Having implicit conversion on Unit class causes much more harm than
doing the conversion some other way -- mainly for multiplication and
division, which is done with builtin operators and thus the result is
unitless. Also it is possible to do wrong "retyping" between degrees and
radians:
Rad a = Constants::pi();
Deg b(a*2);
On GCC >4.5 b is `180`, but on GCC 4.4 it is `2pi`, which is obviously
wrong and not wanted.
The conversion is done using Unit::toUnderlyingType() which is now used
everywhere instead of conversion operator. The conversion operator is
made available only for GCC 4.5, where it can be marked as explicit.
This might cause incompatibilites with `master` branch, where no
Unit::toUnderlyingType() exists, but users should have no need to
convert it to underlying type anyway as all required functions are
available through the library.
This reverts commit 36ac4de5c4 and
partially also 0db3a183ae (UnitTest.cpp).
Due to Matrix/Vector rework and other changes this is fixed on lower
levels already. The tests pass on both 4.5 and 4.4, hopefully it won't
break anywhere else.
These four cover most (but not all) warnings when compiling in Release
mode. Now all the tests pass, but the warnings signalize that something
could go bad somewhere else. However, only one drastic solution comes to
my mind at this time -- disabling strict aliasing completely
`-fno-strict-aliasing`. Don't know performance impact of that.
Awesome bug. In GCC 4.6 it throws plenty of ungoogleable `-pedantic`
warnings and in GCC 4.5 it fails directly with "error: bad array
initializer". Fallback to initialization using for-cycle.
Worked around that in 4.6 by disabling `-pedantic` warnings, must be
done this way on 4.5 & 4.4. Hopefully the performance won't be harmed
too much.
It does too much harm on GCC 4.6 (all these constexpr constructors are
not constexpr now). We can disable that `-pedantic` warning for GCC 4.6
only and live with that.
This reverts commit 2d92d497d9.
Awesome bug. In GCC 4.6 it throws plenty of ungoogleable `-pedantic`
warnings and in GCC 4.5 it fails directly with "error: bad array
initializer". Fallback to initialization using for-cycle.
Hello, performance? You can go home now.
Comparing squared length to 1 is not sufficient to compare within range
[1 - epsilon, 1 + epsilon], as e.g. Quaternion with dot() = 1 + 1e-7
when converted to matrix has column vectors with dot() = 1 + 1e-6, which
is just above 1 + epsilon. Thus it's needed to compare sqrt(dot()) in
range [1 - epsilon, 1 + epsilon] or dot() in range [1 - 2*epsilon +
epsilon^2, 1 + 2*epsilon + epsilon^2]. Because epsilon^2 is way off
machine precision, it's omitted, thus dot() in all isNormalized()
implementations is now compared this way:
abs(dot() - 1) < 2*epsilon;
DualQuaternion and DualComplex has now only rotation() which returns
full rotation part, rotationAngle() and rotationAxis() on Quaternion and
Complex were renamed to angle() and axis().