Cory Benfield
Cory Benfield
So the first thing to note is that, in both cases, the bulk of the time is spent in `time.sleep`. That's...surprising. The top of the "withPyOpenSSL" stack, sorted by "time...
@begoldsm So one real possibility about where this slowdown is coming from is contention for the GIL. Substantially more of the stdlib `ssl` module is written in C, where it...
> In both cases it claims it only executed for < 10 seconds, even though the actual run time (real time) was about the same as it was previously (190...
Generally I suspect we'd rather have the appropriate tools for this in [cryptography](https://cryptography.io) rather than extend PyOpenSSL to be able to do this. @reaperhulk @alex is cryptography equipped to fill...
Thanks for this! However, I think you're approaching this backwards. Without support for EC keys in PKey, I don't think we should allow them to be loaded: I'd much rather...
@swift-nio-bot test perf please
Huh, no comparison there. Weird. I don't expect substantial performance improvements regardless: this only cleans up the dirty memory situation and our benchmark doesn't really test memory pressure.
To be clear, I think the interface to and public nature of `HTTP2StreamMultiplexer` should remain: it just shouldn't be a `ChannelHandler` any more.
In practical cases this _may_ reduce allocations. We'll get no more than one actual read per read cycle if the NIOSSLHandler is in front of us in the pipeline, in...
Note also that we accumulate bytes in more than one place, so you'd want to tweak almost all of these.