ODE Internal error 1: flying FW 190 then sploded

[wimsey]

P-38 messed me up, took my tail off and continued laying into me. Was flying a Fw190 till I went boom.

Error message: "Assertion 'bNormalizationResult' failed in ../../include/ode/odemath.h line 321"

Additional info, I was able to nab a screenshot where the GPS module says

"No Towns Near:

NaN.Nan.NanS Nan.Nan.NanW"

The tilde window reads

"G: 1.#R AOA: 1.$ Slip: 1.$ deg"

The minimap shows me being zoomed into "1m" and is completely black.

Can upload the screenshot if you like.

[wimsey]

Mission eventually showed up in CSR as an RTB, but I was dead, very much dead.

http://csr.wwiionline.com/scripts/services/persona/sortie.jsp?sid=1634130&pid=715642&page=1

[GOPHUR]

Sent to coders.

[ratatosk]

I just got the same error and the game froze. What was strange is that I wasn't even in the game world, I was eating lunch and looking at the brigade list.

i7 quad, 9GB, win7/64, GTX260, yada, yada. this is teh sux, please fix. Thanks.

[AHWULF]

Ramp and I believe this and a common ctd were caused by the Intel compiler and its funky optimizations. The identical code on Mac compiled by GCC compiler (old but reliable) has never had either problem so it's a good bet it's the compiler.

For 1.34 we have switched to the MS2010 compiler and so far neither has happened. Performance didn't change but the code output seems more stable. We won't know for sure until a broader open beta.

Floating point code can be very sensitive to incorrectly optimizations and ODE (physics library) is virtually all floating point code. Probably one line of code isn't doing what the coder wrote...

[ratatosk]

thanks for the reply and explanation, ahwulf. I appreciate your efforts.

[GOPHUR]

Indeed. Nice work. We've seen none of these crashes in beta so far.