RE: [ft] Font family/style name issue (for msgothic shipped with vista).

[Rajeev Pahuja]

Thanks Adam for clarification.
This is a bug in freetype, i checked current CVS version also, there is
no change w.r.t. this. 

One of the possible fix for this in file sfobj.c, method
tt_face_get_name() could be as follows

...
rec = face->name_table.names + found_win;
switch ( rec->encodingID )
{
 case TT_MS_ID_UNICODE_CS:
 case TT_MS_ID_SYMBOL_CS:
 case TT_MS_ID_UCS_4:   //should be renamed
     convert = tt_name_entry_ascii_from_utf16;
     break;

 //case TT_MS_ID_UCS_4:
  //convert = tt_name_entry_ascii_from_ucs4;
// break;

 default:
;
}
...  


Thanks
Rajeev


-----Original Message-----
From: Adam Twardoch [mailto:address@hidden 
Sent: Thursday, November 30, 2006 1:13 PM
To: Werner LEMBERG
Cc: Rajeev Pahuja; address@hidden
Subject: Re: [ft] Font family/style name issue (for msgothic shipped
with vista).


>> is being treated as UCS4.  Where as actual string in this font is
utf16.
>>
>> Is "Unicode full repertoire (as per OTF spec.)"  means UCS4 ?
>>     

No. UTF-16 can represent the full Unicode repertoire (up to 1,114,111). 
Non-BMP codepoints are represented using surrogate pairs.

UCS-4, UTF-7 and UTF-8 all can also represent the full Unicode range. 
The only transformation form that cannot is UCS-2, which is essentially 
UTF-16 minus the surrogates, and is almost never used. Codepoints beyond

1,114,111 are not and never will be encoded in Unicode.

In OpenType fonts, all Unicode strings are encoded using UTF-16BE (i.e. 
the Motorola big-endian flavor).

-- 

Adam Twardoch
| Language Typography Unicode Fonts OpenType
| twardoch.com | silesian.com | fontlab.net