Re: Possible typo in an error message - Mailing list pgsql-hackers

From Daniel Gustafsson
Subject Re: Possible typo in an error message
Date
Msg-id 318D357F-9EB7-4CAC-A4C0-CF996958461A@yesql.se
Whole thread
In response to Re: Possible typo in an error message  (Andreas Karlsson <andreas@proxel.se>)
List pgsql-hackers
> On 10 Jul 2026, at 09:31, Andreas Karlsson <andreas@proxel.se> wrote:
>
> On 7/10/26 06:52, Kyotaro Horiguchi wrote:
>> While translating this message, I noticed the following error text:
>> parse_jsontable.c:644:
>>> errdetail("PATH name was %s not found in nested columns list.",
>> Is this perhaps meant to be
>> PATH name %s was not found in nested columns list.
>> instead?
>
> Agreed, or maybe:
>
> "Path name %s not found in nested columns list."
>
> From what I can see both forms are used in our error messages.

I would favor "was not found" since the JSON_TABLE support has a similar
wording for missing PLAN clauses: "PLAN clause for nested path %s was not
found".  Consistency within a feature is good.

>> Also, I wasn't quite sure whether the uppercase "PATH" in messages
>> such as "PATH name" has any particular significance. For the
>> translation, I ended up treating it as the ordinary word "path", since
>> I couldn't tell whether it was intended to refer to the SQL keyword or
>> just the general concept.
>
> Agreed, it does not look like it is a keyword so it should not be written in upper case.

I think it does refer to the PATH keyword in a NESTED PATH clause.  See below
for an example query from our regression tests:

SELECT * FROM JSON_TABLE(
       jsonb '[]', '$' AS path1
       COLUMNS (
               NESTED PATH '$' COLUMNS (
                       foo int PATH '$'
               )
       )
       PLAN DEFAULT (UNION)
) jt;

--
Daniel Gustafsson




pgsql-hackers by date:

Previous
From: Jim Jones
Date:
Subject: Re: Proposal: INSERT ... BY NAME
Next
From: Andreas Karlsson
Date:
Subject: Re: Proposal: INSERT ... BY NAME