mysql-binlog-connector-java icon indicating copy to clipboard operation
mysql-binlog-connector-java copied to clipboard

Fix in "upstream" not "synced" to this repo

Open walro opened this issue 5 years ago • 2 comments

We recently switched over to using your fork rather than the original version, as a result of that I checked the difference between original/"upstream" and this fork, using: https://github.com/osheroff/mysql-binlog-connector-java/compare/master...shyiko:master

It appears that the fix for https://github.com/shyiko/mysql-binlog-connector-java/issues/321 is not included in this repo, as it was commited after "time-of-fork". In particular it's the following two commits that could potentially be of use here too: https://github.com/shyiko/mysql-binlog-connector-java/compare/0.20.1...c54b4c1

Cherry-picking the first commit, e59006e, worked without any hassle. Things got a bit hairy when trying to apply c54b4c1 as lines in the same file were touched later on.

I will try to put up a PR, but preemptively created this issue in case I fail :-)

Thanks for keeping the project alive!

walro avatar Oct 26 '20 13:10 walro

ugh, yeah. c54b4c1 has a lot of refactoring / unrelated changes that are deeply in conflict. do you think we could find a more surgical fix for whatever the deadlock issue there was?

osheroff avatar Oct 29 '20 22:10 osheroff

Looking at the latter commit, ignoring white-space changes, one ends up with: https://github.com/osheroff/mysql-binlog-connector-java/commit/c54b4c18ca1c5299792a4be32fa0296a49259362?w=1 which is a bit more manageable perhaps. I'll assume that the SSL stuff is unrelated. Also I imagine one could live without the renaming of the ensureEventDataDeserialize method.

I'll see if I can can even get a re-pro going for starters, as no tests were added. (Internally, we actually run without keepAlive set to true, so we are probably very unlikely to ever run into this problem)

walro avatar Oct 30 '20 08:10 walro