[texhax] TeX Queries (2): Artificial Break

[Paul Stanley]

 > How can the declared width be less than the real width?
[snip]

The "declared width" is independent of the shape of the glyph.  It 
also takes side bearings into account.  Its sole purpose is to tell 
where to place the next glyph.  Thus it's also called "advance 
width". BTW, it refers to the baseline, not to a 
rectangle.  Therefore, if a font is slanted, extra space is needed 
between the last slanted glyph and the next upright glyph.

There is no canvas limiting the size of a glyph because a glyph 
description is a mathematical construct rather than something 
physical.  Glyph descriptions can even have negative coordinates 
(required for descenders, at least).  The only limitation is that 
coordinate pairs in Type1 and TrueType fonts are integer numbers with 
a limited range.

I suppose on the computer screen the painting-on-a-canvas analogy 
would translate as a rectangular area containing the glyph.  I was 
under the impression that the declared (or advanced)width measured 
the distance between the two sides of this rectangle.  As it only 
measures a portion of the line occupied by the glyph at the baseline 
(including the side bearings) the canvas analogy is a bit misleading. 
Actually, more than just a bit.

Are descenders glyphs that *descend* below the baseline?

You are wondering why the physical width of an "f" exceeds advance 
width in some fonts.  The reason is that if you derive the advance 
width from the bounding box of the glyph shape, you need zillions of 
kerning pairs.  An "f" is usually followed by another lowercase 
character.  If the latter is an "i" or "l", ligatures are used 
anyway. In all other cases it's fine that the bow of the "f" 
surmounts the following glyph.  If an "f" is followed by an uppercase 
letter, kerning has to be applied, of course.

Just to confirm, does kerning concerns both the adding and 
subtracting from the advanced width of the glyph?

Could you please elaborate why many more kerning pairs would be 
needed?  I'm not sure if i understand the problem.

Fonts are usually optimized to work with the least amount of kerning 
pairs.  Font designers also have to take into account that Microsoft 
refuses to support kerning in Type1 fonts at all, deliberately.

yes, kerning pairs can be an accessibility pain in the ... They are 
either invisible to the screen reader or show up as strange control 
characters (which comes to the same thing).

Many thanks,
Paul