[Fontconfig] Webdings and other MS symbol fonts don't display

[Behdad Esfahbod]

Thanks.  Pushed to master.

On 15-05-18 01:58 AM, Akira TAGOH wrote:
> looks good to me.
> 
> On Mon, May 18, 2015 at 4:48 PM, Behdad Esfahbod <behdad at behdad.org
> <mailto:behdad at behdad.org>> wrote:
> 
>     Revised patch attached, to declare zero language coverage for symbol fonts.
> 
>     On 15-05-18 12:29 AM, Behdad Esfahbod wrote:
>     > Patch attached.  Needs documentation and comments, but otherwise looks about
>     > right to me.  Please review.
>     >
>     > b
>     >
>     > On 15-05-17 05:09 AM, Raimund Steger wrote:
>     >> On 05/14/15 21:38, Behdad Esfahbod wrote:
>     >>> On 15-05-14 03:32 AM, Raimund Steger wrote:
>     >>>> On 05/13/15 03:45, Behdad Esfahbod wrote:
>     >>>>> On 15-05-12 04:45 PM, Raimund Steger wrote:
>     >>>>>> [...]
>     >>>>
>     >>>> Anyway when I try rendering some text in WPF using WingDings I see
>     that the
>     >>>> glyphs are accessible in two ways:
>     >>>>
>     >>>> * old 8-bit codepoints (0x20..0xff)
>     >>>> * PUA codepoints (0xf020..0xf0ff)
>     >>>
>     >>> What happens if you remove the macroman subtable?  Does it still do
>     the 8-bit
>     >>> mapping?
>     >>
>     >> Yes, same behavior. (To avoid confusion with the stock Wingdings I
>     renamed the
>     >> stripped font and used the WPF FontFamily constructor with a distinct
>     >> directory location, so I'm quite sure.)
>     >>
>     >>> [...]
>     >>>
>     >>> Ok, so detecting such symbol fonts and treating them specially is
>     certainly
>     >>> possible.  We just need to figure out what special treatment is
>     suitable.  I
>     >>> think I'm fine with adding your hack, but also marking the font with a
>     special
>     >>> marker, such that only if a binding=strong family match happens the
>     font is
>     >>> picked up and never as a fallback.  That should address all problems
>     we know
>     >>> of, right? ;)
>     >>
>     >> I think the binding=strong check may not even be necessary. Since Wingdings
>     >> etc. aren't in any alias rules, the only way they would be chosen as
>     fallback
>     >> (even in the presence of strong 'lang' elements) is if there was none other
>     >> available. And that's highly unlikely due to 49-sansserif.conf.
>     >>
>     >> After all, I don't recall any such bugreports for 2.8.0...
>     >>
>     >> Raimund
>     >>
>     >>
> 
>     --
>     behdad
>     http://behdad.org/
> 
> 
> 
> 
> -- 
> Akira TAGOH

-- 
behdad
http://behdad.org/