SRA and inconsistencies in bit-field types

[Richard Kenner]

> Any objections to my introducing ADAC or GNATC or some such
> environment variable, to be preferred over CC to detect Ada support
> and to compile Ada sources in stage 1?

We did that a long time ago (I think it was ADAC), but it led to of
problems, such as the C parts of the Ada front end being compiled with
a different compiler than the Ada parts and running into all sorts of
inconsistency problems.