Anton Kirilov

> ... ones that run profiling before tests start and re-compile themselves based on profiling information. not sure that should be allowed either. I am pretty sure you mean `h2o`...

> i don't think it's an "unfair" advantage when this is a natural advantage JIT has in the real world and is a possible reason to choose JIT over static,...

> the reality i think is only a tiny fraction of devs out there are interested in those kind of micro-optimizations and would prefer to see a **realistic** comparison of...

IMHO all OS-level optimizations should be fair game, be it process priorities, processor affinities, usage of `io_uring`, etc. (and no, personally I would not use all of them). Probably the...

I am quite late to the party, but as the author of the `h2o` implementation (one of the problematic cases called out above) I would like to state categorically that...

@billywhizz Sorry, I don't see it, at least for `h2o` - prepared statements are used for both the fortunes and the single query tests, and in the latter case the...

@billywhizz I agree with your analysis - it presents a quite useful framework-independent technique to demonstrate that the results are consistent with the expectation that the single query test has...

It is probably worth mentioning that due to issue #3804, the fastest frameworks in the cached queries test might also be hitting the network bandwidth limit. For example, according to...

Keep in mind that batching is expressly disallowed by the TFB rules (except for the database updates test and only for the updates themselves). However, pipelining is permitted (and used...

It's impossible to do it with the upstream `libpq` - `PQsendQueryPrepared()` (or any other function) is going to return an error if it's called a second time while the first...