[texhax] Alignments and Knuth's uncertainty principle ...

[Donald Arseneau]

"Philip TAYLOR (Ret'd)" <P.Taylor at Rhul.Ac.Uk> writes:

> In a horizontal alignment, if (for a given column) no row contains
> an entry solely for that column, but instead invariably coalesces
> that column either to the left, or to the right, or both, using
> \span, how does TeX compute the width of that column.  

This situation does lead to surprising behavior that looks
buggy at first.  TeX gives the smallest feasible width to 
that unconstrained column, and that width is likely 
*negative*.  If you give a blank row somewhere with all
independent columns so as to enforce a minimum zero width
per column, then the resulting layout will look a lot more
sensible.

> a similar alignment, there are a number of such columns that
> abut each other, how does TeX distribute the space amongst them ?

You may get columns that have no unique identity, and they
will be dropped -- combined with an adjacent column.
Each under-defined but independent column will be minimized
I'm not sure how this applies to your exact case; you could 
post or send an example to clarify.


-- 
Donald Arseneau                          asnd at triumf.ca