Very cool but also more interesting is using Perl for any newish project.

It is hard to find maintainers or developers knowing Perl nowadays. Especially in data science related projects as python has been the de-facto tool for that field for some time now.

The video from the presentation on this: https://youtu.be/tdx9leN2kBg?si=uYa7xsOvpoonjaxB

This was part of the MySQL Belgian Days that was organized in the days before FOSDEM.

If I were in vgrippa's place, I would have not produced flamegraphs, but treemaps. It's the far superior choice for visualising profiling/query plans and the like, no matter along which property one would measure.

Munging the mysql output into something that is delectable as input to kcachegrind is straight-forward. Screenshots: https://ddg.gg/?q=%21i+kcachegrind

Hah I did this for SQL Server execution graphs at one point (via profiler or extended events) and then I of course realized that anything that goes more than one or two events deep is just hellish SQL anyway, love to see the execution plan version that's much smarter.

Thanks for sharing!

Do you have slides from your FOSDEM presentation? The video has this funny angle :(.

Thanks for posting!