Обсуждение: syntactically correct query gives ERROR: failed to assign all NestLoopParams to plan nodes

Поиск
Список
Период
Сортировка

syntactically correct query gives ERROR: failed to assign all NestLoopParams to plan nodes

От
Stephan Springl
Дата:
Hi,

running
   psql -f NestLoopParams.sql
gives
   psql:NestLoopParams.sql:16: ERROR:  failed to assign all NestLoopParams to plan nodes
   psql:NestLoopParams.sql:24: ERROR:  failed to assign all NestLoopParams to plan nodes

and does not process the seemingly correct query

select u.zielnummer, gnummer.v
   from u
     left join b on b.btext ~ '^verr\. R(\d+)$'
     left join lateral (select regexp_replace (b.btext, 'verr\. R', '')::int v)
       as gnummer on true
     left join lateral (select * from r where buchnummer = gnummer.v)
       as r on true;

No results are given, not even row count or headers, just the error 
message given above.

Everything needed to reproduce is in the file NestLoopParams.sql. The 
tables do not need to be populated with data.  The problem was verified 
with Postgresql version 13.1 on a fresh debian sid x86_64 installation as 
well as a 13.1 compiled under 32 bit i386 (Debian/Linux).  Master from the 
posgresql git repository (as of today's commit 
f9900df5f94936067e6fa24a9df609863eb08da2) shows the same behaviour if 
compiled under 32 bit i386 Debian/Linux.

I will be happy to provide more information if needed or helpful.

Thank you for your work on such an extraordinary piece of software.

Regards
Stephan
Вложения

Re: syntactically correct query gives ERROR: failed to assign all NestLoopParams to plan nodes

От
Tom Lane
Дата:
Stephan Springl <springl-psql@bfw-online.de> writes:
> running
>    psql -f NestLoopParams.sql
> gives
>    psql:NestLoopParams.sql:16: ERROR:  failed to assign all NestLoopParams to plan nodes
>    psql:NestLoopParams.sql:24: ERROR:  failed to assign all NestLoopParams to plan nodes

Thanks for the report!  This turns out to be not entirely trivial
to fix.  I've started a discussion about it at [1].  One way or
another we should have a fix in place for February's releases.

            regards, tom lane

[1] https://www.postgresql.org/message-id/171041.1610849523%40sss.pgh.pa.us