On 09/24/2015 08:16 AM, Markus Armbruster wrote: > Eric Blake <ebl...@redhat.com> writes: > >> The documentation claims that alternates are useful for >> allowing two types, although nothing enforces this. Meanwhile, >> it is silent on whether empty unions are allowed. In practice, >> the generated code will compile, in part because we have a >> 'void *data' branch; but attempting to visit such a type will >> cause an abort(). Add some tests to expose the problems, and > > There's nothing really wrong with degenerate alternates or unions, but I > don't want to spend time on making them work. Outlawing them will do. > The commit message could perhaps be rephrased a bit to better convey > that.
Sure, that sounds like a better wording to weave in. > > What about empty structs and enums? Empty structs: required support (for example, 'Abort' as a member of 'transaction's union). Empty enums: Documented as not very useful but currently supported, and we already have code that tests that we don't choke on them. > >> adjust existing tests that should be failing for other reasons. > > I had to read this a few times until I understood "should be failing for > other reasons" means something like "are meant to test something else, > but could fail for the wrong reason if we reject degenerate alternates / > unions". I like it. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
