Обсуждение: Parser doesn't grok unqualified array element

Has this come up before?  6.4.2 and current sources both have
this problem:
select count(*) from pg_proc where pg_proc.proargtypes[0] = 701;
works, butselect count(*) from pg_proc where proargtypes[0] = 701;
fails with ERROR: Unable to locate type name 'proargtypes' in catalog

The grammar doesn't seem to have a case that allows for a subscripted
attribute name without a relation name in front of it.

It looks like fixing this might be as easy as making the "ColId"
cases in a_expr, b_expr, possibly other places include an
opt_indirection item like columnElem does.  But maybe there's
more to it than meets the eye?
        regards, tom lane

> Has this come up before?  6.4.2 and current sources both have
> this problem:
> 
>     select count(*) from pg_proc where pg_proc.proargtypes[0] = 701;
> works, but
>     select count(*) from pg_proc where proargtypes[0] = 701;
> fails with ERROR: Unable to locate type name 'proargtypes' in catalog
> 
> The grammar doesn't seem to have a case that allows for a subscripted
> attribute name without a relation name in front of it.
> 
> It looks like fixing this might be as easy as making the "ColId"
> cases in a_expr, b_expr, possibly other places include an
> opt_indirection item like columnElem does.  But maybe there's
> more to it than meets the eye?

No, it is that easy.  For some reason, no one has done it yet.  Our TODO
list has:
* array index references without table name cause problems

I am sure the complexity of yacc grammar rules have kept some away from
fixing this. 


--  Bruce Momjian                        |  http://www.op.net/~candle maillist@candle.pha.pa.us            |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026
 

> Has this come up before?  6.4.2 and current sources both have
> this problem:
>   select count(*) from pg_proc where pg_proc.proargtypes[0] = 701;
> works, but
>   select count(*) from pg_proc where proargtypes[0] = 701;
> fails with
>   ERROR: Unable to locate type name 'proargtypes' in catalog
> 
> The grammar doesn't seem to have a case that allows for a subscripted
> attribute name without a relation name in front of it.
> It looks like fixing this might be as easy as making the "ColId"
> cases in a_expr, b_expr, possibly other places include an
> opt_indirection item like columnElem does.  But maybe there's
> more to it than meets the eye?

It has been reported, and is probably on the ToDo list as something. I
have been carrying it on my personal ToDo for a while, just to make
sure it doesn't get lost.

I would try your solution if I were fixing it, which I'm not yet. Go
fer it dude!
                   - Tom