3 things



Hi,

1.  I've successfully built and installed rc4 on an i686-linux-gnu box
    with clisp-2.29

2.  The testsuite worked, with one minor exception in rtest15:

/* ********************** Problem 4. *************** */ 
%Input is
		     1
INTEGRATE(SQRT(- 2 + - + x), x, 0, 1)
		     x


The result is
  4
- -
  3

Using ratsimp
Using ratsimp
This differed from the expected result:
4
-
3


	Actually -4/3 is also correct if you use the negative branch
	of the SQRT function, although I don't know if this is
	intended.

3. It seems that TAYLOR is somewhat more clever than the documentation
   would indicate.  Consider:

(C1) f(x):=1+1/x+exp(-x)*(1+1/x);
				   1		     1
(D1) 		       f(x) := 1 + - + EXP(- x) (1 + -)
				   x		     x
(C2) taylor(f(x),x,inf,2);
		     1		      1		   - x
(D2)/T/ 	 1 + - + . . . + (1 + - + . . .) %E    + . . .
		     x		      x

This is neither a Taylor or Laurent series, but is very useful in
assymptotic analysis of expressions.  I don't know how robust this
behavior is, for example:

(C1) taylor(sin(x)/(1+x),x,inf,3);
TAYLOR encountered an essential singularity in:
SIN(x)
 -- an error.  Quitting.  To debug this try DEBUGMODE(TRUE);)

Rather than giving an error, it might be nice to have the option of
keeping the sin (or any other essentially singular function, like
exp(-x)) explicit and taylor expanding the rest; i.e.,

		     sin(x)/x  -sin(x)/x**2 + ...


In any event this behavior is undocumented.

David