[lltx] Microtypography... now the user asks questions.

[Élie Roux]

2010/2/15 zappathustra <zappathustra at free.fr>:
>
> Thank your very much for all your work here (with special thanks to
> Elie who answered all my mails and set everything in motion, even
> though he was asking something from me in the first place...).

I actually wrote absolutely no line of code for this, so you should
thank Hans (who is in copy) and Khaled for that!

> I've tested font expansion with otf fonts, and I'm very happy that it
> works. Now, I have some questions...
>
> - First, how does font expansion enter TeX's paragraph building
> algorithm? Does it play a role in badness calculation or does it come
> for free? (\tracingparagraphs doesn't give any information). I.e., is
> there a trade-off between space manipulation and character
> manipulation? And are there restrictions on adjacent lines, e.g. a
> fully stretched line can't be next to a fully shrunken one (talking
> about glyph expansion, of course)?

I think Hans is by far the best person to ask!

> - Second, what about letterspacing (once again for better
> implementation)? I know the microtypography package is supposed to do
> that, but I've not had the courage to look at the code yet, and I'm
> not using LaTeX anyway...
> - Third, I understand that the expansion factor is given by the
> following table, which I manipulate with a little bit of \directlua:
>
> fonts.expansions.setups['default'] = {
>
>    stretch = 2, shrink = 2, step = .5, factor = 1,
>
>    [byte('A')] = 0.5, [byte('B')] = 0.7, [byte('C')] = 0.7,
> [byte('D')] = 0.5, [byte('E')] = 0.7,
>    [byte('F')] = 0.7, [byte('G')] = 0.5, [byte('H')] = 0.7,
> [byte('K')] = 0.7, [byte('M')] = 0.7,
>    [byte('N')] = 0.7, [byte('O')] = 0.5, [byte('P')] = 0.7,
> [byte('Q')] = 0.5, [byte('R')] = 0.7,
>    [byte('S')] = 0.7, [byte('U')] = 0.7, [byte('W')] = 0.7,
> [byte('Z')] = 0.7,
>    [byte('a')] = 0.7, [byte('b')] = 0.7, [byte('c')] = 0.7,
> [byte('d')] = 0.7, [byte('e')] = 0.7,
>    [byte('g')] = 0.7, [byte('h')] = 0.7, [byte('k')] = 0.7,
> [byte('m')] = 0.7, [byte('n')] = 0.7,
>    [byte('o')] = 0.7, [byte('p')] = 0.7, [byte('q')] = 0.7,
> [byte('s')] = 0.7, [byte('u')] = 0.7,
>    [byte('w')] = 0.7, [byte('z')] = 0.7,
>    [byte('2')] = 0.7, [byte('3')] = 0.7, [byte('6')] = 0.7,
> [byte('8')] = 0.7, [byte('9')] = 0.7,
> }
>
> and I'd like to understand it. I suppose stretch and shrink are
> percentages, and step is the increment. But what about factor?
> Setting, say, factor=10, is a shorthand for stretch and shrink set to
> 20 and step to 5, or is there something else? Finally, what do the
> values apparently associated with characters represent? I've set
>
> [byte('a')] = 5
>
> (a probably stupid value) and it produced overfull boxes absent
> otherwise.

Once again, I think Hans will be the best person to answer you questions...

> Errrr, as an aside, I've already told this to Elie, but apparently
> something went wrong. I use the latest version of LuaTeX (for
> Windows), i.e. v.0.5, and setting \hsize still modify the page width
> (if \pagewidth isn't specified).

Can you provide a minimal plainTeX example that gives different
results under pdfTeX and LuaTeX?

Thank you,
-- 
Elie