Greg Stein wrote:
>
> On Wed, Aug 16, 2000 at 02:44:41PM -0500, Karl Fogel wrote:
> > Greg Stein <gstein@lyra.org> writes:
> > > I'd highly recommend using a typedef such as svn_vernum_t to represent this.
> > > That gives you one good feature (semantics, leading to readability, type
> > > checking, maintenance, understandability) and one minor feature (ability to
> > > change the type if necessary; e.g. make it unsigned).
> >
> > That seems like a good idea.
> >
> > We had a recent phone conversation with JimB where we tentatively
> > decided not to make special types that just mirror native C types, but
>
> I would agree that making something like "svn_long_t" would be a bad idea :-)
> In all other cases, I actually think it makes a lot of sense because of the
> semantic value-add.
>
> > I think the reasons for that decision don't apply so well in this
> > case.
>
> Yup. A type name lends semantics to the code where a "long" or "char" fail.
>
> > Plus, it's always easier to change back from custom type to
> > native type, than the other direction. :-)
>
> Good point :-)
+1 to all of the above--thanks for making the distinction explicit. :)
--
Daniel Rall <dlr@collab.net>
http://collab.net/ | open source | do the right thing
Received on Sat Oct 21 14:36:06 2006