On Mon, Nov 07, 2016 at 02:47:56PM +0100, Maximiliano Curia wrote: > >I think the mention of backports here was a mistake - the submitter > >specifies that the problem happened when upgrading to 5.5.53-0+deb8u1 > >which was a security update. > > Ups, it seems that I misread the version.
No problem. > In any case, ideally, a stable security fix shouldn't break existing > software, so, if feasible, it would be better to have this fixed in mysqld. Agreed, though in this case it was a deliberate decision by upstream to close a security hole, AIUI. The impact on akonadi wasn't anticipated. Sorry about that. We will consider akonadi's rather special use case separately next time. In terms of a fix in mysqld, I'm not sure what we could do. We can revert the behaviour change (IIRC upstream left us a build-time option for this purpose), but AFAIK this would leave users vulnerable. It wouldn't have helped this time, but we have had regressions in akonadi in the past due to MySQL changes too. Any chance you could add some functional dep8 tests to akonadi packaging, and then we could adjust our processes to try to make sure these get run before landing any changes? Robie
signature.asc
Description: PGP signature