Nemocas
Avoid try-catch in tests
The issue https://github.com/Nemocas/Nemo.jl/issues/2012 is related to a try-catch statement in ConformanceTests. I think it's a very bad idea to use try-catch statements as they just mask problems, in particular in tests.
On a broader front, it looks like they are used extensively in Oscar. Why is it used so extensively?
Generating some opinions before starting to poke around in the code. Ping @fieker @fingolfin @thofma
Bill introduced them as a clever way to try fast algorithms - that may fail - first, think some nmod/ fmpz_mod stuff that might fail if non-units are encountered and are correct otherwise. Then the catch will invoke the slow, but safe method.
Other use case are modular algorithms where I do not know beforehand that the projection to GF(p) is going to work. I think real-lll is also doing this to signal not enough precision
But yes, we were bitten before when we did not carefully catch the correct error...
