Re: [HACKERS] Restrict concurrent update/delete with UPDATE of partition key

On Thu, Apr 5, 2018 at 7:14 AM, Andres Freund <andres(at)anarazel(dot)de> wrote:
> Hi,
>
> On 2018-04-02 11:26:38 -0400, Robert Haas wrote:
>> On Wed, Mar 28, 2018 at 2:12 PM, Andres Freund <andres(at)anarazel(dot)de> wrote:
[....]
> I've attached a noticeably editorialized patch:
>
> - I'm uncomfortable with the "moved" information not being crash-safe /
> replicated. Thus I added a new flag to preserve it, and removed the
> masking of the moved bit in the ctid from heap_mask().
>
> - renamed macros to not mention valid / invalid block numbers, but
> rather
> HeapTupleHeaderSetMovedPartitions / HeapTupleHeaderIndicatesMovedPartitions
> and
> ItemPointerSetMovedPartitions / ItemPointerIndicatesMovedPartitions
>
> I'm not wedded to these names, but I'l be adamant they they're not
> talking about invalid block numbers. Makes code harder to understand
> imo.
>

These names are much better than before, thanks.

One concern -- instead xxxMovedPartitions can we have
xxxPartitionChanged or xxxChangedPartition?

xxxMovedPartitions looks (at least to me) like partitions are moved. In other
databases, there is maintenance command to move a partition from one tablespace
to another, current naming is fine as long as we don't support the
same, but if we do then this names will be confusing, comments/thoughts?

Regards,
Amul