[Gmsh] Why does "Still negative jacobians" mean error?

[Jeremy Theler]

Hi all

I see in contrib/HighOrderMeshOptimizer/OptHomRun.cpp:686 that if the
higher-order optimizer cannot make all the jacobian non-negative, then
an error is raised.

Why is this a hard error and not a warning? I have seen cases where the
optimizer cannot make all of then non-negative, yet the results are
reasonable. Moreover, if not optimizing high order, they still are (and
the general quality is worse). So I do not get why this is an error.

Any comment?

Thanks
--
jeremy
www.seamplex.com