> Aaron Freeman wrote:
> > I see from your link that Sun uses an int, but couldn't
> that be arbitrary?
> I can't speak to whether the expert committee's decision was
> arbitrary or not. :) But once they've made a decision, it
> is a rule that must be implemented.
> > I don't believe Jakarta, and other implementations have
> this limit --
> > I just looked and they are using a long. This is a
> seriously limiting
> > factor and may require us to swap out our underlying servers from
> > Resin (which I really, really don't want to do).
> Where are you seeing this? Jakarta used to have Tomcat,
> which I assume is what you're referring to, but every
> reference I can find in Tomcat 6.x (BaseRequest, Request,
> RequestFascade, ApplicationHttpRequest,
> DummyRequest) returns int for getContentLength(). If someone
> has figured a way around the spec limitation, that's great
> news! Maybe newer versions of Java had changed this, but I
> didn't think you could alter method signatures.
Actually I mispoke, its "struts" not "jakarta". I am confused with their
organization of all the pieces. At any rate here is the thread where they
discuss the problem and resolution:
resin-interest mailing list