Martin Taillefer

@halter73 We run our pools usually with 1024 or 2048 entries so that we can recover from spiky traffic efficiently (if you imagine traffic spiking at 1K RPS and sometimes...

@BrennanConroy Seems like my benchmark wasn't right. I'm updating it now and will report back once done.

I've updated the implementation a bit and have a brand new more detailed benchmark. I'm attaching the benchmark as a zip file, and the README file in there shows the...

@halter73 I added the benchmark code as requested. I had to modify it to remove the comparison with the old version (since the code base only contains the new version...

I care a little, but I'm not overly concerned. In general, if teams care deeply about perf, they will not be using the legacy runtimes. However, we could put back...

This is by design. The intent here is that redaction will occur on the final string. not on the individual items in the enumeration. That is, we do Redact(Stringify(Enumerable)) and...

@alastairtree I believe the value of a Lazy model is for a small cache with a high likelihood of concurrent accesses to the same state. But in a high traffic...

@RobDeVoer It's very similar on the inside, but the API surface has been considerably simplified by hiding all the complications around how monomorphisation is achieved. And we have some ideas...

Anybody got any ideas? Thanks.

There's a PR out for a prototype of this.