On Thu, 2010-11-18 at 09:17 -0800, David Smiley @MITRE.org wrote: > > olegk wrote: > > > > On Thu, 2010-11-18 at 07:12 -0800, David Smiley @MITRE.org wrote: > >> In the example ClientExecuteSOCKS.java you showed me: > >> 1. Why isn't the SOCKS support built into httpclient instead of requiring > >> users to do it themselves? HttpClient is already invovled with Socket > >> level > >> TCP/IP stuff, with all the socket factory stuff going on, which is where > >> the > >> SOCKS support needs to be implemented. > > > > Feature-bloat is one of the biggest issues with HttpClient. I am usually > > very reluctant to include additional features that are conceptually not > > related to HTTP, especially if we do not have enough expertise within > > the project to support them. > > > > SOCKS protocol is not supported out of the box simply because we cannot > > adequately test it. If you are willing to contribute a patch for this > > feature as well as provide a reasonable test coverage for it, I will > > happily review and commit it to the repository. > > > > Testing this will require a SOCKS server to be stood up. A quick google > search points at java implementations, such as > http://sourceforge.net/projects/jsocks/ -- perhaps it could be run in a > test. If you expect this level of testing for SOCKS, presumably you have the > same expectation of HttpClient's support for HTTP proxies? i.e. is http > client standing up an http proxy server in the tests to actually check? If > you're actually not going the distance there, then I argue I shouldn't have > to do the same for SOCKS proxies either.
Mock based unit tests will do. Oleg --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
