On 06/03/2016 06:09 AM, Markus Armbruster wrote: > Most patches are ready as is or with minor touch-ups. > > I'd probably drop PATCH 01. The file name collision no longer exists, > and moving files just to improve their names a bit doesn't seem worth > the bother.
It's a bit more churn to the rest of the series, but I can live with it. It's also more related to the JSON visitor than the clone visitor, so even if I do want to keep it, it makes more sense to keep it with the second half of the series. > > There are memory leaks in PATCH 12 and 24 (see review of PATCH 24). > > PATCH 13 needs more work than I'm comfortable to do on commit. Mostly > missing comment updates, but also a crash bug. I've got that fix ready to post. > > On PATCH 21, I have minor interface design doubts. > > The only difficult question is what to do about invalid UTF-8 strings > and non-finite numbers. Patches before PATCH 29 show opinions, but a > decision isn't made until PATCH 29. Good, because it leaves most of > this series not blocked by the debate on what we should do. > > I guess I could take PATCH 02-12 now, and let you respin the rest. But > perhaps its easier if you respin all of them. I'm posting v5 of 2-15 now, then focusing on your comments on the JSON visitor for a later posting. I also think my subset F series (adding a 'box' parameter) is relatively orthogonal, if you want to start reviewing that: https://lists.gnu.org/archive/html/qemu-devel/2016-05/msg03569.html I'd still LOVE to get blockdev-add working for ALL block devices before 2.7 soft freeze (there's patches for NBD that I need to review, and I know the gluster folks were waiting on qapi patches to be able to use a SocketAddress union as the basis for the gluster branch of the blockdev-add union - and my qapi patches are almost all the way there to permitting it). -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature