DBD::mysql – even more utf8-issues fixed, super CI

In my previous post I explained how many utf-8 related issues are now fixed in the latest DBD::mysql development release, and I asked for feedback and testing from our users.

Even better utf-8 handling

I’m happy to inform you that, thanks to your feedback, we’ve now released yet another development version, with even more UTF8 issues fixed. Tanabe Yoshinori reported that column names and database warnings were not properly encoded and prolific contributor Pali Rohár fixed that issue, and much more.

Serialization issue under taint mode

One other issue which was reported by amavisd users has to do with how perl stores values internally under taint mode, which is the default way amavisd is run. This is now fixed, see for more information https://github.com/perl5-dbi/DBD-mysql/issues/78.

Extreme CI testing setup

And Pali also modified our Travis setup so we now do continuous integration testing not only on many different perl versions, but also using many different MySQL and MariaDB versions. This uncovered many smaller and larger issues and we can compile again all the way back with MySQL against version 4 if you would want that (and we had multiple people wanting that and filing bug reports for this in the past).

This also lead to Pali discovering a use-after-free security issue (CVE-2017-3302) in libmysqlclient which was fixed in MySQL 5.6 and up but still present in 5.5 and also in MariaDB. See for more info this thread on oss-security.

Your feedback is welcome!

Find the full change log below; when all is well on Wednesday 8th of March we’ll release the stable version 4.042, including all these changes, as well as the changes from the previous post.

You can leave your feedback via the DBI-users mailing list, or using our GitHub page.

2017-02-28 change log of version 4.041_2)