From 2.4 to 2.5

- Some overflow errors that were previously undetected are now reported.
- The command for the next prime is NF, for the previous prime is PF.
  If the argument is negative NF returns 2 and PF "no primes under 2".

From 2.3 to 2.4

- New command RF to find the previous prime.
- New command IP to check the primality.
- Erf(x) is negative if x is negative.
- Revised and adapted for fpc v. 2.
- Code optimized for size and speed.

From 2.2 to 2.3

- Fixed some bugs of the factorization routines.

From 2.0 to 2.2

- Factorization of integers up to 50 figures.
- Some optimizations.
- Fixed some small bugs.

From 1.8 to 2.0

- Factorization of integers up to 48 figures.
- The 'next prime' command pf now can operate on numbers up to 250 figures.
- Improved the speed of factorization and primality tests.
- Fixed some bugs in the evaluation of bitwise operations, due to leading
  zeros sometimes present in the buffers.

From 1.7 to 1.8

- Factorization of integers up to 45 figures.
- A new command PF (next prime) to find the smallest prime number
  greater than the result integer part (up to 45 figures).

From 1.6 to 1.7

- Extensively revised and tested. Some optimizations. Stable version,
  recompiled by fpc 1.0.6.
- The author's email and homepage addresses are changed.

From 1.5 to 1.6

- Factorization of integers of up to 19 figures.
- The list of file names displayed by the 'files' command is sorted
  alphabetically.

From 1.4 to 1.5

- Some optimizations.
- Fixed a bug in the evaluation of some expressions involving power
  calculation (thanks to Jan <janm@mit.edu> for pointing it out).

From 1.3 to 1.4

- A new command DF to decompose in factors the result integer part (up to
  a maximum of 17 digits).
- Improved the code, also to increase the speed.
- New faster logarithm routines.
- New faster division routine.
- Improved the square root algorithm.

From 1.2 to 1.3

- The maximum allowed number of digits is now 295 (four lines).
- Some routines have been improved and rewritten in assembly to decrease
  the execution times.
- When possible, the functions accept arguments having any number of digits.
- Fixed the wrong results +/-1 in some hyperbolic cotangent overflow.
- Fixed some small bugs.
