What is wrong in multiplication of matrices

Hello,

I try to multiply matrices by matrix congruence. First introduce the following 
matrices A, C, P, Q:

declare([A,C,P,Q], nonscalar);

and the following block matrices: 
m: matrix([P^^(-1), 0, 0], [0,1,0],[0,0,1]);
M: matrix([-P, transpose(A), transpose(C)], [A, -P^^(-1), 0], [C, 
0, -Q^^(-1)]);

Then multiply it as follows 
r1: m . M . m;
and I got wrong result. Why, because on position r1[1,2] is 
(I obtain it by grind(r1[1,1])$)
'transpose(A)*P^^(-1)

but should be
P^^(-1) . transpose(A)

I see that in the multiplication were used operator * but I declare that 
A,C,P,Q are nonscalars and it should use the . operator. Why it is not done?
Or how I can multiply matrices as block matrices? I want only operate on 
symbolic matrices as A,C,P,Q not as real i.e.
A: matrix([1,2],[3,4])  <---- not on such matrices, where are values

I want to work on symbolic matrices. Is maybe any chance to introduce in 
future versions of maxima something such as: 

declare([A,C,P,Q], symbolic_matrix) 

to tell maxima that I operate on matrices in symbolic way?

I also want to ask how to simplify r1[1,1]
-P*(P^^(-1))^2

it should simplify to 
-P

but there is * not . so it couldn't simplify.

When I write with . it also do not simplify, why?
-P . (P^^(-1))^2
I got:
          <- 1> 2
  - P . (P     )
 
But when I write
-P . P^^(-1);
then I got correct results:
  - 1

Thank you in advance.
Zbigniew