possibly a consequence of https://github.com/MariaDB/server/commit/9b10b4181f70634c9fab96ff0a225147b64396d7
I freely admit not understand what I'm doing, but the build error was gone with this change

The commit message "update libmariadb, and fix debian packaging for client libs" is void of all useful information. It does not state why you did something and what the problem was it fixes or what it attempts to do. There is no comments embedded into the code itself either. So if is difficult to review if your changes were correct or not when there is no messages of intention. The git subject just states the obivious that is already visible in the diff.

The bug report does not either have any context, link to an actually failing build etc so it is hard to debug with this as a starting point. Please provide more information.

I do not remember, sorry. you can revert and see where it fail. that's the easiest way to do. As for "why", we update libmariadb submodule, and this brought a couple of errors, that I had to fix without any idea of the script that I touched. debian packaging is not my strength. The commit that probably broke it is this : https://github.com/MariaDB/mariadb-connector-c/commit/e34a595ce2bff180518df42023dad0b6188b6233 . Since we package with cpack elsewhere, it is only carefully written debian scripts that can ever break.

Hello!

Debian packaging is sometimes a bit tricky and I completely understand if you don't know exactly what you are doing, no blame there. However writing git commits should be your everyday bread and there is no excuse to commit stuff without meaningful commit messages. In the slides 17-20 at https://www.slideshare.net/ottokekalainen/git-best-practices-2016 you can read what I recommend about git commit messages in general.

I will give a shot at building current 10.2 master, see what fails and perhaps fix this.

Fixed for 10.3 in https://github.com/MariaDB/server/commit/f7740b6604ae324be20feab192008f015c139498
and for 10.2 in https://github.com/MariaDB/server/commit/eded6243bc4796ab44e70403edd059d32225f589

I waited for buildbot run to finish over the night, but it didn't and I don't have time to prolong this so I pushed to masters now. The change is pretty safe and very unlikely to produce regressions.