> I think that's fair to say. That branch increases write concurrency compared to the begin-concurrent branch (which is embedded in server database Bedrockdb), but not by as much as we had hoped. On this test: https://sqlite.org/src/artifact/0c6bc6f55191b690 With 2rw 1ro, server-mode write TPS increase 140% compare to begin-concurrent. 160% read/write TPS. I think this is huge improvement. Is the recent begin-concurrent branch get extra 40% TPS compare to 2017-07-31 version ?