Open jentfoo opened 8 months ago
I have considered this further, I have not decided how I want to address this. There are three options:
Right now I lean towards option 2. This does not seem like a behavior that should need configuration. And the Commit / Load situation is probably contrived enough that it's not worth the proactive file size lost to encode the struct id.
We will keep considering this, and welcome feedback if anyone has any.
Not encoding the struct id does not cause an issue with de-serialization, but it can result in future writes to not be able to optimize within the file. For example if one record is added before each
Commit()
, it's likely that there would be no field de-duplication even if all the structs were the same.