INT in the
>>> output schema, so we want to lock down a type there. Hopefully, that
>>> becomes a standard across engines, and IRs. We are doing the same
>>> conversion in Coral [3], but hoping to wrap our discussion here before
>>> proceeding.
>>>
>
Rs. We are doing the same
>> conversion in Coral [3], but hoping to wrap our discussion here before
>> proceeding.
>>
>> * The more systems not agreeing on this conversion (to TINYINT or INT),
>> the more challenging it will be to fix this problem. Sometimes the schema
>> l
TINYINT or INT),
> the more challenging it will be to fix this problem. Sometimes the schema
> leaks transitively across tables and it becomes more difficult to retract,
> so now is the best time to decide on this.
>
> That said, first, I am hoping to get an alignment on supporting TINYI
