[OS X TeX] [lucida] Old style numerals with small caps

[Josep Maria Font]

Many thanks again to Bruno for his very detailed and wise leçons, which more than fulfil my needs. 


JMaF


> Forwarded message from Bruno Voisin:
> 
> Le 6 mai 2012 à 11:28, Josep Maria Font a écrit :
> 
>> It seems it is more difficult to solve the \oldstylenums issue than to just obtain old style numerals as default with small caps (which is what I was about). For the latter purpose, his advice:
>> 
>>> - To get oldstyle nums in small caps, it seems you need to use LY1 encoding, not T1. Namely, replace
>>> 
>>> 	\usepackage[T1]{fontenc}
>>> 
>>> by
>>> 
>>> 	\usepackage[LY1]{fontenc}
>>> 
>> 
>> has worked like a charm.
> 
> To solve these issues, it's needed to look more closely into the working of T1/TS1 (incl. textcomp) and LY1.
> 
> On one hand, T1 and TS1 encodings go together: T1 contains contains letters and numbers with focus on providing accented glyphs for most European languages, while TS1 contains the remaining non-alphabetic glyphs (trademark and copyright symbols, dagger and double-dagger, etc.). The fontenc package with option T1 defines accents etc. for the T1 encoding, and the textcomp package introduces the appropriate definitions (\textregistered etc.) for TS1 encoding.
> 
> On the other hand is LY1, which contains letters and numbers with less accented glyphs, plus the non-alphabetic glyphs all together in one metrics. The fontenc package with option LY1 introduces the appropriate definitions for LY1 encoding.
> 
> Now come the small caps and old-style numbers, for which the two approaches differ:
> 
> - For LY1, old-style numbers are the numbers of a separate expert small-caps font. Hence, selecting the small caps font yields automatically old-style numbers. This is the way the Type 1 Lucida fonts are packaged, with lbrsc the small caps font (and lbdsc the associated bold font), and this is also the way the associated LY1 metrics hlhrc8y is constructed.
> 
> - For T1/TS1, small caps and old-style numbers are totally separate concepts. Old-style numbers are no longer considered as the proper numbers of an expert small caps font, they are considered instead as exotic non-alphabetic glyphs relegated to the TS1 encoding for the standard (ie non small-caps) font. The small caps T1 metrics hlhrc8t for lbrsc borrows its numbers from the regular lbr font, while the old-style digits are relegated to the TS1 metrics hlhr8c for lbr.
> 
> As a result, the solutions to your problem differ:
> 
> - Based on T1/TS1: the packages to load are
> 
> 	\usepackage[T1]{fontenc}
> 	\usepackage{textcomp}
> 
> You will not get old-style digits in small caps; this is the deliberate way the T1/TS1 metrics for Lucida are designed, and I can think of no workaround. For old-style numbers, you get them automatically by using \oldstylenums in math mode, ie with $\oldstylenums{0123456789}$; to get them also in text, simply add
> 
> 	\DeclareEncodingSubset{TS1}{hlh}{1}
> 
> - Based on LY1: the only package to load is
> 
> 	\usepackage[LY1]{fontenc}
> 
> and this yields old-style numbers in small caps. For \oldstylenums to work, you need to copy the definition of \oldstylenums from latex.ltx and replace \rmdefault in it by hlcm, yielding
> 
> 	\makeatletter
> 	\def\oldstylenums#1{%
> 	   \begingroup
> 	    \spaceskip\fontdimen\tw@\font
> 	    \usefont{OML}{hlcm}{\f at series}{it}%
> 	    \mathgroup\symletters #1%
> 	   \endgroup
> 	}
> 	\makeatother
> 
> As a consequence, the old-style digits will be taken from the lbmo alternative italic math font.
> 
>> Will this change in encoding package have any other side effects ? The packages we work with with some fonts effects that I can think of are lucidabr and textcomp.
> 
> Using LY1 instead of T1/textcomp should normally have no side effect. In principle T1/textcomp provides more glyphs than LY1, but in practice I've never encountered any problem with LY1. You can still load textcomp on top of LY1, but I wouldn't advise it.
> 
> The differences are elsewhere: the LY1 metrics hlhr8y etc. are the original metrics designed by Y&Y using afm2tfm when packaging the fonts for TeX, while the T1/TS1 metrics were created much later by Walter Schmidt (I think) using fontinst. Hence line breaks etc. may differ when using one or the other.
> 
> Personally I have a preference for LY1, for which I find the line breaks more pleasing to the eye (in particular, less hyphenations). That is, of course, very subjective.
> 
> Finally here are modified versions of your example for LY1, T1/TS1 and also with the OpenType fonts. The modifications include examples for old-style numbers in italic and bold. Glyph charts for the T1, TS1 and LY1 metrics are also attached.
> 
> Hope this helps,
> 
> Bruno