Benchmark ldump

[girvel]

On a large sample complex data, compared to binser and bitser, with different compression and without, compare serialization/deserialization speed and output size

[girvel]

• more complex data input
• field names based on word dictionary
• extract random module
• sample that uses require (may not work with one of binser/bitser)