In message <c410142a4d.james(a)ix.strcprstskrzkrk.co.uk>
James Bursa <james(a)semichrome.net> wrote:
I'm looking at rewriting the font-related code in NetSurf,
because it's a bit
of a mess at the moment. Reinstating transliteration is definitely part of
this. My ideas are as follows:
1. Get rid of font_set and font_data, and make the nsfont_ calls take a
font name, size, and style directly. The font code can keep an
internal cache of font handles.
2. Implement font substitution, so that missing glyphs are taken from
different font if available. I have some working code for this with
the RISC OS 5 font manager so far. For the RO 3-4 font manager, we will
probably have to scan the Encoding files ourselves and build a mapping
of Unicode -> character code. (Or we could make the Unicode font manager
a requirement for proper Unicode rendering, but it looks like it will
take many more years for everyone to have it.)
3. Transliterate if glyphs are not in any font.
The aim is that glyphs from all installed fonts are available as soon
they are installed, without needing special tables like the UFont resource,
as long as they have a correct Encoding file.
You see any way to specify the encoding of a set of non-large character
set RISC OS fonts (if I can call them like that) outside the font file
and/or to combine such a set of fonts together ? I.e. have the UFont
advantage (but not its disadvantage) ?
Currently PS printing & fonts is also a very weak point (with and without
the Unicode capable font manager), any idea if this is/can be addressed
as well ?
John Tytgat, in his comfy chair at home BASS
John.Tytgat(a)aaug.net ARM powered, RISC OS driven