Re: [HACKERS] [sqlsmith] subplan variable reference / unassigned NestLoopParams (was: [sqlsmith] Failed assertion in joinrels.c)

2015-08-04 Thread Tom Lane 
Andreas Seltenreich seltenre...@gmx.de writes:
 Tom Lane writes:
 Well, I certainly think all of these represent bugs:
 3 | ERROR:  plan should not reference subplan's variable
 2 | ERROR:  failed to assign all NestLoopParams to plan nodes

 These appear to be related.  The following query produces the former,
 but if you replace the very last reference of provider with the literal
 'bar', it raises the latter error.

Fixed that, thanks for the test case!

 ,[ FWIW: git bisect run ]
 | first bad commit: [e83bb10d6dcf05a666d4ada00d9788c7974ad378]
 | Adjust definition of cheapest_total_path to work better with LATERAL.
 `

There's still something fishy about your git bisect results; they don't
have much to do with what seems to me to be the triggering condition.
I suspect the problem is that git bisect doesn't allow for the possibility
that the symptom might appear and disappear over time, ie it might have
been visible at some early stage of the LATERAL work but been fixed later,
and then reintroduced by still-later optimization efforts.

regards, tom lane


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

[HACKERS] [sqlsmith] subplan variable reference / unassigned NestLoopParams (was: [sqlsmith] Failed assertion in joinrels.c)

2015-08-02 Thread Andreas Seltenreich 
Tom Lane writes:

 Well, I certainly think all of these represent bugs:

  3 | ERROR:  plan should not reference subplan's variable
  2 | ERROR:  failed to assign all NestLoopParams to plan nodes

These appear to be related.  The following query produces the former,
but if you replace the very last reference of provider with the literal
'bar', it raises the latter error.

select 1 from
  pg_catalog.pg_shseclabel as rel_09
  inner join public.rtest_view2 as rel_32
  left join pg_catalog.pg_roles as rel_33
  on (rel_32.a = rel_33.rolconnlimit )
  on (rel_09.provider = rel_33.rolpassword )
left join pg_catalog.pg_user as rel_35
on (rel_33.rolconfig = rel_35.useconfig )
where ( ((rel_09.provider ~~ 'foo')
  and (rel_35.usename ~* rel_09.provider)));

,[ FWIW: git bisect run ]
| first bad commit: [e83bb10d6dcf05a666d4ada00d9788c7974ad378]
| Adjust definition of cheapest_total_path to work better with LATERAL.
`

regards,
Andreas


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers