2D Matrix, Compacting problem

Question

Hi!
I'm about to implement a matrix class that need som kind of "compacting" support, meaning:
If the the following matrix (3x4) is compacted:
[1][0][2][8]
[7][0][3][9]
[0][0][0][0]
The resulting matrix (2x3)would be:
[1][2][8]
[7][3][9]
Column 2 will be removed since it's empty (only zeros)
and row 3 will be removed as well sinces its empty.
I've tried a number of solutions, but they are all inneficient and/or cumbersome.
I'm thinking like this:
Find out which rows and columns should be removed.
Create a new matrix with the new size, and iterate over the first matrix copying the values.
Any ideas that come to mind?
What about choice of datastructure?
Is the best I can do to use a long[][] or maybe a
combination of Lists of some sort?
Any ideas for the algorithm?
Oh, the size of the matrix will in most of the times be around 80x80 to maybe 120x120 (not necessarily same rowcount as colcount)
And it must support getting/adding at specific cells, because of the way it's filled with values (from database)
/ Ronnie

Answer 1

Find out which rows and columns should be removed.Well, that's easy. Two arrays of booleans, initially all false, iterate through the entire matrix setting appropriate booleans to true if you find a non-zero element.
Create a new matrix with the new size, and iterate over the first matrix copying the values.Convert each array of booleans into an array of ints, which are the subscripts of the boolean arrays which were true. E.g. {f, t, ,t f, t} -> {1, 2, 4}.
The sizes of these arrays give the matrix size. Two nested loops to copy elements across.