Обсуждение: [COMMITTERS] pgsql: Doc: desultory copy-editing for v10 release notes.

Doc: desultory copy-editing for v10 release notes.

Improve many item descriptions, improve markup, relocate some items
that seemed to be in the wrong section.

Branch
------
master

Details
-------
https://git.postgresql.org/pg/commitdiff/749eceff4a1f9740391b126c81af9fd4bf3b1eaa

Modified Files
--------------
doc/src/sgml/release-10.sgml | 825 ++++++++++++++++++++++---------------------
1 file changed, 432 insertions(+), 393 deletions(-)

On 2017/07/10 9:11, Tom Lane wrote:
> Doc: desultory copy-editing for v10 release notes.
>
> Improve many item descriptions, improve markup, relocate some items
> that seemed to be in the wrong section.
>
> Branch
> ------
> master
>
> Details
> -------
> https://git.postgresql.org/pg/commitdiff/749eceff4a1f9740391b126c81af9fd4bf3b1eaa

I see you updated text for the partitioning item:

@@ -1574,7 +1553,7 @@
       <para>
        Add table <link linkend="SQL-CREATETABLE-PARTITION">partitioning
        syntax</> that automatically creates partition constraints and
-       <command>INSERT</> routing (Amit Langote)
+       handles routing of tuple insertions and updates (Amit Langote)
       </para>

Although I like the new text better, I'm afraid that we don't support
routing updates yet, only inserts.

create table p (a int) partition by list (a);
create table p1 partition of p for values in (1);
insert into p values (1);

update p set a = a + 1;
ERROR:  new row for relation "p1" violates partition constraint
DETAIL:  Failing row contains (2).

Routing of updates is being worked on for PG 11 [1].

Attached patch removes "and updates".

Thanks,
Amit

[1] https://commitfest.postgresql.org/14/1023/

Amit Langote <Langote_Amit_f8@lab.ntt.co.jp> writes:
> On 2017/07/10 9:11, Tom Lane wrote:
>> Doc: desultory copy-editing for v10 release notes.

> I see you updated text for the partitioning item:

>         syntax</> that automatically creates partition constraints and
> -       <command>INSERT</> routing (Amit Langote)
> +       handles routing of tuple insertions and updates (Amit Langote)

> Although I like the new text better, I'm afraid that we don't support
> routing updates yet, only inserts.

Hm?  We correctly handle updates that don't change the partition key
columns, as well as deletes, no?  The previous text made it sound
like *only* the insert case worked properly.

It might be worth mentioning that you can't move a row into another
partition via UPDATE.  Or maybe that's more detail than we need here.
Not sure.

            regards, tom lane

On 2017/07/10 12:36, Tom Lane wrote:
> Amit Langote <Langote_Amit_f8@lab.ntt.co.jp> writes:
>> On 2017/07/10 9:11, Tom Lane wrote:
>>> Doc: desultory copy-editing for v10 release notes.
>
>> I see you updated text for the partitioning item:
>
>>         syntax</> that automatically creates partition constraints and
>> -       <command>INSERT</> routing (Amit Langote)
>> +       handles routing of tuple insertions and updates (Amit Langote)
>
>> Although I like the new text better, I'm afraid that we don't support
>> routing updates yet, only inserts.
>
> Hm?  We correctly handle updates that don't change the partition key
> columns, as well as deletes, no?

That's true.

>  The previous text made it sound like *only* the insert case worked properly.

So, as long as an UPDATE doesn't change the partition key of a tuple, it's
being "routed" correctly, that is, put back into the same partition.

To me, the phrase "routing updates" meant "re-routing" a tuple when the
partition key change requires it, but I may be wrong.

> It might be worth mentioning that you can't move a row into another
> partition via UPDATE.  Or maybe that's more detail than we need here.
> Not sure.

Maybe, we can leave that detail out of the release notes.

Thanks,
Amit