The TAG_Compound now uses an OrderedDict internally for its mapping
between tag names and tags, to allow for much faster lookup by tag name.
Looking up tags by index number functions as before.
The .tags list has been changed into a read-only property (now a tuple).
The method for adding new tags to a TAG_Compound is by direct item
assignment.
A number of tests used the internal .tags list of TAG_Compound, they
have been changed to use item assignment.
This is very much a change in API, I don't know what other projects use this library, assuming that .tags is a mutable list. Certainly some (but not many) of the tests that make arbitrary compounds used that.
I'm building a project that is using the anvil library, and making these changes to NBT cut the run times in half.
The TAG_Compound now uses an OrderedDict internally for its mapping between tag names and tags, to allow for much faster lookup by tag name. Looking up tags by index number functions as before.
The .tags list has been changed into a read-only property (now a tuple). The method for adding new tags to a TAG_Compound is by direct item assignment.
A number of tests used the internal .tags list of TAG_Compound, they have been changed to use item assignment.
This is very much a change in API, I don't know what other projects use this library, assuming that .tags is a mutable list. Certainly some (but not many) of the tests that make arbitrary compounds used that.
I'm building a project that is using the anvil library, and making these changes to NBT cut the run times in half.
So it might be useful to those upstream?