On 12 May 2016 at 08:40, Christian Borntraeger <borntrae...@de.ibm.com> wrote: > Maybe a topic for this years QEMU summit could be to talk about > release process and release criterias.
Yeah, I'm happy to talk about what we could do better with releases (both on the mailing list and at the summit). A couple of notes: This time around we had to delay by at least a week because of the timing of the CVEs -- we had to allow a reasonable time before raising the embargo for distros to prepare fixes, and the bugs came in pretty close to when we'd otherwise have done our final rc for the release. I think one significant difference between us and Linux is that we have fewer people testing our rcs, so I worry that if we put more stuff in then we are less likely to notice bugs in it in time. The intended purpose of the "few days gap then final release is same as last rc" approach is to give a safety valve in case the patches that went into the final rc had some horrendous bug in them. In this case the last rc only had the CVE fixes so was pretty safe, but in previous releases we've often had a few more patches than that in final-rc. I don't think two extra days before reopening the tree is a very big cost. thanks -- PMM