Facebook open sources Haxl

We previously had a custom DSL and it outgrew it's DSL-ness. The DSL was really good at one thing (implicit concurrency and scheduling io), and bad at everything else (cpu, memory, debugging, tooling). The predecessor was wildly successful and created new problems. Once all those secondary concerns became first order, we didn't want to start building all this ecosystem stuff for our homemade DSL. We needed to go from DSL to, ya know, an L. So the question is which...

If you understand the central idea of Haxl, I don't know of any other language that would let you do what Haxl in Haskell does. The built in language support for building DSLs (hijacking the operators including applicative/monadic operations) -really- shines in this case. I would -love- to see haxl-like implicit concurrency in other languages that feel as natural and concise. Consider that a challenge. I thought about trying to do it in C++ for edification/pedagogical purposes but it's an absolutely brutal mess of templates and hackery. There may be a better way, though.

Interpreted code was no longer cutting it for perf reasons, and any time you create your own language you end up reinventing the entire tool chain (debuggers, profilers, etc.). Haskell provides so much functionality in the language itself and has mature solutions to the other issues plaguing us in FXL, so it was a natural choice.

That said, the way it currently works is more like the first. You can think of the entire haxl run (program) as an AST that is given to the execution. It expands as much of the AST as possible (anything that's not IO), and anywhere it needs IO it enqueues those requests to be scheduled. Once it's explored as much as possible, it aggressively schedules the IO (deduping, batching, and overlapping the calls). Once it all comes back, it unblocks the AST where it can, and repeats the process.

This isn't necessarily the optimal scheduling (as you point out, unblocking each part of the tree as each result comes in might be better). It was specifically designed to make it easy to play with this kind of stuff later. Since the concurrency is entirely implicit the implementation is entirely abstracted away.

Query sharding is at the data source layer, which Haxl doesn't delve into. It's up to each data source integration with Haxl to do the appropriate routing/etc.

Hope you find it useful!

I actually find "distilled reference with links to source" a fantastically valuable view. I've no objection to providing some sort of combined view, but let's not lose what we have in a quest for consolidation. I've no idea if that's what you meant or not, and don't mean to put words in your mouth of course, just expressing a concern.