SE++ 0.17 runs in circles

[mkuemmel]

I found another case where SE++ runs in circles doing nothing. The full dataset is available here (280MB):
https://we.tl/t-dbmTeDzXxl

The catch is the python config file asks for Sersic fit, but the only measurement image does not have PSF specified. Also the ASCII config file has no PSF, hence SE++ runs in circles.

Giving a PSF in the config file "psf-filename=...", "psf-pixel-sampling=..." or likely giving a PSF with the image solves the problem.

Should either ask for a PSF or at least give an error.

No big deal, but....

[ayllon]

Probably related to #438. My guess is that an exception is thrown and the thread pool does not recover properly from it.
Let me check.

[mkuemmel]

Well, #438 appeared to be solved with Marc's commit. Do you think that was only kind of accidental?

[ayllon]

No, not at all. The exception itself was fixed, but the problem with handling exceptions may still be there. I think the thread pool fails to finish gracefully when threads start dying and just hangs.

[mkuemmel]

That version is no longer relevant and the recent v0.21 does not show this behaviour.