[clisp-list] writing FP numbers

[Robert B. <br...@go...>]

I would like to write a value of type single-float into an array of 8-bit
bytes using the binary format that C code would use.  Something similar to:

  float foobar;
  foobar = 3.14159;
  memcpy(buffer, &foobar, 4);

I can decode a single-float with integer-decode-float in order to get the
correct bits, but things get a little bit tricky when the float is
denormalized.  Also, I'm unsure how integer-decode-float handles NaNs.

Is there a non-ANSI internal clisp function that does what I want quickly
and reliably?  For example, SBCL has sb-kernel:single-float-bits, which
returns an integer containing the bit representation of its floating point
argument.

Also, given the raw bits for a floating point number, what clisp function
can be used to construct a single-float or double-float?

            bob