Máté Kelemen

> If the solution doesn't converge for the preceding case, the code breaks and can't resume for the next case. By breaking, do you mean it 1) raises an Exception...

Ah ok so there are no errors but the simulation stops advancing after a step fails to converge. Can you please copy your ```ProjectParameters.json``` here? (you can put it in...

Oh god; good chance to replace this with a regex?

I'm curious: shouldn't this have already caused segfaults? For example, [this test](https://github.com/KratosMultiphysics/Kratos/blob/55d962ddeff800edcc6ae4c9f3dd3cda5fb2be28/kratos/tests/test_function_parser_utility.py#L28) begins with an ```x``` but passed the CI on every compiler for years now.

> With clang's implementation this breaks We also have a clang run in the CI, don't we? Maybe this only becomes an issue in release builds? Edit: oh wait, is...

@miguelmaso it's ready for review. Ping @KratosMultiphysics/technical-committee

> ``` > typedef PointerVectorSet SetIdentityFunction, > std::less, > std::equal_to, > DofType* > DofsArrayType; > ``` ```IndexObject::operator()``` is a template and (as far as I can tell) has no specializations,...

Just a guess, but the ```pointer``` and ```reference``` member aliases of ```const_iterator_adaptor``` probably need to inherit the ```const```ness: ```cpp using pointer = const Parameters*; using reference = const Parameters&; ```

@roigcarlo ah found it, your aliases are ```private```, so ```iterator_traits``` can't find them.

> Moving to cpp: We could do something similar to forward declarations. It may be a bit overly convoluted, but we could organize templates like this: - A header containing...