[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: newmath.dtx patches

>> - Should all layouts with nonstandard size scalings set 
>> \delimiterfactor, so that \big remains \big and doesn't
>> suddenly become \Big when switching the math layout?

> Could you provide the necessary \delimiterfactor values for that ?  
> We might even need more than one value for Lucida, depending on the
> smallscale/noscale/scale option, but I haven't done any experiments
> yet.

For MathTime, I derived the factor by hand.  For the other layouts,
I'll have to see.  The algorith was something like

\setbox0=\hbox{$\displaystyle x^2$}

Take the maximum of (ht0 - axis_height) or (axis_height - dp0).
Then choose a value so that (2 * max) * \delimiterfactor < 12.0pt.
(Assuming that \big delimiters are 12pt high)

Another test case was

\setbox0=\hbox{$\displaystyle \frac{\displaystyle x^2}{2}$}

Take the maximum of (ht0 - axis_height) or (axis_height - dp0).
Here, choose a value so that (2 * max) * \delimiterfactor < 24.0pt.
(Assuming that \bigg delimiters are 24pt high)

The axis_height comes from \fontdimen8\textfont2, I believe.

If you want an even stricter criterion than the above, repeat
the procdedure with $x^{(0)}$ instead of $x^2$.

Finally, choose the smallest version of \delimiterfactor required 
to fulfill the criteria in all test cases.  For CM the default
is 901, for the other layouts, it should probably range somewhere
between 900 and 850.

Cheers, Ulrik.

P.S.  I recall that I did another test with my Mma implementation
of the encodings, in which I used the big delimiters from Math2
as the default size, ignoring the delimiters from Times or Math1.
I believe I had to reduce \delimiterfactor to 870 to ensure 
that $\left[ f \right]$ didn't trigger a bigger size.