Re: jsonpath: Missing Binary Execution Path? - Mailing list pgsql-hackers

From David E. Wheeler
Subject Re: jsonpath: Missing Binary Execution Path?
Date
Msg-id C9C53667-D6B4-4B0A-ADD0-292388FFD2DB@justatheory.com
Whole thread Raw
In response to Re: jsonpath: Missing Binary Execution Path?  (Chapman Flack <jcflack@acm.org>)
List pgsql-hackers
On Jun 13, 2024, at 22:31, Chapman Flack <jcflack@acm.org> wrote:

> It's baked right into the standard grammar: || can only have a
> <JSON boolean conjunction> on its right and a <JSON boolean disjunction>
> on its left.
>
> && can only have a <JSON boolean negation> on its right and a
> <JSON boolean conjunction> on its left.

Wow.

> The case for ! is even more limiting: it can't be applied to anything
> but a <JSON delimited predicate>. That can be either the exists predicate,
> or, any other <JSON path predicate> but wrapped in parentheses.
>
> The language seems sort of gappy in the same way XPath 1.0 was. XPath 2.0
> became much more consistent and conceptually unified, only by that time,
> XML was old school, and JSON was cool, and apparently started inventing
> a path language.

I suppose that’s the reason for this design. But if these sorts of limitations were changed in XPath, perhaps SQL-Next
couldfix them, too. 

Thanks for citing the standard; super helpful.

D




pgsql-hackers by date:

Previous
From: Maxim Orlov
Date:
Subject: Bugfix and improvements in multixact.c
Next
From: Nazir Bilal Yavuz
Date:
Subject: Re: CI and test improvements