On Fri, 15 Jun 2012 11:29:03 -0400 Corey Bryant <coreyb@linux.vnet.ibm.com> wrote:
I'm not clear on what the problem is with returning the same error class for two different errors. Could you explain? I don't have a problem changing it if it's an issue.
Because an mngt app/user won't know if what has happened was that the fd name is invalid or if the fd name already exists.
I agree, it makes sense so the management app can program based on the error class. It just seems like the error classes in general should be able to be used multiple times if it fits, especially if we're trying to avoid having too specific errors.
Maybe (in the future) something like a sub-class could be used? In other words to allow using the same error class with unique sub-classes.
There's a discussion going on which is more or less the same topic: http://lists.gnu.org/archive/html/qemu-devel/2012-06/msg02243.html