Re: TTF2TFM
- From: vojta@xxxxxxxxxxxxxxxxx (Paul Vojta)
- Date: Tue, 20 Dec 2005 08:54:22 +0000 (UTC)
In article <dnvalm$1ba$1@xxxxxxxxxxxxxxxxxxxxx>,
Michael Zedler <Michael.Zedler@xxxxxx> wrote:
>Paul Vojta schrieb:
>> In article <dnu2op$ari$1@xxxxxxxxxxxxxxxxxxxxx>,
>> Michael Zedler <Michael.Zedler@xxxxxx> wrote:
>>> I never quite understood what's the advantage of ttf2tfm over
>>> ttf2afm+fontinst.
>>
>> Well, TeX uses numbers and not names to refer to glyphs.
>> Ttf also uses numbers and not names to refer to glyphs.
>
>Please have a look at what role encodings play, for TTF t1-wgl4.enc is
>of particular interest.
OK, let's look at t1-wgl4.enc. First of all, there's the line:
/T1Encoding [ % now 256 chars follow
Now ttf fonts are 16-bit (at least), and Omega is 16-bit (at least), so
I don't see why one would want to thread font access in Omega through
an 8-bit needle.
Next there's the line:
/grave /acute /circumflex /tilde
In a PostScript font, the strings "grave", "acute", etc. are the procedure
names for the glyph-drawing routines in the font. They are the primary
means for addressing the glyph routines.
In TrueType, however, glyph names (as I understand it) are optional.
>> If you use ttf2afm, you have to assign names to your glyphs so that they
>> can be removed by afm2tfm. Sounds kind of pointless to me.
>
>I referred to fontinst, not afm2tfm. The former is quite useful to
>generate T1/TS1 encoded fonts as well as to correct kerning flaws.
>For the user who is mainly interested in a quick installation the most
>obvious advantage is the automatic generation of fd+map files.
OK, so the names are removed by fontinst instead of by afm2tfm.
>> I can understand why ttf2afm is useful (i.e., for pdftex, since its
>> font model is based on PostScript), but not for dvi creation.
>
>This is not correct, please have a look at how your dvi
>viewer/dvips/pdftex treat fonts. Neither is it related to the question
>ttf2tfm vs. ttf2afm+fontinst.
What exactly is not correct?
My dvi viewer (which is "mine" perhaps in a greater sense than you
anticipated) is xdvi (non-k). It is set up to treat ttf fonts via
mktexpk, which calls ttf2pk (a program related to ttf2tfm). It does not
use tfm files directly.
dvips (at our site) handles ttf fonts in exactly the same way.
Look at another thread currently running on comp.text.tex to see how
dvips handles (or doesn't handle) ttf fonts.
Yes, pdftex does use PostScript-style encoding files for ttf fonts,
but I don't know any reasons for doing so other than for uniformity
with handling of PostScript fonts.
The OP (in a subsequent message) mentioned that s/he wanted to add
OpenType (hence ttf) support natively in Omega. Therefore, my answer
above is framed in terms of what a good design would be, rather than
what is in place right now.
I have no objections to encoding files per se, but enc files in the
style of t1-wgl4.enc were designed for PostScript fonts, and are not
well suited for use with ttf fonts. If one uses encoding files with
ttf fonts at all, they should be in a ttf-specific format. However,
I don't see a need for them, since TrueType fonts already contain an
encoding. If you don't like the encoding, then use a virtual font.
Of course, the encoding is 16-32 bits wide, but Omega can handle that.
--Paul Vojta, vojta@xxxxxxxxxxxxxxxxx
.
- Follow-Ups:
- Re: TTF2TFM
- From: Michael Zedler
- Re: TTF2TFM
- References:
- TTF2TFM
- From: gul791
- Re: TTF2TFM
- From: Michael Zedler
- Re: TTF2TFM
- From: Paul Vojta
- Re: TTF2TFM
- From: Michael Zedler
- TTF2TFM
- Prev by Date: eps über Kopfzeilenlinie
- Next by Date: Re: eps über Kopfzeilenlinie
- Previous by thread: Re: TTF2TFM
- Next by thread: Re: TTF2TFM
- Index(es):
Relevant Pages
|