Re: [HACKERS] Is this a bug? or am I doing some thing wrong?

Terry Mackintosh wrote:

> [...]
> shell_connection-> o.accno = c.accno and
> shell_connection-> o.stockno = s.stockno;
> ERROR: DefineQueryRewrite: rule plan string too big.
> shell_connection=>
>
> Is this a bug? or am I doing some thing wrong?
> And if it is a bug, is it fixed in 6.4.1?

It's not a real bug, it's a side effect from the fact that a
tuple cannot span multiple disk blocks.

>
> What's a 'rule plan string'? the part I typed? if so, how long can it be?
> I also tried typing it all as one (wraped) line, but same thing.

First it's wrong. The system stores a string representation
of the rules querytree in pg_rewrite (not the rules plan).
Anything the rule system deals with are querytrees. Those
ones coming from the parser for the actual query and those
ones generated for the rule actions and qualifications. A
querytree consists of cascaded node structures in memory. The
function nodeToString() builds a string which is stored in
pg_rewrite and stringToNode() is used to rebuild the in
memory structures from that.

These strings are very explanative and thus very long
compared against the query they represent. They must fit into
one text type attribute and from the limitation above, the
resulting pg_rewrite tuple must fit into one (8k by default)
disk block.

While fixing the rule system for v6.4 I thought about
implementing continuation rows for rewrite rules. But there
are at least ideas out how tuples could eventually span disk
blocks in the future, so I left that problem untouched
waiting for this general solution.

Jan

--

#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#======================================== jwieck(at)debis(dot)com (Jan Wieck) #