fpprec and output precision



Hi all, 

As a follow up to something that was asked in May 2005 but not answered
(as far as I can tell).  fpprec doesn't seem to provide consistent
output, particularly with scientific notation and tex().  Please review
these examples of what I consider to be strange behavior.

(%i75) fpprec:4;
(%o75)                                 4
(%i76) tex(5.0011613);
$$5.001$$
(%o76)                               false
(%i77) tex(5.0011613e-5);
$$5.0011613 \times 10^{-5}$$
(%o77)                               false
(%i78) tex(5.0011613e5);
$$500116.$$
(%o78)                               false
(%i79) 5.0011613e5;
(%o79)                              500116.
(%i80) 5.0011613e-5;
(%o80)                             5.0012E-5
(%i81) 5.0011613;
(%o81)                               5.001

Is there a way to have consistent output, and by consistent I mean
report a specific number of significant figures in all output,
regardless of the input style?  Perhaps I am making assumptions about
fpprec, and there is something else available for this purpose?  Thanks
a bunch,

Pete


PS
Below is the previous question:
----------------------------------
Like this:
(%i2360) FPPREC:6;

(%o2360)                               6
(%i2361) tex(float(1/3));

$$0.3333$$
(%o2361)                             FALSE
(%i2362) tex(float(1/3*10^-24));

$$3.3333333333333335 \times 10^{-25}$$
(%o2362)                             FALSE
(%i2363) tex(float(1/3*10^-2));

$$0.0033$$
(%o2363)                             FALSE

The output of (%i2362) is not wanted. Why the setting
of FPPREC doesn't work when the value is too little?

However, 
(%i2364) float(1/3*10^-24);

(%o2364)                          3.33333E-25
It works without "tex"!

Is there any other way to make the "tex"ed value to be
shorter. It is for the convenience of reading.

Thanks for consideration.

YC Zou