You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* coerce all integer types to int64_t
- to remove integer ambiguity around 32/64bit integers, we coerce all
ints to int64_t.
* handle int64_t constraint/ ImageSequenceReference
* plumb uint64_t support all the way through the SDK
- When a uint64_t is serialized, it is bitwise & with INT64_MAX first to
avoid overflowing the integer. In order to do that comparison at
serialization time, the uint64_t needs to be stored as a uint64_t all
the way through the python bindings into the any mapping.
- the next step is to move the & operation up into the python bindings
and prevent people from storing invalid numbers in OTIO files.
* add negative numbers to test and compare values
Please enter the commit message for your changes. Lines starting
* add in C++ sdk tests
* add unicode unit test as well
* clean up big integers test
* add serialization support for nan, inf, etc
- this is to allow parity with the values in python files
* [de]serialize a variety of numeric types correctly
- add support in the serializer/deserializer for NaN, Inf, -Infinity etc
from rapidjson
- block pybind11 from automatically casting numbers that don't fit into
double or int64_t into bool
- add a unit test and test data to test the serialization and behavior
* Clean up invalid type/value exceptions in Any
- Adds specific messages that test if something is an out of range
integer, or an invalid type
- includes a list of supported types
* cleanup
* Add note to docs about supported numeric types
* Lint fixes
* Fix ImageSequenceReference cloning for int64_t
- The 'Writer' part of SerializableObject needed the same tightening as
the RapidJSON dispatcher
- this is used for things that serialize and compare JSON (like cloning
and the `is_equivalent_to` method.
- Because the ImageSequenceReference schema has `int` in it (which
should at some point be changed to int64_t), it was tripping this
check on certain platforms
* Some more minor tightening. ready to ship!
* use the big_int_type for python2 support
Co-authored-by: ssteinbach <[email protected]>
0 commit comments