Re: [HACKERS] Bug in get_partition_for_tuple

On 2017/03/27 23:54, Robert Haas wrote: > On Wed, Mar 22, 2017 at 4:00 AM, Amit Langote > wrote: >>> You're right, fixed that in the attached updated patch. >> >> Added this to the partitioning open items list, to avoid being forgotten >> about. >> >>

Re: [HACKERS] Bug in get_partition_for_tuple

On Wed, Mar 22, 2017 at 4:00 AM, Amit Langote wrote: >> You're right, fixed that in the attached updated patch. > > Added this to the partitioning open items list, to avoid being forgotten > about. > >

Re: [HACKERS] Bug in get_partition_for_tuple

On 2017/03/13 14:41, Amit Langote wrote: > On 2017/03/13 14:31, Jeevan Ladhe wrote: >> However, In following comment in your test: >> >> -- check routing error through a list partitioned table when they key is >> null >> >> I think you want to say: >> >> -- check routing error through a list

Re: [HACKERS] Bug in get_partition_for_tuple

Hi Jeevan, On 2017/03/13 14:31, Jeevan Ladhe wrote: > Hi Amit, > > I was able to reproduce the crash, and with the attached patch the crash > goes > away. Also, "make check-world" passes clean. > > Patch looks good to me. Thanks for the review. > However, In following comment in your test: >

Re: [HACKERS] Bug in get_partition_for_tuple

Hi Amit, I was able to reproduce the crash, and with the attached patch the crash goes away. Also, "make check-world" passes clean. Patch looks good to me. However, In following comment in your test: -- check routing error through a list partitioned table when they key is null I think you want

[HACKERS] Bug in get_partition_for_tuple

Just observed a crash due to thinko in the logic that handles NULL partition key. Absence of null-accepting partition in this case should have caused an error, instead the current code proceeds with comparison resulting in crash. create table p (a int, b char) partition by list (b); create table