For what it's worth: I'm +1 on Sean's JUnit 3.7 patches, thanks to both Sean and dIon for taking care of this. I'm suprised we haven't addressed that sooner.
I'm -0ish on the default header patch. I haven't had a chance to look at it in detail (which is why I haven't commented on it before) but IIRC, it makes both HttpClient and BaseHttpMethod store a list of default headers, and when HttpClient.executeMethod is invoked, the default headers are in the client are added to the method, yes? I'm bothered by that for a couple of reasons, but primarily by the hidden, persistent change that HttpClient.executeMethod then makes to HttpMethod. (I also think that the state management of HttpMethod is probably in need of revision anyway, but that's a different thread.) IMO, storing a set of default headers in HttpClient, and adding them (as regular headers) to HttpMethod in executeMethod is OK, even desirable. Storing default headers (that are not cleared by recycle()) in HttpMethod is questionable. Storing it in both places, and having HttpClient populate the *default* headers of HttpMethod is a bad idea. Thirdly, if anyone has a patch for the connection timeout issue that doesn't require JDK 1.4, or can make the build script only compile the JDK 1.4 extension to HttpConnection when 1.4 is available, I think that'd be good (so we could produce 1.2/1.3 and 1.4 builds with code modification). Or for that matter, if anyone would like to create a FAQ for that, explaining how to do the 1.4 version, that'd be good too. Frankly I never really see the timeouts, so it's not a big itch for me, and it's a bit hard for me to test. - Rod PS: My wife is 39 weeks pregnant with our second child, so I haven't had much time for jakarta stuff in the past couple of weeks, and I'll likely disappear for a couple of weeks real soon, but I'll be back.