Re: [Springframework-developer] IJdbcTemplate / IHibernateTemplat e / ISqlMapTemplate

[Colin S. <col...@ex...>]

I used to start my interface names with I, but a year or two ago stopped=20
when I realized that in fact I was almost always working against the=20
interfaces anyways, so it was silly to always have the prefix there (now=20
I generally add 'Impl' to the end of the implementation object, if there=20
is only one variant).

The I prefix is not going to be pretty here where most of our other=20
interfaces don't have it, but on the other hand, I don't like interface=20
names which seem to have no relation to the impl. objects either... I=20
would probably leave them as Ixxxx and add a note to the javadoc that it=20
is that way for historical reasons, as opposed to having a totally=20
different interface and impl name...


Kopylenko, Dmitry wrote:

>I'm not a big fan of Ixxx naming pattern for interfaces. +1 for renaming
>them to something else.
>
>Dmitriy.
>
>-----Original Message-----
>From: j=FCrgen h=F6ller [werk3AT] [mailto:jue...@we...]=20
>Sent: Thursday, February 05, 2004 11:45 AM
>To: spr...@li...
>Subject: [Springframework-developer] IJdbcTemplate / IHibernateTemplate =
/
>ISqlMapTemplate
>
>
>Everybody,
>=20
>Due to a recent request, I've just extracted IHibernateTemplate and
>ISqlMapTemplate interfaces, analogous to IJdbcTemplate that Rod introduc=
ed a
>while ago. They are intended to ease mocking for unit-testing DAOs, for
>those who want to mock at this level. For typical DAO implementations, t=
he
>standard template classes are good enough, allowing for mocks of the
>underlying persistence tool and live tests against a database.
>=20
>However, I wonder about the naming. We don't have the I* pattern anywher=
e
>else in the framework; on the other hand, we need to keep the old templa=
te
>names for compatibility (and some templates don't lend themselves for su=
ch
>an interface, like JdoTemplate). IHibernateTemplate etc doesn't worry me=
,
>but maybe a different name like HibernateOperations would be better?
>Essentially, those interfaces specify a set of data access operations fo=
r a
>particular persistence tool.
>=20
>The naming isn't too important, though, as it will not be the main API t=
hat
>people will work with; the template classes will be. We'll just let the
>templates implement those operation interfaces to allow for mocking them=
 if
>desired. BTW, iBATIS SQL Maps is all based on concrete classes rather th=
an
>interfaces, not allowing to mock the persistence tool itself.
>=20
>Juergen
>=20
>