On 03/27/2015 03:11 AM, Markus Armbruster wrote: > Eric Blake <ebl...@redhat.com> writes: > >> ...or an array of dictionaries. Although we have to cater to >> existing commands, returning a non-dictionary means the command >> is not extensible (no new name/value pairs can be added if more >> information must be returned in parallel). By making the >> whitelist explicit, any new command that falls foul of this >> practice will have to be self-documenting, which will encourage >> developers to either justify the action or rework the design to >> use a dictionary after all. >> >> Signed-off-by: Eric Blake <ebl...@redhat.com> >> ---
> Since there's just one whitelist, all schemata share it, and that means > it's too permissive for each of them. Sloppy, but good enough. > > If the sloppiness bothers us, here are two alternatives: Doesn't bother me enough to use either alternative, but I _will_ update the commit message to at least remind someone of the potential for tightening things, should the need arise in the future. > > * Program takes the whitelist as argument, say > > scripts/qapi-commands.py --legacy-returns qmp-legacy-returns ... Early exit (generator fails if you don't whitelist), but noisy (have to touch a Makefile in addition to the .json schema). > > * Leave enforcing to C > > If a command 'frobnicate' returns a non-dictionary, generate something > like > > #ifndef FROBNICATE_LEGACY_RETURN_OK > #error Command 'frobnicate' should return a dictionary > #endif > > Then manually define the macros necessary to keep the current use > working in a suitable header. Late exit (failure does not occur at python generation time, but only at C compilation time) - and the point of this series was to promote several late exits into early exits. But slightly cleaner than having to touch Makefiles to add in the whitelist. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
