Fix a bunch of warnings

[illwieckz]

Fix a bunch of warnings, also delete an unused macro.

[slipher]

crnlib: fix -Wimplicit-const-int-float-conversion looks really ugly. Seems like we'd be better off disabling this warning

I also dislike the crnlib: fix -Wmaybe-uninitialized one. The assignment of dummy values which are never supposed to be used obscures the meaning of the code. Also it prevents sanitizer tools that can detect uninitialized memory use from uncovering possible bugs in the algorithm in the case that a value was unexpectedly not assigned.

[slipher]

How about just replacing this with static_assert? CRNLIB_ASSUME is a pre-C++11 implementation for static assertions.

[illwieckz]

Just doing this?

#define CRNLIB_ASSUME(p) static_assert(p, "")

[slipher]

yes

[illwieckz]

Now done.

[illwieckz]

crnlib: fix -Wimplicit-const-int-float-conversion looks really ugly. Seems like we'd be better off disabling this warning

This is about warnings like that:

warning: implicit conversion from 'int' to 'float' changes value from 2147483647 to 2147483648
        if (((*m_pVec3F)[component] < INT_MIN) || ((*m_pVec3F)[component] > INT_MAX)) {
                                                                          ~ ^~~~~~~

I guess this MSVC warning spammed in MSVC AppVeyor CI is the same one:

warning C4244: 'argument': conversion from 'T' to 'T', possible loss of data
          with
          [
              T=float
          ]
          and
          [
              T=int
          ]

[slipher]

crnlib: fix -Wimplicit-const-int-float-conversion looks really ugly. Seems like we'd be better off disabling this warning

This is about warnings like that:

warning: implicit conversion from 'int' to 'float' changes value from 2147483647 to 2147483648
        if (((*m_pVec3F)[component] < INT_MIN) || ((*m_pVec3F)[component] > INT_MAX)) {
                                                                          ~ ^~~~~~~

I guess this MSVC warning spammed in MSVC AppVeyor CI is the same one:

warning C4244: 'argument': conversion from 'T' to 'T', possible loss of data
          with
          [
              T=float
          ]
          and
          [
              T=int
          ]

I looked at that commit closer and it is actually completely wrong. For example in

int i = m_float;
if ((i < INT_MIN) || (i > INT_MAX)) {

The condition is equivalent to if (false). An int cannot be less than INT_MIN or greater than INT_MAX

[illwieckz]

The condition is equivalent to if (false). An int cannot be less than INT_MIN or greater than INT_MAX

Right, this is stupid indeed!

I rewrote that commit by doing (float)INT_MAX instead.

[slipher]

More CRNLIB_ASSUME bits that can be deleted above

[illwieckz]

Done.

[slipher]

As I said earlier I don't think it's an improvement to treat Wmaybe-uninitialized. LGTM to the other commits

[illwieckz]

As I said earlier I don't think it's an improvement to treat Wmaybe-uninitialized. LGTM to the other commits

Thanks, I now removed this one.