Confused writing a function and its result - depending on reevaluation of the function

I have written the follwoing function which basically does some function
discussion:

kurvendiskussion(func) :=
block(
[ab1,ab2,ab3,ab4,N,E,W,l,erg,abl,x,f,xi],
f: ratsimp(func),
f(x):=''f,
ab1:diff(f(x),x),
ab2:diff(f(x),x,2),
ab3:diff(f(x),x,3),
ab4:diff(f(x),x,4),
l:realroots(f(x)=0),
N: makelist([xi:float(ev(x,l[i])), float(f(xi))], i , 1, length(l)),
l:realroots(ab1=0),
E: makelist([xi:float(ev(x,l[i])), float(f(xi))], i , 1, length(l)),
l:realroots(ab2=0),
W: makelist([xi:float(ev(x,l[i])), float(f(xi))], i , 1, length(l)),
abl: [ab1,ab2,ab3,ab4],
erg:
matrix(["Funktion","Ableitung","Nullstellen","Extremwerte","Wendepunkt"],
[func, transpose(abl), transpose(N),transpose(E),transpose(W)]),
return (erg)
)$


I call it in this way:

f: (x-3)*(x+4)*(x-7)/20;
ergebnis: kurvendiskussion(f)$
disp("","Aufgabe",ergebnis)$

When I call it first I'll get:

"Aufgabe"
matrix(["Funktion","Ableitung","Nullstellen","Extremwerte","Wendepunkt"],[((
x-7)*(x-3)*(x+4))/20,matrix([(3*x^2-12*x-19)/20],[(6*x-12)/20],[3/10],[0]),m
atrix([[-4.0,0.05*(x^3-6.0*x^2-19.0*x+84.0)]],[[3.0,0.05*(x^3-6.0*x^2-19.0*x
+84.0)]],[[7.0,0.05*(x^3-6.0*x^2-19.0*x+84.0)]]),matrix([[-1.214550226926804
,0.05*(x^3-6.0*x^2-19.0*x+84.0)]],[[5.214550226926804,0.05*(x^3-6.0*x^2-19.0
*x+84.0)]]),matrix([[2.0,0.05*(x^3-6.0*x^2-19.0*x+84.0)]])])

Which I do not really want as I want the real values. Now I evaluate the
above function definition again and rerun the previous call. Now I get:
"Aufgabe"
matrix(["Funktion","Ableitung","Nullstellen","Extremwerte","Wendepunkt"],[((
x-7)*(x-3)*(x+4))/20,matrix([((x-3)*(x+4))/20+((x-7)*(x+4))/20+((x-7)*(x-3))
/20],[(x+4)/10+(x-3)/10+(x-7)/10],[3/10],[0]),matrix([[-4.0,0.0]],[[3.0,0.0]
],[[7.0,0.0]]),matrix([[-1.214550226926804,4.821701928786462]],[[5.214550226
926804,-1.821701928786462]]),matrix([[2.0,1.5]])])
That's actually what I like...

Any idea what I do wrong?

Attached also a file in wxm Format which contains the function definitions.

Thanks for help on this.

BTW: Does anyone know a good description about Maxima programming?

Klemens



-------------- next part --------------
A non-text attachment was scrubbed...
Name: kurvendiskussion.wxm
Type: application/octet-stream
Size: 1279 bytes
Desc: not available
URL: <http://www.math.utexas.edu/pipermail/maxima/attachments/20110426/55bffe34/attachment.obj>;