Sorry this is old, I'm taking some time on an airplane (sigh, yes, again) to reread and respond to some saved email.
> In conversations with the team, perhaps a warning > rather than an error might be wisest for the binary rpm package > of the LSB runtime test suite (for example if it cannot find > the LSB dynamic > linker), although there is a counter argument that says we should > not allow our certification test suite to even install if > the required dependencies in the system are not met, since > otherwise how would we be able to tell that the system under test > and the journal sent in, is not in fact from a system > not providing the required dependencies. > The good news is that other parts of the certification process will > require rpm packages to be installed that do test for the > dependency. A question, to what extent are we able to have package (rpm) level dependencies? I'm completely ignorant as to what happens when a package is imported in some way for installation to a non-RPM system. Are the same sorts of dependencies maintained when a tool like alien (isn't that what Debian calls it?) is used? I realize that the above is just a comment, not spec wording, so I'm not criticising, but we have to continue to be careful not to imply that it's an rpm-based world, much though it would make our lives easier if there were a single common packaging system. In Re: the suggestion of a warning, did a bug need to be filed to track this to some agreement on the proper behavior? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with subject of "unsubscribe". Trouble? Email [EMAIL PROTECTED]
