Thread: Allow elog_node_display() to show INSERTs

Hello,

I just made this patch so I could debug and otherwise
poke around.

It allows elog_node_display() and friends to show
INSERT parse trees.

AFIK it works, in my use and after applying
to the latest cvs with:
   ./configure --enable-cassert ; make check

The patch does not have a corresponding regression
test.  I understand how to make a regular SQL-esque
regression test, but this is sorta a metadata kind
of test and I'm unclear on how to proceed or whether
there's even a testing framework for this sort of thing.

Karl <kop@meme.com>
Free Software:  "You don't pay back, you pay forward."
                  -- Robert A. Heinlein

"Karl O. Pinc" <kop@meme.com> writes:
> I just made this patch so I could debug and otherwise
> poke around.
> It allows elog_node_display() and friends to show
> INSERT parse trees.

This seems rather pointless unless we were going to undertake to make
outfuncs.c support *all* raw-grammar node types.

[ looks... ]  The comment at the head of that file suggests that we do,
but obviously we don't.  How many other node types would need to be
added to make the coverage complete?  Is it worth the trouble?

            regards, tom lane

I'm not interested in pushing for inclusion, this is
just an observation.

On 05/16/2007 03:20:54 PM, Tom Lane wrote:

> This seems rather pointless unless we were going to undertake to make
> outfuncs.c support *all* raw-grammar node types.

Isn't having something better than having nothing?
It's not like it matters at all one way or the other
to the end-user.  It's just for the developers. (?)


Karl <kop@meme.com>
Free Software:  "You don't pay back, you pay forward."
                  -- Robert A. Heinlein