Remembering Johannes Doerfert

I have very fond memories of meeting johannes for the first time in code when working on Polly. His writeup on reduction dependece elimination in the larger dependence analysis was very well thought out, and left a lasting impression on me on "how to document" polyhedral optimizatioms.

I met him that year at euroLLVM, and we spent a long time chatting about loop optimization, all the while making me try some local weissbeer :) he really made me feel welcome, since I didn't know many people in LLVM back then (it was my second month or so of contributing to LLVM).

We've run into each other at different LLVM dev events ever since, and he's always offered me sound advice, as I've walked through academia (Johannes, should I stay in academia or move to industry? - doesn't matter sid, you're managing your manager anyway, so pick a place where you work on something you like!)

I wish I'd grabbed another beer with him the last time I met him, when we had a friendly shouting match about polyhedral analysis for cache modelling. I miss his technical depth, openness, and friendliness.