The package *C_CALC* was written by Hans Boehm, Vernon Lee and Alan J. Demers in 1988-9 using the programming language *Russell*.
It provides a calculator which does precise arithmetic on real numbers using functions to represent the numbers.
The package is available from Xerox or from Rice University.
It may well be defunct now, but Hans Boehm, now at HP, has implemented a constructive reals calculator in Java, which runs on the web.
According to Boehm, in this implementation:

A constructive real numberxis represented abstractly as a functionf, such that_{x}fproduces a scaled integer approximation to_{x}(n)x, with an error of strictly less than 2^{n}.

More precisely: |f(_{x}n) * 2^{n}- x| < 2^{n}

Sheng Liang and Rajiv Mirani wrote a Haskell implementation of Boehm's constructive reals. A version that works with Haskell 1.3 can be found at yale: symalg

David Plume has recently implemented exact real arithmetic in Haskell and his calculator performing exact real arithmetic is available on the web for all to try. He has some web pages on this project which point you to his project report, to the online calculator and also to a page of relevant links.

Keith Briggs of BTExaCT has an implementation of exact real arithmetic in C++.

Exact Real Computation in ProofPower