Re: Fix array-element quoting in postgres_fdw import statistics - Mailing list pgsql-hackers

From Etsuro Fujita
Subject Re: Fix array-element quoting in postgres_fdw import statistics
Date
Msg-id CAPmGK16FT3N2RA0+m5tN8+R+9Kx=2dEyj+T-cVoFscOQrdaDrQ@mail.gmail.com
Whole thread
In response to Re: Fix array-element quoting in postgres_fdw import statistics  (Etsuro Fujita <etsuro.fujita@gmail.com>)
List pgsql-hackers
On Sun, Apr 12, 2026 at 3:46 PM Etsuro Fujita <etsuro.fujita@gmail.com> wrote:
> On Sun, Apr 12, 2026 at 12:13 PM SATYANARAYANA NARLAPURAM
> <satyanarlapuram@gmail.com> wrote:
> > build_remattrmap() used quote_identifier() to format column names
> > for a text[] array literal passed to the remote pg_stats query.
> > quote_identifier() applies SQL identifier quoting, which doubles
> > double-quote characters but does not escape backslashes.  However,
> > inside a PostgreSQL array literal, backslash is an escape character.
> >
> > Column names containing backslashes (e.g. "a\b") were silently
> > mangled by the array parser—"a\b" became "ab"—causing the
> > WHERE attname = ANY('{...}') filter to miss those columns.  The
> > statistics import would then fail with a WARNING about missing
> > attribute statistics. This is a very corner cases because usually
> > backslash is not included in the column names. Anyways attached
> > a draft patch. Please take a look and let me know what you think.

> Good catch!  I will look into this as well.

I think your analysis is correct, but a simpler solution for this
issue is to use deparseStringLiteral() for column names, instead of
quote_identifier(), as proposed in [1].

Best regards,
Etsuro Fujita

[1] https://www.postgresql.org/message-id/CAJTYsWWGhVDFjr%2BsmdYdU-Q_TT9YMzXA4QcLCr7rizDOyrEEow%40mail.gmail.com



pgsql-hackers by date:

Previous
From: Alexander Korotkov
Date:
Subject: Re: Fix bug with accessing to temporary tables of other sessions
Next
From: David Rowley
Date:
Subject: Re: Cleanup shadows variable warnings, round 1