Re: [Freefont-bugs] OS/2 tables

[Josef W. Segur]

On 7 May 03, at 18:48, Primoz wrote:

> On Tue, 6 May 2003, Josef W. Segur wrote:
> 
> > > What could then be the reason that Internet Explorer doesn't offer
> > > FreeMono as a Greek font? The CodePageRange table for FreeMono reads
> > > "800000bf 00000000", which I interpret as covering Latin 1, Latin 2,
> > > Cyrillic, Greek, Turkish, Hebrew, Windows Baltic and Symbol Character Set
> > > pages.
> >
> > The problem may be caused by that "Symbol Character Set" bit being
> > set. I do run Win95 but don't have Internet Explorer, so I can't
> > actually check.
> > That flag bit is meant for fonts with arbitrary non-Unicode glyphs,
> > such as WingDings and WebDings. I suggest clearing it for
> > FreeMono.ttf, FreeSans.ttf, and FreeSerif.ttf. The italic, bold, and
> > other variants are OK already.
> 
> Is it? I thought that the bit should be set if the fonts contains glyphs
> from the set of Microsoft "symbol" font. At least both Symbol and
> Wingdings fonts have this bit set, while, say, Times and Tahoma do not:
> 
> symbol.ttf               CodePageRange=80000000 00000000
> tahoma.ttf               CodePageRange=2000009f 00000000
> tahomabd.ttf             CodePageRange=2000009f 00000000
> times.ttf                CodePageRange=400001ff ffff0000
> timesbd.ttf              CodePageRange=400001ff ffff0000
> timesbi.ttf              CodePageRange=400001bf dff70000
> timesi.ttf               CodePageRange=400001bf dff70000
> wingding.ttf             CodePageRange=80000000 00000000
> 
> FreeMono.ttf, FreeSans.ttf, and FreeSerif.ttf indeed contain "symbol"
> characters; the italic, bold and bold italic variants currently don't (and
> there is little point having, say, slanted box-drawing characters :) So I
> believe these fonts should have the "Symbol Character Set" bit set.

Yes, but Microsoft's Symbol, Wingdings, and Webdings "symbol" characters
are accessed in the 0 to 0xFF range by old programs or the U+F000 to U+F0FF
part of the Unicode Private Use Area. Those fonts also use a 'cmap'
subtable for platform of 3 and encoding ID of 0, rather than the encoding
ID of 1 which is used for textual fonts. All the "symbol" characters in
the Free UCS fonts are Unicode code points like any other characters.

I believe no font supplied by Microsoft has the "Symbol character set"
flag bit combined with normal text ranges. I guess it likely that Internet
Explorer may assume that flag bit simply means a non-text font.

> My concern is that Microsoft have set numerous other bits as well (mostly
> referring to the MS-DOS code pages) which are not set in the Free fonts,
> and this might be the culprit.
> 
> It bothers me, though, why IE should rely on the old MS-DOS code pages.

It does seem unlikely. Some of the fonts which were on the Microsoft
"Core fonts for the web" page before it was withdrawn don't have the
MS-DOS code page coverage, but do have the Windows 1253 for Greek.
Specifically: Comic Sans MS, Georgia, Trebuchet MS, and Verdana.

-- 
                                                 Joe