[Maria-developers] which version MDEV-6991 GROUP_MIN_MAX optimization is erroneously applied in some cases
Hi Sergei, I suggest to move MDEV-6691 to 10.1: Reasons: 1. It will be easier to fix it on top of: MDEV-6991 GROUP_MIN_MAX optimization is erroneously applied in some cases So fixing before 10.0 is not desirable 2. The relevant code is different in 10.1. So it's not desirable to fix before 10.1. 3. The bug is not much important Thanks.
Hi, Alexander! MDEV-6691 seems to be unrelated. What bug did you mean? Regards, Sergei On Jun 02, Alexander Barkov wrote:
Hi Sergei,
I suggest to move MDEV-6691 to 10.1:
Reasons:
1. It will be easier to fix it on top of:
MDEV-6991 GROUP_MIN_MAX optimization is erroneously applied in some cases
So fixing before 10.0 is not desirable
2. The relevant code is different in 10.1. So it's not desirable to fix before 10.1.
3. The bug is not much important
Hi Sergei, On 06/02/2015 07:51 PM, Sergei Golubchik wrote:
Hi, Alexander!
MDEV-6691 seems to be unrelated. What bug did you mean?
Sorry for confusing. I meant this one: (MDEV-8229) GROUP_MIN_MAX is erroneously applied for BETWEEN in some cases
Regards, Sergei
On Jun 02, Alexander Barkov wrote:
Hi Sergei,
I suggest to move MDEV-6691 to 10.1:
Reasons:
1. It will be easier to fix it on top of:
MDEV-6991 GROUP_MIN_MAX optimization is erroneously applied in some cases
So fixing before 10.0 is not desirable
2. The relevant code is different in 10.1. So it's not desirable to fix before 10.1.
3. The bug is not much important
Hi, Alexander! On Jun 02, Alexander Barkov wrote:
Hi Sergei,
I suggest to move MDEV-6691 to 10.1:
Reasons:
1. It will be easier to fix it on top of: MDEV-6991 GROUP_MIN_MAX optimization is erroneously applied in some cases So fixing before 10.0 is not desirable 2. The relevant code is different in 10.1. So it's not desirable to fix before 10.1. 3. The bug is not much important
Okay. Taking into account that the bug was here, like, forever, and it only shows up when a date is stored in a string not as a pure date (not equal to the canonical string representation of a date), let's fix it in 10.1 together with other refactorings. We can always reconsider this, if needed. Regards, Sergei
participants (2)
-
Alexander Barkov
-
Sergei Golubchik