Congrats on the kid!

I came away with the impression that the focus was on trying something new in order to learn and that a lack of transactions in existing libs was merely the excuse to experiment. There is still the performance issue with his approach, which I am investigating out of curiosity, so I wouldn't worry about it replacing yours.

It's hard not to take criticism personally but there will always be people who think you code is shit and approach all wrong.

BTW - I'm still running a patched version of the latest redis-client in production. Any thoughts about the patch to fix the internal buffer length calculations (changing "Math.max(currentLength * 2, atLeast * 1.1);" to "Math.max(currentLength * 2, offset + atLeast * 1.1);")?