#23 bug in mat.cpp

[jens röhner]

bug in:

Mat<T>::operator!=(const Mat<T> &m) const

the line (1001), \c

if (data[i]!=m.data[i]) return false;

should return true

further remarks:

line 1042: one ';' sign too much;

i optimized the operator \c

Mat<T>::operator*=(const Mat<T> &m)

for better memory usage. maybe you have some need for
it ?
\code

template<class T> inline
void Mat<T>::operator=(const Mat<T> &m)
{
it_assert1(no_cols == m.no_rows,"Mat<T>::operator=:
wrong sizes");
Vec<T> tmp(m.no_cols);

int i,j,k, pos=0, m_pos=0;

for (i=0; i<no_rows; i++) {
    tmp.zeros();
    for (j=0; j<m.no_cols; j++) {
        m_pos=0;
        for (k=0; k<no_cols; k++) {
        tmp(j) += data[pos+k] *

m.data[m_pos+j];
m_pos += m.row_offset;
}
}
set_row( i, tmp);
pos += row_offset;
}
}

\endcode

i tested it vor \c mat and \c cmat ...

[Tony Ottosson Gadd]

Logged In: YES
user_id=326034

This bug has actually already been discovered, but not
reported into the bug-system. A fix is already into the CVS.

The more memory efficient version of oper*=() is of
interest, but will have to wait until after 3.7.0 since
there is an internal change of storage (from row-major to
column-major).

/Tony