On Tue Sep 25 14:52:00 2007, forbin wrote:
Show quoted text> for a large blob of JSON, it would probably be more efficient to pass it
> in as a scalar ref than having to copy it.
it would still have to be copied when the underlying JSON module gets it, so this like a micro
optimization to me.
Is it benchmarkable? Compared to the number of allocations done to inflate that memory, and
the fact that perl's data structure probably have a 10x overhead, saving one allocation +
memcpy even for something like a 1mb json document seems like it won't make a real
difference.