* Use Concurrent Dictionaries and ignore case on them to avoid ToLower
* remove single array allocation
* allocate GetClosures differently
* Use JsonReader for closures
* add comment
* use isdefined
* more readonly with less allocations
* sorts
* fmt
* use element type when making an array
* first pass
* Fix reading for object and array
* serialization works?
* Fix closure parsing
* things are way faster
* Deserialize is async
* fmt
* renames
* remove deserialize threads
* fmt
* faster to use ordinal compare
* serialization looks okay
* fix closure writing
* fixes
* use possibly different values for compute id
* use closure parser on download
* memory test for blobs
* decompose serialization
* fmt
* fmt
* for decomposing, values should be used instead of original
* set id after computing it
* redo more closure parsing
* fix memory test
* don't throw on try get deserialized
* fmt
* fix integration tests
* fix tests
* disable memory blob storage by default
* put back ?
* merge fixes and delete worker threads
* fmt
* serialization of old floats pass
* serialization of old floats pass
* rename class
* uncomment and fix tests
* don't allow things to exist in the closure table that doesn't exist
* detach blob tests
* rename serializer
* revert
* fix blob tests again
Adam Hathcock
22f4945d71