maria-developers team mailing list archive

[Sergei Golubchik <serg@xxxxxxxxxxxx>]

Hi, Zardosht!

On Jan 15, Zardosht Kasheff wrote:
> Hello all,
> 
> In MySQL/MariaDB 5.1, when a timestamp field was used as a key, we
> used an integer compare to compare the fields. We assumed all integer
> fields (including date, time, and timestamp) could use an integer
> comparison to compare the fields, and that they were either 1, 2, 3, 4
> or 8 bytes long.
> 
> We now see in MariaDB 5.3 and beyond that changes have been made to
> the timestamp field, such that the length of the field may be
> something other than 1, 2, 3, 4,or 8 bytes.
> 
> Given a timestamp field of N bytes, how are we meant to compare them?
> Are some meant to be integer comparisons? Are others supposed to be
> memory comparisons?

It's stored as 4-byte bigendian number of the integer part of the
timestamp, and then as 0- to 4-byte bigendian number for the fractional
part. Because both parts are stored as bigendian numbers, I suppose, you
can compare timestamps either as numbers or with memcmp.

See Field_timestamp_hires::store_TIME().

But really the API-correct way is to use Field::key_type() method and
compare the values using the specified method. It's more future-proof
than to hard-code comparison method mapping based on the field types.

Regards,
Sergei