On 03/02/2017 11:39 PM, Markus Armbruster wrote: > Eric Blake <ebl...@redhat.com> writes: > >> On 03/02/2017 03:43 PM, Markus Armbruster wrote: >>> Bad error handling, memory leaks, and lack of blockdev-add support. >> >> How hard are we trying to get blockdev-add working in 2.9? Or is this >> series 2.10 material now? > > Definitely not 2.10: seven patches fix or document bugs, one improves > error messages, five are straightforward cleanups. The series touches > only these two block drivers, including QAPI schema parts not used > anywhere else: > >>> block/gluster.c | 127 +++++++-------- >>> block/sheepdog.c | 436 >>> +++++++++++++++++++++++++++++++++++++-------------- >>> qapi-schema.json | 38 +++++ >>> qapi/block-core.json | 73 +++------ >>> 4 files changed, 443 insertions(+), 231 deletions(-) > > With the pending pull requests merged, blockdev-add *is* working, except > for sheepdog. I'm considering that a bug, and I need *two* patches to > fix it. One touches only sheepdog.c, and the other only adds to the > QAPI schema.
I can agree to that. It may help if the v2 is explicit that the series is for-2.9 in the subject. > > If we decide not to fix the bug, I'd recommend to declare blockdev-add > supported in 2.9 anyway, with a release note that sheepdog support is > broken. > Yes, I think we can still treat the rename of 'x-blockdev-del' to 'blockdev-del' as a bug fix appropriate for 2.9, finally declaring the feature stable. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature