Ross A. Baker

I think splitting it up is the way, but before we invest more in it: 1. Are the dissenters satisfied? It was controversial in the day, and got a particularly...

That is a great question that I have not measured to adequately answer. Will also be hard to benchmark, because whatever impact we see isolating it may not matter in...

Preliminary benchmark on a response prelude (i.e., version / status / headers). We're winning on throughput and memory pressure. ``` [info] Benchmark Mode Cnt Score Error Units [info] RenderBench.http1Encoder thrpt...

/cc @zmccoy, who got a tour of a half-baked version on Friday

An unresolved tension here is that our parser is character-based, and our encoding would become byte-based. This is going to rear its head with HTTP/2, I think.

Sketched out a plausible `Http1Decoder`. One implementation wraps cats-parse. Compared it to Ember's parser. We use cats-parse extensively for individual headers, but all the backends have their own hand-written parsers...

> A lot of http is defined in terms of side-effecting algorithms, Do you have an example? Things like HPACK in HTTP/2 updating a table?

I'm kind of skittish about this one. There's no "DSL laws", but we have advertised that these are substitutable, and that will no longer be so: ```scala Ok() === Response(Ok).pure...

That strategy is roughly what I had in mind. The relationship is already broken: `Date`, `Server`, and `User-Agent` are all added by various backends. All of this is good for...

This is just a fit check for the otel4s API. Don't take it too seriously... yet.