I am getting some super-good benchmarks here:

Mosquitto

$ ./benchmarks/single_pub_sub.js --header -q 1 -r 10000
mean, standard deviation, median, mode, runs
1317.389, 2621.9008969056513, 1088, 165, 10000

Mosca

$ ./benchmarks/single_pub_sub.js --header -q 1 -r 10000
mean, standard deviation, median, mode, runs
208.6827, 252.84635984089678, 185, 170, 10000

It seems Mosquitto implementation of the persistance does not really scale to multiple runs.

Thanks also to @davedoesdev and its outstanding https://github.com/davedoesdev/qlobber.

Excellent! What would be interesting is benchmarks for topics using wildcards.
Are you sure mosca was using TrieAscoltatore? Doesn't ascoltatore build MemoryAscoltatore unless it's specified?

Edit: Ah got it - you made the change in Ascoltatori to change the default.

Edit: Actually you forgot to do it ascoltatori.build so I don't think you were testing using TrieAscoltatore? I've made the changes in moscajs/ascoltatori#62

I am directly using Qlobber in the persistance support.
The latest version of Ascoltatori uses TrieAscoltatore (I think 0.7.3)

Il giorno mercoledì 26 giugno 2013, David Halls ha scritto:

Excellent! What would be interesting is benchmarks for topics using
wildcards.
Are you sure mosca was using TrieAscoltatore? Doesn't ascoltatore build
MemoryAscoltatore unless it's specified?

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/36#issuecomment-20029049
.

I think you missed out ascoltatori.build

Yes, you are right! I'm merging in your PR asap.

Il giorno mercoledì 26 giugno 2013, David Halls ha scritto:

I think you missed out ascoltatori.build

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/36#issuecomment-20030563
.