Sylvain Wallez wrote: > Daniel Fagerstrom wrote: >> Carsten Ziegeler skrev: >>> Hmm, I think it will, e.g. j.s.h.HttpServletRequest#getSession returns >>> a j.s.h.HttpSession while o.a.c.e.Request#getSession returns a >>> o.a.c.e.Session and overriding just the return type is not possible >>> in java. >> Didn't think about that. >> >> If we let o.a.c.e.Request#getSession have j.s.h.HttpSession as return >> type and then let it return something that implements o.a.c.e.Session >> You can always cast the return type of to a o.a.c.e.Session. >> >> Wouldn't the need to add a few casts be a rather minor inconvenience >> that is worthwhile for simplifying the framework and make it easier to >> connect to the rest of the world? > > This isn't a minor inconvenience, as it will break every code out there > that gets the session from a oace.Request, and there is *a lot*. Exactly.
> So -1, although I was initially in favor of this change. Yeah, I wasn't against it either and did not see this problem earlier on, but as soon as Daniel listed the changes in this vote thread, I noticed the problem :( > > One of the possible solutions is to make generally and officially > available the ServletRequest and ServletResponse entries of the object > model that are populated by the HttpEnvironment. This creates a > migration path towards the standard servlet interfaces. > They are already officially available in an http environment, so we could end up with the question what to do in other environments. Generating them? Carsten -- Carsten Ziegeler - Open Source Group, S&N AG http://www.s-und-n.de http://www.osoco.org/weblogs/rael/
