Re: Backpatch FK changes to 7.3 and 7.2? - Mailing list pgsql-hackers

Bruce Momjian wrote:
>
> Seems like a small reasonable patch to me, and several folks want it.
>

I am bit worried with this regression issue. I posted this about a week ago
but didn't get any response. Anyone else tested this? I re-run the test on a
fresh download of 7.3.2. Same. Jan, you mentioned something concerning an
error messages - is this issue causing the regression error?

This is the message I posted before:

I applied the patch to a 7.3.2 installation, and did a make clean, make,
make check. There is one regression error. Is this an expected behaviour? Or
did I do something wrong? See regression diffs:

*** ./expected/foreign_key.out  Sun Sep 22 02:37:09 2002
--- ./results/foreign_key.out   Sat Apr 12 20:44:54 2003
***************
*** 882,888 **** ERROR:  $1 referential integrity violation - key in pktable still
referenced from pktable -- fails (1,1) is being referenced (twice) update pktable set base1=3 where base1=1;
! ERROR:  $1 referential integrity violation - key referenced from pktable
not found in pktable -- this sequence of two deletes will work, since after the first there
will be no (2,*) references delete from pktable where base2=2; delete from pktable where base1=2;
--- 882,888 ---- ERROR:  $1 referential integrity violation - key in pktable still
referenced from pktable -- fails (1,1) is being referenced (twice) update pktable set base1=3 where base1=1;
! ERROR:  $1 referential integrity violation - key in pktable still
referenced from pktable -- this sequence of two deletes will work, since after the first there
will be no (2,*) references delete from pktable where base2=2; delete from pktable where base1=2;

Regards,
Michael Paesold





> --------------------------------------------------------------------------
-
>
> Jan Wieck wrote:
> > Stephan Szabo wrote:
> > >
> > > On Tue, 8 Apr 2003, Tatsuo Ishii wrote:
> > >
> > > > > > The changes I committed to address most of the FK deadlock
problems
> > > > > > reported can easily be applied to the 7.3 and 7.2 source trees
as well.
> > > > > >
> > > > > > Except for a slight change in the text of the error message that
gets
> > > > > > thrown "if one tries to delete a referenced PK for which a FK
with ON
> > > > > > DELETE SET DEFAULT exists" (it's a rare case, believe me), this
patch
> > > > > > would qualify for backpatching. The unnecessary FOR UPDATE lock
of
> > > > > > referenced rows could be counted as a bug.
> > > > > >
> > > > > > Opinions?
> > > > >
> > > > > Since I seem to suffer from these horrible deadlock problems all
the
> > > > > time, I'd like it to be backported to 7.3...
> > > >
> > > > Me too!
> > >
> > > As a note, this'll solve some of the deadlocks on fk update (generally
the
> > > key values aren't touched) but not insert related ones (two rows
inserted
> > > to the same primary key causing one to wait and possible deadlocks)
> > >
> > > In any case, why don't we get a patch against 7.3, and make an
> > > announcement and let people who are interested use it and test it.
With
> > > in-field testing it'd probably be safe enough. :)
> >
> > Here it is.
> >
> >
> > Jan
> >
> > --
> > #======================================================================#
> > # It's easier to get forgiveness for being wrong than for being right. #
> > # Let's break this rule - forgive me.                                  #
> > #================================================== JanWieck@Yahoo.com #