[Firebird-devel] RE: Error msg when inserting into a computed field.

[Claudio V. C. <cv...@us...>]

> -----Original Message-----
> From: fir...@li...
> [mailto:fir...@li...]On Behalf Of Frank
> Schlottmann-Goedde
> Sent: Viernes 29 de Junio de 2001 12:04
> To: fir...@li...
>
> > There is another failure in  CF_ISQL_14 where
> >
> > create table t5 (af computed by (1+2));
> > insert into t5 values(10);
> >
> > will fail with
> > > Statement failed, SQLCODE = -804
> > >
> > > Dynamic SQL Error
> > > -SQL error code = -804
> > > -Count of columns does not equal count of values
> > >
> >
> > instead of
> >
> > < Statement failed, SQLCODE = -151
> > <
> > < attempted update of read-only column

Some weeks ago, the firebird queen (the same that shared dolls with Ada
Lovelace in her infancy <g>) asked me to provide a modification to allow the
following to happen:

create table tc(a int, b computed (a + 1));
commit;
insert into tc values(a);

In few words, for lazy people, to allow them insert in a table without
writing the list of fields EVEN THOUGH the table has computed fields, that
will be ignored. Now, you come with the paroxysm of the test cases,
emulating a fool that wants to insert a value in the unique and read-only
field of the table. Well, I have two suggestions:
- Is there any damage if I add a SHORT or USHORT to the struct req is
dsql.h? I think SSHORT is enough since the limit of field count in a table
is 32767. This field, called
req_computed_fcount
would be augmented in define_insert_action(), so instead of my code
if (field->fld_flags & FLD_computed)
	    continue;
it will be
if (field->fld_flags & FLD_computed)
{
   ++request->req_computed_fcount;
   continue;
}
hence by the time an error condition is detected:
if request->req_computed_fcount is > 0
then the message Ann wants is shown
else the traditional message Frank loves is shown.

Flames, diatribes, flack?
Anyway, do I need to create a new error message, Ann? I never have seen the
one you describe.

C.