On 03/19/10 12:51 PM, Mike Kupfer wrote: > On 03/18/10 02:06 PM, Mark J. Nelson wrote: > >> Do you want to propagate any failures from copyin_crypto() through >> crypto_passhtrough(), and avoid attempting to copyout_crypto() if there >> were failures? > > Good question. The error handling for Tonic builds could be more > uniform, in general. For example, we mark the build as failed if a > signing build is unable to generate the crypto tarball (see > crypto_from_proto()), but we don't mark the build as failed if a > non-signing build has a corresponding failure (crypto_passthrough(), > copyin_crypto()). > > I'm starting to think that the build should be marked as failing if any > of the OpenSolaris deliverables cannot be generated, for whatever > reason. Does that sound reasonable?
Yes, absolutely. Otherwise, users of -O (both of them?) get a false sense of security. --Mark _______________________________________________ tools-discuss mailing list tools-discuss@opensolaris.org
