PlateSolve2 problem

thank you to having integrated this new solver, it seems to work very well … except for one on my target.
Last night I wanted to continue an old sequence with ngc6960 and it was impossible so solve any one of my previous pictures … while pictures from other targets was easily solved (same camera and scope). I switched to Elbrus and Astrometry, both solved this picture very quickly.
I tested with standalone PlateSolve2 program and there is no problem, image solved in few seconds.
It seems that there is a little problem somewhere, may be in the fits header. To be able to use the standalone PlateSolve2 correctly, I have to replace the comma by a dot in my windows system, may be there is something with the regional parameters…
I am curious to know…
PS: I uploaded one frame in 8bit fit because of the upload size limitation, it can be converted to 16bit to do some tests

uplNGC6960_300sec_1x1_OIII_003425 (2).zip (887.4 KB)

Thanks Pascal. I have identified an issue that would cause failed solved if decimal numbers were provided with a “,” separator. This has been addressed and will be out in the next beta.

thank you Ken.