QGenericMatrix? why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?

opengpu

why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?

opengpu

QGenericMatrix?
can i use like this QGenericMatrix<VARIANT>
VARIANT is from msdn

opengpu

actually, i just want to map the content(which are all number) into an abstract layer.
so i want to use QGenericMatrix which also has row and column, and can get numer-data by row&coloumn.
is there better way?
now i want to store the real VARIANT, and not store the QString which is convert from the number-data

JKSH

@opengpu said in QGenericMatrix? why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?:

why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?

QMatrix is obsolete. Don't use it.

Use QGenericMatrix. It is a template class, like the Boost matrix.

@opengpu said in QGenericMatrix? why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?:

now i want to store the real VARIANT

Why not just store a number? (Like int32_t)

opengpu

@JKSH said in QGenericMatrix? why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?:

VARIANT

other than Qt type is not adivised to use?

jsulm

@opengpu Can you explain what you want to do? Why do you want to store a variant in a QGenericMatrix? What is the use case?

opengpu

@jsulm VARIANT is the data i get from 3rdParty

jsulm

@opengpu But what is it? A struct/class? A number? Or something like QVariant?

opengpu

@jsulm
https://docs.microsoft.com/en-us/windows/desktop/winauto/variant-structure

jsulm

@opengpu Why do you want to put VARIANT in a matrix? Do you want to do matrix based calculations? Maybe a two dimensional QVector would be enough?

opengpu

@jsulm said in QGenericMatrix? why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?:

dimensional QVector

dimensional QVector?
yes, i just want to get the Var besed on row&column

jsulm

@opengpu

QVector<QVector<VARIANT>> _2dArray;

JKSH

@opengpu said in QGenericMatrix? why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?:

yes, i just want to get the Var besed on row&column

I agree with @jsulm -- you should use QVector<QVector<T>>.

QGenericMatrix and the old QMatrix are designed to be used as transformation matrices: https://en.wikipedia.org/wiki/Transformation_matrix. They are not designed to hold row+column data.

opengpu

@jsulm said in QGenericMatrix? why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?:

QVector<QVector<VARIANT>> _2dArray;

_2dArray[iRow][iCol] = 1.0f; //this is ok
but how should i set the presice size of _2dArray?
_2dArray.resize(rowCount * colCount); //this is not correct, right?

jsulm

@opengpu said in QGenericMatrix? why not QMatrix a template? is there anything like boost::numeric::ublas::matrix<AAA>?:

_2dArray.resize(rowCount * colCount); //this is not correct, right?

No, it isn't

_2dArray.resize(rowCount);
// And now for each row
_2dArray[row].resize(colCount);

opengpu

This post is deleted!

opengpu

this is right

_2dArray.resize(rowCount);
for (int i = 0; i <_2dArray.size(); ++i)
{
_2dArray[i].resize(colCount);
}

VRonin

If you need a 2d container for type T with fixed dimensions and having all rows with the same number of columns you should use boost::multi_array<T,2>.
QVector<QVector<T>> is fine but it's more flexible than you need and hence less efficient. I would stick with the good old boost:

boost::multi_array<VARIANT,2> _2dArray(boost::extents[rowCount][colCount]);