Tom Lane wrote:
> Is it intentional that 5.0.19 breaks binary compatibility with clients
> built against 5.0.18? The addition of a field to the MYSQL struct has
> had that effect, since a client that has allocated its own MYSQL struct
> will have not made it large enough. This is for example causing crashes
> in MySQL-python for me. It woulda been nice if there were at least a
> large warning in the release notes about it.
No, this was not intentional. We've narrowed down the field, know what
it was included for (embedded server fixes), and are looking into this,
especially since we plan on building 5.0.20 really soon now.
Suggestions would be to bump up the version number of libmysqlclient.so
and of course, never break ABI in GA releases. Thanks for pointing this
out to us.
Please hang in there, and I'll provide you with more information as soon
as I have it.
Colin Charles, Community Engineer
MySQL AB, Melbourne, Australia, www.mysql.com
Office: 4528@stripped / +46 18 174 400 ext. 4528
Mobile: +614 12 593 292 / Skype: colincharles
Using MySQL for a FLOSS application?