Page 19 of 22

Re: OpenMW 0.41.0

Posted: 22 Dec 2016, 14:45
by scrawl
Um, I made that scan for the master branch, not 0.41 :roll: Don't worry about it.

FWIW, the atan2 one is a false positive, the others are now fixed.

Re: OpenMW 0.41.0

Posted: 22 Dec 2016, 14:57
by Allofich
For the atan2 false positive, I don't understand what's happening so I might be saying something stupid, but does Coverity just think it's wrong because it sees variable names with x and y in the opposite order? Would it make sense to just store x and y as oppositely named variables and then pass them on in the order Coverity expects, maybe with an explanatory comment? Might help avoid more misunderstandings in the future.

Re: OpenMW 0.41.0

Posted: 22 Dec 2016, 15:02
by psi29a
scrawl wrote:Um, I made that scan for the master branch, not 0.41 :roll: Don't worry about it.

FWIW, the atan2 one is a false positive, the others are now fixed.
Ah, purely coincidence that the tag was made around the time the scan was triggered. ;)

Good to know and thanks Scrawl, that is what I wanted to hear. 8-)

I'll get on the release for main PPA and Debian when I get back from posting Christmas cards...

Re: OpenMW 0.41.0

Posted: 22 Dec 2016, 15:06
by Jyby
scrawl wrote:FWIW, the atan2 one is a false positive
Because you've flipped the x and y vectors in the rest of the code? This looks suspicious because atan2 is a piecewise function. You might not be getting what you expect. I haven't looked into what we're doing exactly though..

Image

To those who don't know, you can take away from this function the differing behavior if you flip x and y.

atan2(1,0) = pi/2
atan2(0,1) = 0

If you think your usage is correct, it would be wise to suppress the coverty result, perhaps with local variables set to the dirx diry, and make a comment about it incase someone looks at that code again.

Re: OpenMW 0.41.0

Posted: 23 Dec 2016, 00:01
by Chris
Jyby wrote:Because you've flipped the x and y vectors in the rest of the code? This looks suspicious because atan2 is a piecewise function.
Many applications tend to use 0 degrees as a forward-facing angle. Given that, atan2's Y parameter maps to the right-facing vector component, and the X parameter maps to the forward-facing vector component. However, most 2D and 3D applications use X as the right-facing vector component, and either Y, +Z, or -Z as the forward-facing vector component. So properly mapping that to atan2 makes the X component fit the Y parameter, and the Y/+Z/-Z component fit the X parameter.

Re: OpenMW 0.41.0

Posted: 23 Dec 2016, 01:17
by K1ll
OpenMW 0.41.0 Linux targz package release builds:

32 Bit
64 Bit

Re: OpenMW 0.41.0

Posted: 23 Dec 2016, 01:54
by sg1efc
K1ll wrote:OpenMW 0.41.0 Linux targz package release builds:

32 Bit
64 Bit
Thanks everyone! 8-)

Re: OpenMW 0.41.0

Posted: 24 Dec 2016, 00:42
by raevol
Shoot, I think I am still waiting on the OSX release package. I am leaving in a few hours and may not be able to do the release until Monday. :| Sorry for the delay everyone.

Re: OpenMW 0.41.0

Posted: 24 Dec 2016, 00:48
by Atahualpa
No worries, raevol. A release on Monday would be really nice -- just after the Christmas celebrations when everyone is full up and satisfied. :)
The change log and the videos are ready; let's hope that there are no errors popping up.

Re: OpenMW 0.41.0

Posted: 24 Dec 2016, 09:42
by weedfreak
0.41 is now released on Arch! No need to wait until Monday.