On Nov 19, 2007 11:44 AM, Simon Laws <[EMAIL PROTECTED]> wrote: > > > On Nov 19, 2007 11:29 AM, ant elder <[EMAIL PROTECTED]> wrote: > > > On Nov 19, 2007 10:22 AM, Simon Laws <[EMAIL PROTECTED]> wrote: > > > > > I've just committed changes to the Jetty and Tomcat host containers. > > > (r596243 and r596246) to ensure that URLs are formed correctly when > > > context > > > is included in the Node URI, e.g. > > > > > > http://localhost:8080/somecontext > > > > > > is combined with > > > > > > mycomponent/myservice > > > > > > to give > > > > > > http://localhost:8080/somecontext/mycomponent/myservice > > > > > > Tomcat was working OK in most circumstances but the original changes > > to > > > jetty meant it was exposing services as > > > > > > http://localhost:8080/somecontext/somecontext/mycomponent/myservice > > > > > > I've reverted the original changes so that the context is not set on > > the > > > server but is added to the servlet path in getURLMapping(String) as > > it > > > needs to be set back into the binding uri. If we want to set the > > context > > > to > > > the server we will have tomake more changes so that the binding can > > > manipulate the complete URL and not get it confused with the context > > when > > > the servlet is registered. > > > > > > > After this the calculator-distributed sample no longer works for me nor > > Tomcat using nodes with paths in the urls. Whats the reasons we can't > > get > > Jetty to work with the context path on the http server? I'm wondering > > if > > that might be easier, it also has to be the way the webapp host works as > > we > > can't change the context path of the webapp, but i don't want to go off > > looking at this if you've already tried it and hit blocking issues? > > > > ...ant > > > I'm seeing the calculator-distributed issue too. I've not hit a blocking > issue. To make contexts work as intended we need to be a bit smarter at the > way the URL is resolved as we need to resolve the URL provided by the > servler with the default URL that the server now has. If you want to look > into that that would be great as I haven't got back to it yet. > > Simon > I have a fix for the calculator-distributed problem. It's unrelated to the jetty problem and it to do with accidentally having loaded a composite twice. Need a test to prevent that. Just doing a full build now.
Simon Simon
