On Wed, Aug 7, 2013 at 10:10 PM, Glyph <glyph@twistedmatrix.com> wrote:

On Aug 7, 2013, at 1:06 PM, Laurens Van Houtven <_@lvh.io> wrote:

FWIW, Maciej commented that they don't expect to be *faster*, because they actually care about the JSON spec or checking return values. Things which ultrajson conveniently ignores :-) It's pretty easy to do the wrong thing fast, but that seems to be a common theme in this thread.

I just said "improved", i.e. 2.1 ought to be faster than 2.0, no claims relative to anything else :).

I'm aware that ultrajson is pretty bad, but cjson had comparable (if slightly slower) performance. Does it also have correctness and security problems?

It's also important to note that in a microbenchmark, JIT warmup can be a significant factor in PyPy. If you're interested in the performance of short-running applications (eg. command-line tools) then this is very relevant, but for a long-running server process you're unlikely to care about the warmup overhead but rather the sustained performance thereafter.

--
mithrandi, i Ainil en-Balandor, a faer Ambar