From: Tom Lane <tgl@sss.pgh.pa.us>
To: exclusion@gmail.com
cc: pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #18340: BitString may break nodetoString() conversion for a
 raw parse tree
In-reply-to: <18340-4aa1ae6ed4121912@postgresql.org>
References: <18340-4aa1ae6ed4121912@postgresql.org>
Comments: In-reply-to PG Bug reporting form <noreply@postgresql.org>
	message dated "Tue, 13 Feb 2024 09:00:00 +0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2565844.1707838472.1@sss.pgh.pa.us>
Date: Tue, 13 Feb 2024 10:34:32 -0500
Message-ID: <2565845.1707838472@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/2565845.1707838472%40sss.pgh.pa.us>
Precedence: bulk

PG Bug reporting form <noreply@postgresql.org> writes:
> When a server compiled with -DWRITE_READ_PARSE_PLAN_TREES, the following
> query:
> SELECT x' 0';
> produces:
> ERROR:  did not find '}' at end of input node

> Observed starting from commit 40ad8f9de, which implemented
> WRITE_READ_PARSE_PLAN_TREES for raw parse trees.

Bleah.  That's presumably my fault, will take a look.

> More interesting results may be obtained (e.g.,
> WARNING:  outfuncs/readfuncs failed to produce an equal raw parse tree
> ), but with raw trees only, not with node trees stored in the database,
> considering core code.
> Still codesearch.debian.net shows that nodeToString() used by several
> extensions: pg_qual_stats, pgpool2, hypopg, pglogical; perhaps some of
> them use that conversion unsafely (without processing a node with
> transformExpr() or so).

I doubt there's any reason to consider this a critical problem
right now.  We should fix it of course, in the expectation that
we might need it to work someday.

			regards, tom lane