Not much happening this week, what with the holidays. I took advantage of the break to handle some long-overdue code reviews and code improvements to Myria.

• Jingjing Wang has added resource profiling to Myria. We can now measure the resource consumption of each operator during query execution. (Unfortunately, these data are not yet available from the website.) See myria#656 and myria-web#241.

• Dominik Moritz implemented binary copy for Postgres. Database inserts are now substantially faster, especially for floating point or doubles where we could not use the COPY mode before (turns out double → string → double is not an identity-preserving transform. Duh.) See myria#667 and myria#669.

  Dominik also improved the performance of the profiler using the new interface. myria#672

• Shumo Chu is getting Myria ready for Apache licensing — we just need to get rid of a few pesky GPL’ed dependencies. Tentatively, it looks like we have easy switch-in replacements that might also yield a bit faster query execution. myria#658

• I upgraded our continuous integration to Travis-CI’s new Docker-based containers. This lets us re-enable caching and also tests that execute multicore and faster! myria#673 and myria#674.

• As part of these upgrades I cleaned up and better tested several operators.

In short: nothing much to see here, but Myria will hopefully work faster and better. One of Sophie’s sample queries sped up from 12m to 4m, and the 3x improvement is entirely due to the binary inserts!

We have also begun interviewing candidates for the Myria software engineer position.