[XeTeX] lineskip is inconsistent

Jonathan Kew jonathan_kew at sil.org
Sun Feb 19 19:53:36 CET 2006


On 19 Feb 2006, at 5:54 pm, Steffen Wolfrum wrote:

> Hi,
>
> coming from ConTeXt I am still experimenting with different  
> features of Xe(Con)TeX(t).
>
> One thing I have found today makes me wonder:
> Even though the lineskip is defined as 14 pt it differs totally  
> from font to font. And only Times and Helvetica are on the grid.
>

Does ConTeXt redefine \lineskip to mean something other than its  
built-in TeX meaning? If not, then I think you're using the wrong  
parameter.

In TeX, the "normal" spacing of text lines is given by the parameter  
\baselineskip; successive lines of text will have their baselines  
exactly this far apart, *provided* there aren't any particularly tall/ 
deep characters.

If TeX determines that the bottom of one line and the top of the next  
would come closer than \lineskiplimit (default in Plain TeX: 0pt),  
then it abandons \baselineskip and instead inserts \lineskip  
*between* the adjacent lines (not their baselines, but between the  
bottom of one and the top of the next).

An additional complication with XeTeX is that the height and depth of  
each line is computed not on the basis of the actual shapes of the  
characters, but is based on the "ascent" and "descent" metrics  
provided by the font. In some cases, these metrics include some  
whitespace above/below the normal extent of the glyphs, and so the  
lines will logically "touch" (and \lineskip will kick in) even before  
the actual visible glyphs would have touched.

When using such fonts, it can be helpful to set \lineskiplimit to a  
*negative* value, to allow a certain amount of "overlap" between the  
logical extent of adjacent lines, before the normal \baselineskip  
spacing is abandoned and \lineskip used instead.

OK, that's the Plain TeX picture (actually, these are built-in TeX  
parameters, not even Plain definitions). I don't know, however, to  
what extent ConTeXt overrides this stuff. I'd be a bit surprised,  
though, if it actually redefined these command names to mean  
something quite different.

JK




More information about the XeTeX mailing list