#2 Floats with big exponents converted to binary data

[Jan Adámek]

When a big float number (with exponnt over about 32)
is passed to teng, it is converted to about 32 digits
of the original float with some binary data (looks
like a part of libteng's data segment) appended
instead of remaining digits.

This bug was found in Python module, but probably
exists also in PHP extension and C++ library.