists.
Example: (ee-aref '((1 . one) (2 . two) (3 . three)) 2)
                                -> two"
  (cdr (assoc idx alist)))

(defun ee-font-alias (fontname &optional use-ucs)
"Expand FONTNAME when it is an alias; when USE-UCS is non-nil, use \"iso10646\".
The list of aliases has been built by hand from data found at this file:
  (find-file \"/usr/share/fonts/X11/misc/fonts.alias\")"
  (let ((aliases '(
    ("fixed" . "-misc-fixed-medium-r-semicondensed--13-120-75-75-c-60-iso8859-1")
    ("variable" . "-*-helvetica-bold-r-normal-*-*-120-*-*-*-*-iso8859-1")
    ("5x7" . "-misc-fixed-medium-r-normal--7-70-75-75-c-50-iso8859-1")
    ("5x8" . "-misc-fixed-medium-r-normal--8-80-75-75-c-50-iso8859-1")
    ("6x9" . "-misc-fixed-medium-r-normal--9-90-75-75-c-60-iso8859-1")
    ("6x10" . "-misc-fixed-medium-r-normal--10-100-75-75-c-60-iso8859-1")
    ("6x12" . "-misc-fixed-medium-r-semicondensed--12-110-75-75-c-60-iso8859-1")
    ("6x13" . "-misc-fixed-medium-r-semicondensed--13-120-75-75-c-60-iso8859-1")
    ("6x13bold" . "-misc-fixed-bold-r-semicondensed--13-120-75-75-c-60-iso8859-1")
    ("7x13" . "-misc-fixed-medium-r-normal--13-120-75-75-c-70-iso8859-1")
    ("7x13bold" . "-misc-fixed-bold-r-normal--13-120-75-75-c-70-iso8859-1")
    ("7x13euro" . "-misc-fixed-medium-r-normal--13-120-75-75-c-70-iso8859-15")
    ("7x13eurobold" . "-misc-fixed-bold-r-normal--13-120-75-75-c-70-iso8859-15")
    ("7x14" . "-misc-fixed-medium-r-normal--14-130-75-75-c-70-iso8859-1")
    ("7x14bold" . "-misc-fixed-bold-r-normal--14-130-75-75-c-70-iso8859-1")
    ("8x13" . "-misc-fixed-medium-r-normal--13-120-75-75-c-80-iso8859-1")
    ("8x13bold" . "-misc-fixed-bold-r-normal--13-120-75-75-c-80-iso8859-1")
    ("8x16" . "-sony-fixed-medium-r-normal--16-120-100-100-c-80-iso8859-1")
    ("9x15" . "-misc-fixed-medium-r-normal--15-140-75-75-c-90-iso8859-1")
    ("9x15bold" . "-misc-fixed-bold-r-normal--15-140-75-75-c-90-iso8859-1")
    ("10x20" . "-misc-fixed-medium-r-normal--20-200-75-75-c-100-iso8859-1")
    ("12x24" . "-sony-fixed-medium-r-normal--24-170-100-100-c-120-iso8859-1")
    )))
    (setq fontname (or (ee-aref aliases fontname) fontname))
    (if use-ucs (setq fontname (replace-regexp-in-string
				"iso8859" "iso10646"
				fontname)))
    fontname))

(defun ee-set-frame-font (fontname &optional use-ucs)
  "Like `set-frame-font', but uses `ee-font-alias' to expand FONTNAME."
  (set-frame-font (ee-font-alias fontname use-ucs)))

;; For tests
(defun ee-insert (&rest rest)
  "Insert characters, strings, or ranges of characters.
Example: (ee-insert '(?a ?z) 10 \"Foo!\")"
  (while rest
    (let ((o (car rest)))
      (cond ((stringp o) (insert o))
	    ((numberp o) (if (char-valid-p o) (insert o)))
	    ((consp o) (mapc 'ee-insert (apply 'number-sequence o)))
	    (t (error "Not string/int/pair: %S" o))))
    (setq rest (cdr rest))))

;; Tests:
;; (ee-font-alias "fixed")
;; (ee-font-alias "fixed" t)
;; (ee-set-frame-font "5x7" t)
;; (ee-set-frame-font "6x10" t)
;; (ee-set-frame-font "fixed" t)
;; (ee-set-frame-font "9x15" t)
;; (ee-set-frame-font "10x20" t)
;; (ee-insert "\n" '(32 126) "\n" '(913 981) "\n" '(8592 9675) "\n")



Part 3 - Suggestion
-------------------

It would be nice to have an Emacs function that would do this
expansion - like the (ee-aref aliases fontname) above - cleanly
and reliably...


Cheers,
  Eduardo Ochs
  eduardoochs@gmail.com
  http://angg.twu.net/

Re: `set-frame-font' doesn't support font aliases

[Kenichi Handa]

In article <87fxrxpw0z.fsf@dekooning.omnisys.int.com.br>, Eduardo Ochs <eduardoochs@gmail.com> writes:

> The current code for `set-frame-font' doesn't support font
> aliases - for example, these sexps don't work:

>   (set-frame-font "5x7")    ;; error: "Font `5x7' is not defined"
>   (set-frame-font "fixed")  ;; uses some unexpected font

I've just installed a fix to check also an alias name in X
font-backend.  But, as Emacs accepts a fontconfig-style font
name too, for instance, "fixed" is parsed as a font of
family "fixed" which may match with many fonts.

Is keeping backward compatibility of (set-frame-font
"fixed") important enough to drop the support of
fontconfig-style name?

---
Kenichi Handa
handa@ni.aist.go.jp

Re: `set-frame-font' doesn't support font aliases

[Stefan Monnier]

>> The current code for `set-frame-font' doesn't support font
>> aliases - for example, these sexps don't work:

>> (set-frame-font "5x7")    ;; error: "Font `5x7' is not defined"
>> (set-frame-font "fixed")  ;; uses some unexpected font

> I've just installed a fix to check also an alias name in X
> font-backend.  But, as Emacs accepts a fontconfig-style font
> name too, for instance, "fixed" is parsed as a font of
> family "fixed" which may match with many fonts.

> Is keeping backward compatibility of (set-frame-font
> "fixed") important enough to drop the support of
> fontconfig-style name?

Can you recover the Emacs-22 behavior by restricting the backends to
just `x'?  If so, I think it's OK.


        Stefan

Re: `set-frame-font' doesn't support font aliases

[Kenichi Handa]

In article <jwvy75nrn8b.fsf-monnier+emacs@gnu.org>, Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>> The current code for `set-frame-font' doesn't support font
>>> aliases - for example, these sexps don't work:

>>> (set-frame-font "5x7")    ;; error: "Font `5x7' is not defined"
>>> (set-frame-font "fixed")  ;; uses some unexpected font

> > I've just installed a fix to check also an alias name in X
> > font-backend.  But, as Emacs accepts a fontconfig-style font
> > name too, for instance, "fixed" is parsed as a font of
> > family "fixed" which may match with many fonts.

> > Is keeping backward compatibility of (set-frame-font
> > "fixed") important enough to drop the support of
> > fontconfig-style name?

> Can you recover the Emacs-22 behavior by restricting the backends to
> just `x'?

No, because the font-name parser is independent of font
backends; i.e. even if you use only x font-backend, you can
specify a font name as, for instance,
"fixed-12:foundry=misc".

It is possible to implement a new method
`resolve_alias_name' for a font-backend, and add a new
function font-resolve-alias-name.

Then one can do:
 (set-frame-font (font-resolve-alias-name "fixed")).

Another idea is to do:
  (set-frame-font (font-spec :alias "fixed)).
or
  (set-frame-font ":alias=fixed")

---
Kenichi Handa
handa@ni.aist.go.jp

Re: `set-frame-font' doesn't support font aliases

[Stefan Monnier]

>> Can you recover the Emacs-22 behavior by restricting the backends to
>> just `x'?

> No, because the font-name parser is independent of font
> backends; i.e. even if you use only x font-backend, you can
> specify a font name as, for instance,
> "fixed-12:foundry=misc".

Then we need to detect the ambiguity, and provide a way for the user to
resolve the ambiguity.

> Another idea is to do:
>   (set-frame-font (font-spec :alias "fixed)).
> or
>   (set-frame-font ":alias=fixed")

That seems clean from the user's POV.  Tho maybe aliases should take
precedence, rather than the reverse.


        Stefan

Re: `set-frame-font' doesn't support font aliases

[Kenichi Handa]

In article <jwvr6bfq6a2.fsf-monnier+emacs@gnu.org>, Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>> Can you recover the Emacs-22 behavior by restricting the backends to
>>> just `x'?

> > No, because the font-name parser is independent of font
> > backends; i.e. even if you use only x font-backend, you can
> > specify a font name as, for instance,
> > "fixed-12:foundry=misc".

> Then we need to detect the ambiguity, and provide a way for the user to
> resolve the ambiguity.

> > Another idea is to do:
> >   (set-frame-font (font-spec :alias "fixed)).
> > or
> >   (set-frame-font ":alias=fixed")

> That seems clean from the user's POV.  Tho maybe aliases should take
> precedence, rather than the reverse.

Something like this for "Bitstream Vera Sans"?
  ":family=Bitstream Vera Sans"

I don't like it because the new format is incompatible with
fontconfig's way.  So, the original question arises:

> Is keeping backward compatibility of (set-frame-font
> "fixed") important enough to drop the support of
> fontconfig-style name?

Or, isn't it good enough to ask users to use this form
instead:

(set-frame-font ":alias=fixed")

---
Kenichi Handa
handa@ni.aist.go.jp

Re: `set-frame-font' doesn't support font aliases

[Stefan Monnier]

>> That seems clean from the user's POV.  Tho maybe aliases should take
>> precedence, rather than the reverse.

> Something like this for "Bitstream Vera Sans"?
>   ":family=Bitstream Vera Sans"

> I don't like it because the new format is incompatible with
> fontconfig's way.  So, the original question arises:

>> Is keeping backward compatibility of (set-frame-font
>> "fixed") important enough to drop the support of
>> fontconfig-style name?

> Or, isn't it good enough to ask users to use this form
> instead:

> (set-frame-font ":alias=fixed")

Maybe you're right and it's sufficient.


        Stefan