On Thu, Feb 24, 2011 at 7:25 AM, Sage Weil <sage@xxxxxxxxxxxx> wrote: > On Wed, 23 Feb 2011, Colin McCabe wrote: >> Yeah, should add something like that. Maybe something like this: >> >> int rados_pool_list(rados_t cluster, char ***pools) >> void rados_pool_list_free(char **pools) >> >> Also should rename >> rados_snap_set_read -> rados_snap_set_read_context for consistency. > > Well, on the read side, it's not actually a context. You just need to > specify a snapid. On the write side, the context tells the OSDs which > snapshots to keep/cow when doing a write. You can't write to a snapshot. > I'm not sure the details are relevant to API users, but at the very least > I don't think there is a clear symmetry between the read/write side to > make the calls look consistent. This way it's clear they're not... > How about: void rados_pool_snap_set_read(rados_pool_t pool, rados_snap_t snap); int rados_pool_selfmanaged_snap_set_write_ctx(rados_pool_t pool, rados_snap_t seq, rados_snap_t *snaps, int num_snaps); The second one is getting long, but it ought to be clear that setting the write context can only be done with selfmanaged snaps. > ... > > The argument order for these is weird... not quite sure what I was > thinking. Let's do whatever pread and pwrite do? pwrite has buf, count, offset, so we could go with that. C. -- 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
