2014-03-08 16:26 GMT+04:00 Pierre Chapuis <catwell@xxxxxxxxxxxx>: > Hello, > > this is interesting. Do you have a benchmark that shows that? > > I wrote that FFI-based implementation but as far as I know > François Perrad's pure Lua implementation matches its speed > most of the time. > Hey, I even wanted to write you about my results so as to encourage you to keep up the good job you've done :) But as I did not have a clear benchmark with numbers, I hesitated to do that and later other things drew my attention... Unfortunately I still do not have an isolated benchmark... But I've read your README and you suggested to use pure Lua implementation for new projects. So we tried to go that route, but we got numerous trace aborts due to some NYI (I do not remember which exactly, but there were two of them with v2.0). Then we tried your implementation, and got everything JIT compiled, and it was faster. May be with v2.1 that situation will change. Our use case is a bit specialized, we have a known structure (defined in another messagepack objects elsewhere) and use only arrays. So in order to avoid ipairs we've supplied table classifier that says that any table is an array, not a hash. -- Regards, Konstantin