Timestamp fractional seconds internal oracle representation.

[acautin]

Hi I am seeing wrong results when using oci_util:from_dts/1 function for the fractional part. The problem seems to be matching using little endian instead of big when decoding the value.

Also #5 doesn't seem to be correct as I can read the data in binary format.

[c-bik]

@acautin

In

erloci/src/oci_util.erl

Line 287 in 480fe13

, Ns:4/little-unsigned-integer-unit:8>>

and

erloci/src/oci_util.erl

Line 295 in 480fe13

, Ns:4/little-unsigned-integer-unit:8, TzH:1/integer-unit:8, TzM:1/integer-unit:8>>

...is it giving you correct result if you replace Ns:4/little-unsigned-integer-unit:8 with Ns:4/big-unsigned-integer-unit:8

[c-bik]

#5 is an old a bug, it probably doesn't apply any more. I will try to verify and close that later!
👍 for find it!

[c-bik]

Also need to update this test case with better verification for correctness of API

erloci/test/erloci_test.erl

Lines 963 to 966 in 480fe13

	{{C2Y,C2M,C2D}, {C2H,C2Min,C2S}} = oci_util:from_dts(C2),
	{{C3Y,C3M,C3D}, {C3H,C3Min,C3S}, C3Ns} = oci_util:from_dts(C3),
	{{C4Y,C4M,C4D}, {C4H,C4Min,C4S}, C4Ns, {C4TzH,C4TzM}} = oci_util:from_dts(C4),
	{{C5Y,C5M,C5D}, {C5H,C5Min,C5S}, C5Ns} = oci_util:from_dts(C5),

Currently API correctness isn't checked.

[acautin]

For the fractional part (Which was what I was interested at the moment) yes it gives me the correct results if I use big there. For now I am using only the datetime conversion taking the first 7 bytes and adding the fraction (if there is any) myself. We can replace it later when this is fixed. https://github.com/K2InformaticsGmbH/dderl/blob/master/src/dderloci_utils.erl#L142-L146

[c-bik]

This will deprecate most of oci_util APIs

For example: