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
I have an Oracle 11g database, where EVENT ID's (integers) have been stored as the Oracle data type NUMBER(10,0). This indicates 10 digits of precision, at a scale of 0 (i.e. 10^0). The scale value being non-negative means that this value will always be integer, but when I look at the result of JDBC.coltypes(JDBC.Source(csr)) the column has been assigned the DataType of Float64.
This is also happening if I look at Tables.schema(JDBC.Source(csr)).
It's possible to convert these columns to Int, but I worry that if I had very large precision integers, and JDBC.jl treated them as Float64, then I might lose precision, which on ID's is a major problem.
The text was updated successfully, but these errors were encountered:
Looks like we have decimals being converted to floats, probably because nobody ever got around to implementing decimals. This is definitely a really bad and dangerous behavior. It seems a little odd that JDBC allows you to call getFloat here without warning or error, which is what is happening.
I'm not sure what Java would return here, probably this. I'll try to look into this, however it'll be difficult for me to set up a test for it, so if anyone else can aid with this it would be appreciated.
I have an Oracle 11g database, where EVENT ID's (integers) have been stored as the Oracle data type NUMBER(10,0). This indicates 10 digits of precision, at a scale of 0 (i.e. 10^0). The scale value being non-negative means that this value will always be integer, but when I look at the result of
JDBC.coltypes(JDBC.Source(csr))
the column has been assigned the DataType ofFloat64
.This is also happening if I look at
Tables.schema(JDBC.Source(csr))
.It's possible to convert these columns to
Int
, but I worry that if I had very large precision integers, andJDBC.jl
treated them asFloat64
, then I might lose precision, which on ID's is a major problem.The text was updated successfully, but these errors were encountered: