Hi Ted,
> 'eqn' does not put the "z" in, so I think between us we have uncovered
> a deficiency in 'eqn'. Quite where in the 'eqn' code this lurks is not
> obvious (to me at the moment).
http://git.savannah.gnu.org/cgit/groff.git/tree/src/preproc/eqn/box.cpp#n268
void set_script_size()
{
if (minimum_size < 0)
minimum_size = 0;
if (script_size_reduction >= 0)
printf(".ps \\n[.s]-%d>?%d\n", script_size_reduction, minimum_size);
else
printf(".ps (u;\\n[.ps]*7+5/10>?%d)\n", minimum_size);
}
> Or maybe, as you suggest, it is only there because of the "/10" and
> 'eqn' has no intention of setting a lower limit to point-size in
> superscripts (though I think that could be a flaw in itself, from the
> typesetting point of view).
No, I think it intends to. The first 5 is just to aid rounding before
the /10. For the second 5 to be connected to the /10 it would suggest
50 had some significance. Instead, I think it's meant to be 5pts
minimum. minimum_size above is points one time and scaled points the
other. (I couldn't get git-on-the-web to show me the long-term history
of that bit of source.)
Cheers, Ralph.
