Re: Gnus problem with nil font in face definition

[Ted Zlatanov]

On Fri, 29 Feb 2008 16:28:34 +0900 Kenichi Handa <address@hidden> wrote: 

KH> In article <address@hidden>, Reiner Steib <address@hidden> writes:
>> On Thu, Feb 28 2008, Ted Zlatanov wrote:
>> > It seems that the cause is faces that have the font set to nil, instead
>> > of unspecified.  I don't know if this is a Gnus-only problem, and it may
>> > be caused by an Emacs problem, so I am cc-ing both emacs-devel and ding.

>> I'm quite sure that nothing in Gnus/Message has changed in this area,
>> so it is probably related to the unicode merge (new font back end?)

>> For those not using Gnus, a test case might be:

>> - emacs -Q -fn '-misc-fixed-medium-r-normal--20-200-75-75-c-100-iso10646-1'

KH> I found what is wrong with the current font-backend code.

KH> % xlsfonts -fn -misc-fixed-*-*-*--20-200-75-75-c-100-iso10646-1

KH> lists these fonts.

KH> -misc-fixed-bold-r-normal--20-200-75-75-c-100-iso10646-1
KH> -misc-fixed-bold-r-normal--20-200-75-75-c-100-iso10646-1
KH> -misc-fixed-bold-r-semicondensed--20-200-75-75-c-100-iso10646-1
KH> -misc-fixed-medium-o-normal--20-200-75-75-c-100-iso10646-1
KH> -misc-fixed-medium-o-semicondensed--20-200-75-75-c-100-iso10646-1
KH> -misc-fixed-medium-r-normal--20-200-75-75-c-100-iso10646-1
KH> -misc-fixed-medium-r-normal--20-200-75-75-c-100-iso10646-1
KH> -misc-fixed-medium-r-normal--20-200-75-75-c-100-iso10646-1
KH> -misc-fixed-medium-r-normal--20-200-75-75-c-100-iso10646-1
KH> -misc-fixed-medium-r-semicondensed--20-200-75-75-c-100-iso10646-1

KH> but the result of

KH> % xlsfonts -fn -misc-fixed-*-*-*--*-*-*-*-c-*-iso10646-1

KH> doesn't contain this font:
KH> -misc-fixed-bold-r-normal--20-200-75-75-c-100-iso10646-1

I don't know the font selection semantics, but this seems
counterintuitive.

KH> I'm now working on handling such a case correctly.
KH> By the way, if you specify this font (i.e. a font whose bold
KH> versoin also exists without auto-scaling):

KH> -misc-fixed-medium-r-normal--18-120-100-100-c-90-iso10646-1

KH> Emacs can find the correct bold version (at least in my
KH> environment).

Was my guess correct that a nil font in the face specification is the
problem?  Or is it specifically a problem with the font I use, as I
think you're saying?  Or both?

Ted