Documentation misleading concerning the order when combining transforms

[pvigier]

Hello,

If I understand well the source code of sf::Transform, when we do:

auto A = sf::Transform(...);
auto B = sf::Transform(...);
auto C = A.combine(B);

C is equal to A*B with * the matrix multiplication.

And when we want to transform a point:

auto u = sf::Vector2f(...);
auto v = C.transformPoint(u);

We have v = A*B*u.

So it is B that is applied first on u, then A.

However, in the documentation of combine, it is said that:

The result is a transform that is equivalent to applying *this followed by transform.

According, to this sentence, C would be equal to B*A (applying A followed by B). I think it is misleading.

Correct me if I am wrong.

Best regards,
Pierre

[LaurentGomila]

I don't remember this kind of stuff that was written a long time ago, so the best thing would probably be to test it. A rotation combined with a translation should be a good case to figure out which one is applied first.

[pvigier]

Indeed, I have done exactly that before posting to be sure.

Here is the snippet:

auto transform = sf::Transform();
transform.translate(2.0f, 3.0f);
transform.rotate(90.0f);

Then, I print the entries of the matrix:

-4.37114e-08 -1 0 2 
1 -4.37114e-08 0 3 
0 0 1 0 
0 0 0 1 

Which corresponds to rotation first, translation then.

I think that the tutorial on transforms is also misleading. It is said that :

You can apply several predefined transformations to the same transform as well. They will all be combined sequentially:

I think it should be written: "They will all be combined sequentially in reversed order".

[Bromeon]

Thanks for the report, I created a PR in both SFML and SFML-Website to address this.