General Guidance for how to handle Integers greater than 64 bits?

[krpatter-intc]

Is there any general guidance on how to use the frictionless schema when it comes to representing fields that are BitInt's or Python numbers greater than 64 bits?

It seems maybe the integer field makes no assumptions on size, so perhaps it is reasonable to use it as is? However, I noticed some tools referencing this spec (like pandas) seem to convert these to string instead.

I was considering the use of object but the field states it must be a valid JSON object and I guess strictly speaking a string isn't really an object.

Any advice in this space?

[fjuniorr]

I've bumped into this problem before but never found any general advice.

For my use case at the time ended up creating a custom boolean bigint property and handling it accordingly.

[Seddryck]

My understanding of the specifications is that the integer type does not define the size of the integer. If you want to specify the size, there's no need to create a new type: you can use the format field instead.

It's common practice to extend the set of accepted format values for integers with options like i8, i16, i32, i64, or i128.

[krpatter-intc]

Thanks, yes, I am thinking to proceed with this as the specified type.