Hi Oystein.
Øystein Grøvlen wrote:
> Evgeny Potemkin wrote:
>> Hi Oystein,
[skip]
>> But anyway, I suggest you to make the calculation more clear by
>> setting the key_length to 0 and then to sum all key part lengths. Te
>> same as the group part in the create_tmp_table do.
>
> Good point. I have committed a new version of the patch which compute
> the key_length the way you suggest.
Ok.
>
> Note that key_length from the group part that you refer to is probably
> not correct since it does not take the length bytes into account.
Most probably no one groups over a blob field, thus this bug isn't manifested
itself. Feel free to create a test case & report it.
> However, it does not seem that key_length is used at all in that context
> since assigning an arbitrary value to key_length does introduce any
> failures in the test suite.
You see, the code you've fixed exists in 5.0+ also, but 6.0 started relying on
it. So it's better to fill them correctly.
>
> Thanks,
>
> Øystein
>
>>
Regards, Evgen.
