On Mon, May 14, 2012 at 10:39 AM, Noah Watkins <jayhawk@xxxxxxxxxxx> wrote: > > On May 14, 2012, at 10:34 AM, Sage Weil wrote: > >> On Mon, 14 May 2012, Noah Watkins wrote: >>> On May 14, 2012, at 10:03 AM, Tommi Virtanen wrote: >>>> On Fri, May 11, 2012 at 6:21 PM, Noah Watkins <jayhawk@xxxxxxxxxxx> wrote: >>>>> Return an error rather than a segfault if a client uses the interface >>>>> when unmounted. The Java bindings were explicitly enforcing the >>>>> mounted state, but this was overly complicated, uncommon, and better >>>>> handled with a simple exception that can be checked at libcephfs level. >>>> ... >>>>> + if (!cmount->is_mounted()) >>>>> + return -EDOM; >>>> >>>> Noah, this is not really about your patch, as existing code already >>>> has return -EDOM, but... oh boy does the API abuse POSIX errors. >>>> >>>> "Numerical argument out of domain". >>> >>> :) Yeh, i stared at errno.h for a while. Note that -EDOM is already >>> returned by mount() when it is already mounted. >> >> Given the small number of libcephfs users, let's not be afraid to change >> it to something better. > > -ENOTCONN replace -EDOM for this patch > -EISCONN replace -EDOM for mount() when already mounted? I haven't had the chance to go over everything thoroughly (I notice a few other users of EDOM that I want to check out), but those definitely sound better to me. More better review comments later today. -Greg -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
