Re: [cx-oracle-users] float vs. int

[Marcos P. <msa...@gr...>]

But other interfaces do return the right thing, =BFdon't they? I mean,
odbc and SQL*Plus. (Can't check right now though)

El lun, 06-10-2003 a las 19:17, Anthony Tuininga escribi=F3:
> On Sat, 2003-09-27 at 21:27, Geoff Gerrietts wrote:
> > Quoting Geoff Gerrietts (ge...@ge...):
> > > What are the criteria for deciding when a NUMBER will be returned as
> > > a float, or when it will be returned as an int or long?
> >=20
> > Terribly sorry; found the answer in the mailing list archives.
>=20
> No problem. This isn't the first time the problem has appeared. It is
> one of those things that annoys me every time I think about it. But I
> haven't found a really useful solution so far.
>=20
> > I think for my uses, anything with a scale 0 < sys.maxint should be an
> > integer, and anything with a scale 0 > sys.maxint should be a long.
> > This does not violate the principle of least surprise, and it also
> > relieves me of the burden of post-processing every query result so
> > CORBA doesn't barf.
>=20
> That should be quite reasonable and simple to implement. I am assuming
> that you mean "value" when you state "scale" above, right? Does anyone
> have any objections to that implementation? Of course, it doesn't solve
> the problem that most people immediately complain about, which is "Why
> do I get a float back when I issue the query 'select count(*) from
> table'"? Unfortunately, Oracle returns a scale of "0" and a precision of
> "0" meaning no information is available as to whether an integer or a
> floating point number is being returned.
>=20
> > With decimals, it would be nice to work with fixed point rather than
> > floats, but that would likely be a substantial change in my
> > application: floats are everywhere, and in several places, I expect
> > them to be floats.
>=20
> True. I've considered adding an extension which would allow
> specification of what is to be returned. I still haven't come up with a
> reasonable name for it or whether I should overload setoutputsize() in
> some way. Essentially, it would mean the following:
>=20
> cursor.somefunc(1, cx_Oracle.INTEGER)
> cursor.somefunc(2, cx_Oracle.FLOAT)
> cursor.somefunc(3, cx_Oracle.FIXED_POINT)
>=20
> where of course, the constants defined above don't exist yet and
> "somefunc" needs to be named properly so that it makes sense. Any
> suggestions?
>=20
> > Thanks,
> > --G.
--=20
Marcos S=E1nchez Provencio <msa...@gr...>
www.burke.es