On 05/06/2011 08:32 AM, Matthias Bolte wrote: > To be really sure, we might add a whitelist aside the backlist and > once the generator sees something that's not on one of the lists it > complains with an error that tells the programmer to either add the > function to whitelist and check that the generated code does the right > thing, or add it to the blacklist and add a function body manually. I like that idea the most. > On the other hand, stuff like flags parameter being unsigned in the > public API but signed in the XDR protocol need manual special cases in > the generator. We could add some general sanity checks to avoid > something like this in the future. The generator could complain when a > parameter is called flags but is signed. I'll work on that too. Also a good idea. It's not an API change if we switch existing offenders to unsigned, is it? Even for C++, where there is function overloading, we are using 'extern "C"' declarations. > > So, are you just inclined to ACK or do you ACK this? :) > I'm with danpb here - Treat this as an ACK, and push now, to maximize the testing exposure. Do your proposed cleanups (documentation, sanity checking, whitelisting) as followups. -- Eric Blake eblake@xxxxxxxxxx +1-801-349-2682 Libvirt virtualization library http://libvirt.org
Attachment:
signature.asc
Description: OpenPGP digital signature
-- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list