[m-dev.] reporting inconsistencies when building interface files

[Julien Fischer]

On Wed, 11 Sep 2019, Zoltan Somogyi wrote:

> Would anyone object to the second approach described by the
> long comment in the attached partial diff? The comment also describes
> the issues involved, and my reason for why bugs such as defining a type
> as being both a du and an equivalence type should be detected and
> reported at interface file build time. Since such errors are rare in
> my experience, which way we handle them do not much matter,
> but having all the relevant definitions available all at once
> will simplify the code that makes type representation decisions.

I have no objection to the second approach.

Julien.