On 04.11.2017 01:52, Nikita Glukhov wrote:
>> On 03.11.2017 15:07, Michael Paquier wrote:
>>
>> About the patch set, I had a look at the first patch which is not that
>> heavy, however it provides zero documentation, close to zero comments,
>> but adds more than 500 lines of code. I find that a bit hard to give
>> an opinion on, having commit messages associated to each patch would
>> be also nice. This way, reviewers can figure what's going out in this
>> mess and provide feedback.
> Sorry that comments and commit messages are still absent. I am going
> to do it
> in the next version of these patches where SQL/JSON constructors
> displaying
> will be fixed.
>
Attached the new version of the patches where displaying of SQL/JSON
constructor nodes was fixed. I decided not to invent new nodes but to
extend
existing FuncExpr, Aggref, WindowFunc nodes with new formatting fields that
give us ability to override default displaying in ruleutils.c. Also new
invisible CoercionForm was added for hiding casts in which FORMAT and
RETUNING
clauses are transformed.
>> Making things incremental is welcome as
>> well, for example in the first patch I have a hard way finding out why
>> timestamps are touched to begin with.
> Timestamp's code was touched to add support of two features needed for
> SQL/JSON
> .datetime() item method by standard:
> - TZH and TZM template patterns
> - datetime components recognition
>
> I absolutely agree that this should be in a separate patch.
>
TZH and TZM template patterns were moved into a separate patch.
--
Nikita Glukhov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company