Mathieu Amiot

Right now both our parsers are kinda on-par (sometimes faster and sometimes slower) but yours is destroying mine on JSON fragments. I should look into it as well, I might...

Latest parser work: Again, some outliers mostly due to my laptop getting random wakeups ``` Running target/release/deps/nitox_parser_benchmark-0f39421b3886644e connect_parse time: [544.82 ns 546.93 ns 549.13 ns] change: [+7.4161% +8.2490% +9.0100%] (p...

I might have a solution working in https://github.com/YellowInnovation/nitox/commit/1a2037acca608b4908857e033e787ad3adb93d8b, what do you think @thedodd?

Now is fine, I've added a test with a typical use case and it passes but I'm not sure if it cover yours as well.

Hey! The other crate supports it simply for the reason that the TLS implementation is based on OpenSSL and it is kinda broken, especially if your local libssl is old...

Also, quick note but when acting as a client, you don't need to provide a root cert, only the host name is needed. The current code works fine in production...

Wow indeed, just looked at the mechanism that `gnatsd` is using and after looking around I didn't find any documentation related to this behavior (besides the issue you mentioned). (Ridiculous...

@wallyqs Alright, thanks for the clarification, duly noted! Also, same context, but what happens when the subsequent `INFO` messages have a `tls_required: false`. Should we downgrade to pure TCP? Or...

I agree that how it works currently is definitely not foolproof and relies on the user giving proper params for each tls/tcp mode. tls = url form / tcp =...

So far it doesn't complete the integration tests :(