When you have a lots of objects with the same keys, JSON requires each key to be represented
over and over again, no matter how many times it is used.
UBJSON does exactly the same thing, which is very odd for a "binary" format is trying to save
space.
Just for the purpose of discussion, a decoder could maintain an array of 256 values.
: would set cache[] to value.
! would use cache[].
Pretty much this idea is used in the Erlang-oriented UBF, and it pays off very well there.
Was anything like this ever considered for UBJSON, and if not, what is the argument against it?
When you have a lots of objects with the same keys, JSON requires each key to be represented
over and over again, no matter how many times it is used.
UBJSON does exactly the same thing, which is very odd for a "binary" format is trying to save
space.
Just for the purpose of discussion, a decoder could maintain an array of 256 values.
: would set cache[] to value.
! would use cache[].
Pretty much this idea is used in the Erlang-oriented UBF, and it pays off very well there.
Was anything like this ever considered for UBJSON, and if not, what is the argument against it?