MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
References: <20091028141650.GA3488@irwin.vpn.uni-freiburg.de>
            <19176.37836.964673.628156@morse.mittelbach-online.de>
            <20091028231433.GA4001@irwin.uni-mannheim.de>
            <19177.18303.198573.261636@morse.mittelbach-online.de>
            <20091029135720.GA4353@irwin.vpn.uni-freiburg.de>
            <Pine.LNX.4.61.0911031941100.8953@linux.bop.eps.gda.pl>
            <20091103230212.GA22544@irwin.uni-mannheim.de>
Message-ID:  <19185.48121.439810.144092@morse.mittelbach-online.de>
Date:         Wed, 4 Nov 2009 18:38:01 +0100
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
Sender: Mailing list for the LaTeX3 project <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
From: Frank Mittelbach <frank.mittelbach@LATEX-PROJECT.ORG>
Subject: Re: Encoding subset for lm* and tg*
To: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE
In-Reply-To:  <20091103230212.GA22544@irwin.uni-mannheim.de>
Precedence: list
Status: R

Heiko Oberdiek writes:
 > On Tue, Nov 03, 2009 at 08:14:10PM +0100, Boguslaw Jackowski wrote:
 > 
 > > HO> Result:
 > > HO> * lmtt misses
 > > HO>  * \textperthousand
 > > HO>  * \texttrademark
 > > HO>  * \textservicemark
 > > 
 > > This is on purpose. Squashed (because of monospacing) glyphs do not fit 
 > > to the overall CM look-and-feel, moreover, the typesetting of trademark 
 > > and servicemark is trivial not only in TeX, but even in Word, and the 
 > > glyph perthousandzero (slot 24 in the Cork aka ec encoding) can be used 
 > > to obtain perthousand, permyriad, etc.

I can understand the designers feeling here, but from a document perspective
point I can't quite agree with Jacko. encoding something or not is not really
a question of whether something can be easily faked or not, it is simply a
question of how can reliably be a font replaced with some other font without
loosing information. The whole NFSS system in LaTeX was build to get around
the font encoding peculiarities of Don's original CM fonts where gyphs changed
from font to font turning dollars into pounds as the worst example

 > Summary: the following fonts also have subencoding 0:
 > * Latin Modern: lmdh, lmss, lmssq, lmvtt
 > * TeX Gyre: qag, qbk, qcr, qcs, qpl, qtm, qzc, qhvc

good, thanks, so those can be added to textcomp

 > Open issue: lmtt misses \textperthousand, \texttrademark, \textservicemark
 > I agree, they would look ugly squeezed into the width of the other glyphs.
 > But what to do with textcomp's subencoding?
 > * virtual font that provides the missing glyphs
 > * new subencoding
 > Frank?

I really don't think there is a point in trying to invent yet another
subencoding with all its disadvantages. That means, there are two
possibilities

 - Jacko forgets the designers qualms :-) and accepts that a \texttrademark
   symbol on a monospaced font looks like one that is from a mono-spaced font
   -> then we get a full set of glyphs and can use 0

 - else we use whatever fits which is going to be subencoding 2 even if that
   means a good number of existing symbol become unavailable (unless maually
   re-added)

frank