Re: Define jsonpath functions as stable - Mailing list pgsql-hackers

From Chapman Flack
Subject Re: Define jsonpath functions as stable
Date
Msg-id 29dc62f5-731c-6857-1633-495131a46b92@anastigmatix.net
Whole thread Raw
In response to Define jsonpath functions as stable  (Alexander Korotkov <a.korotkov@postgrespro.ru>)
Responses Re: Define jsonpath functions as stable
List pgsql-hackers
Hi,

On 7/29/19 10:25 AM, Alexander Korotkov wrote:

>  * like_regex predicate uses our regular expression engine, which
> deviates from standard.

I still favor adding some element to the syntax (like a 'posix' or 'pg'
keyword in the grammar for like_regex) that identifies it as using
a  different regexp flavor, so the way forward to a possible compliant
version later is not needlessly blocked (or consigned to a
standard_conforming_strings-like experience).

That would also resolve much of the case against calling that
predicate immutable.

It looks as if, in my first implementation of XQuery regexps, there
will have to be a "not-quite-standard" flag for those too, because
it turns out the SQL committee made some tweaks to XQuery regexps[1],
whereas any XQuery library one relies on is going to provide untweaked
XQuery regexps out of the box. (The differences only affect ^ $ . \s \S)

Regards,
-Chap


[1]
https://wiki.postgresql.org/wiki/PostgreSQL_vs_SQL/XML_Standards#XML_Query_regular_expressions



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: how to run encoding-dependent tests by default
Next
From: Robert Haas
Date:
Subject: Re: Is ParsePrepareRecord dead function