• Li Jie <cpunion gmail.com> Apr 13 2006
• Thomas Kuehne <thomas-dloop kuehne.cn> Apr 16 2006
• Li Jie <cpunion gmail.com> Apr 16 2006

Li Jie <cpunion gmail.com> writes:

I write a matrix class:

# class Matrix(T, int R, int C) {
#   //...
# }

How to write the opMul? It uses like this:

# Matrix!(int, 3, 4) m1;
# Matrix!(int, 4, 8) m2;
# Matrix!(int, 3, 8) m3 = m1 * m2;

Thanks,

- Li Jie

Thomas Kuehne <thomas-dloop kuehne.cn> writes:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Li Jie schrieb am 2006-04-14:
 I write a matrix class:

 # class Matrix(T, int R, int C) {
 #   //...
 # }

 How to write the opMul? It uses like this:

 # Matrix!(int, 3, 4) m1;
 # Matrix!(int, 4, 8) m2;
 # Matrix!(int, 3, 8) m3 = m1 * m2;


I think that isn't (yet) possible.

If there are many different matrix sizes you may try:

# class Matrix(T){
#     this(T[][] data){
#         // ...
#     }
# 
#     typeof(this) opMul(typeof(this) arg){
#         typeof(this) result;
#         // ...
#         return result;
#     }
# 
#     Matrix!(T) opDiv(Matrix!(T) arg){
#         Matrix!(T) result;
#         // ...
#         return result;
#     }
# }
# 
# int main(){
#     Matrix!(int) a;
#     Matrix!(int) b;
#     // ...
#     auto c = a * b;
#     Matrix!(int) d = a / b;
# 
#     return 0;
# }


If you only deal with a few matrix sizes you may try:

# template opMul(int R2, int C2){
#     Matrix!(T, R, C2) opMul(Matrix!(T, R2, C2) arg){
#         Matrix!(T, R, C2) result;
#         // ...
#         return result; 
#     }
# }
# 
# class Matrix(T, int R, int C){
#     // ...
# 
#     mixin opMul!(4, 8);
#     // and other mixins
# }
# 
# int main(){
#     Matrix!(int, 3, 4) a;
#     Matrix!(int, 4, 8) b;
#     // ...
#     auto c = a * b;
# 
#     return 0;
# }

Thomas


-----BEGIN PGP SIGNATURE-----

iD8DBQFEQqLS3w+/yD4P9tIRAiuRAKDDhz8rQvR9RvNNQ5dr1695S/sxXwCggY2J
UZmiwvY/ij2UnJpFXHi2b1E=
=Odkq
-----END PGP SIGNATURE-----

Li Jie <cpunion gmail.com> writes:

In article <ifkah3-dl7.ln1 birke.kuehne.cn>, Thomas Kuehne says...
I think that isn't (yet) possible.

If there are many different matrix sizes you may try:

# class Matrix(T){
#     this(T[][] data){
#         // ...
#     }
# 
#     typeof(this) opMul(typeof(this) arg){
#         typeof(this) result;
#         // ...
#         return result;
#     }
# 
#     Matrix!(T) opDiv(Matrix!(T) arg){
#         Matrix!(T) result;
#         // ...
#         return result;
#     }
# }
# 
# int main(){
#     Matrix!(int) a;
#     Matrix!(int) b;
#     // ...
#     auto c = a * b;
#     Matrix!(int) d = a / b;
# 
#     return 0;
# }


If you only deal with a few matrix sizes you may try:

# template opMul(int R2, int C2){
#     Matrix!(T, R, C2) opMul(Matrix!(T, R2, C2) arg){
#         Matrix!(T, R, C2) result;
#         // ...
#         return result; 
#     }
# }
# 
# class Matrix(T, int R, int C){
#     // ...
# 
#     mixin opMul!(4, 8);
#     // and other mixins
# }
# 
# int main(){
#     Matrix!(int, 3, 4) a;
#     Matrix!(int, 4, 8) b;
#     // ...
#     auto c = a * b;
# 
#     return 0;
# }

Thomas


Thanks Thomas.
Very depressing, Aha.. I think that only this way:

# template dotMul(T, int R, int C, int C1){
#   Matrix!(T,R,C1) dotMul (Matrix!(T,R,C) lhs, Matrix!(T,C,C1) rhs) {
#     // ...
#   }
# }
# 
# auto m1 = Matrix!(int,3,4);
# auto m2 = Matrix!(int,4,8);
# auto m3 = dotMul!(int,3,4,8)(m1,m2); // OK
# auto m4 = dotMul(m1,m2); // Compile error..