Re: SQL/JSON in PostgreSQL

2017-09-30 1:06 GMT+02:00 Nikita Glukhov <n(dot)gluhov(at)postgrespro(dot)ru>:

> On 29.09.2017 20:07, Pavel Stehule wrote:
>
> 2017-09-29 12:15 GMT+02:00 Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com>:
>
>>
>> 2017-09-29 12:09 GMT+02:00 Nikita Glukhov <n(dot)gluhov(at)postgrespro(dot)ru>:
>>
>>>
>>>
>>> I have some free time now. Is it last version?
>>>
>>> Regards
>>>
>>> Pavel
>>>
>>> Yes, this is still the latest version. Now I am working only on
>>> unfinished WIP
>>> patch no. 9, but I think it should be reviewed the last.
>>>
>>>
>>
>> ok
>>
>> Thank you
>>
>
> I have few queries and notes
>
> 1. Why first patch holds Gin related functionality? Can be it separated?
>
> Yes, it can be easily separated. Attached archive with separated GIN patch
> no.2.
>
> 2. Why Json path functions starts by "_" ? These functions are not removed
> by other patches.
>
> Originally, these functions were created only for testing purposes and
> should
> be treated as "internal". But with introduction of jsonpath operators
> jsonpath
> tests can be completely rewritten using this operators.
>

yes - it should be removed.

Probably separation to jsonpath and sqljson is not happy (or sqljson part
should not contains JSON_QUERY and related functions).

Why this code is in patch?

*+/********************Example functions for
JsonPath***************************/++static Datum+returnDATUM(void *arg,
bool *isNull)+{+<->*isNull =
false;+<->return<>PointerGetDatum(arg);+}++static Datum+returnNULL(void
*arg, bool *isNull)+{+<->*isNull = true;+<->return Int32GetDatum(0);+}+*
Regards

Pavel

> 3. What is base for jsonpath-extensions? ANSI/SQL?
>
> Our jsonpath extensions are not based on any standards, so they are quite
> dangerous because they can conflict with the standard in the future.
>
> This patch is pretty big - so I propose to push JSONPath and SQL/JSON
> related patches first, and then in next iteration to push JSON_TABLE patch.
> Is it acceptable strategy?
>
> I think it's acceptable. And this was the main reason for the separation
> of patches.
>
> I am sure so JSON_TABLE is pretty important function, but it is pretty
> complex too (significantly more complex than XMLTABLE), so it can be
> practiacal to move this function to separate project. I hope so all patches
> will be merged in release 11 time.
>
>
> --
> Nikita Glukhov
> Postgres Professional: http://www.postgrespro.com
> The Russian Postgres Company
>