<div class="gmail_quote">On Wed, Aug 18, 2010 at 11:32 PM, Martin Pool <<a href="mailto:mbp@canonical.com">mbp@canonical.com</a>> wrote: <blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> <div class="im">On 19 August 2010 16:14, Alexander Belchenko <<a href="mailto:bialix@ukr.net">bialix@ukr.net</a>> wrote: >> Interesting. Can you elaborate on this or direct me to a thread >> describing this known issue? For example, is this a temporary known >> issue, or is it an issue-by-design? I am asking because I've been >> assuming (oops!) that all stable releases of bzr are subjected to pretty >> much the same regression suite on all supported platforms. Is that a >> bad assumption? </div>We test on Windows, but we test before it's packaged up in to an exe. <div class="im"> > It's because of how py2exe'd python application works. > > Standalone bzr.exe is python application packaged into binary > executable with py2exe tool. So it can work without Python interpreter > installed in your system. > > Some parts of bzrlib tests expects to have the real Python interpreter > to succeed. This is just not true in the case of bzr.exe. So you can > expect those tests to fail. </div>I think it would be reasonable for those to be marked as skipped tests in the case where we don't have a regular python interpreter. If someone wants to tackle this you just need to add a new test dependency feature in tests.features, then call self.requireFeature in the relevant tests. -- Martin </blockquote></div> Thanks for clarifying that, Martin and Alexander. That makes perfect reason. A few months back I was running the selftest on Windows (using native python package of course) and I was wondering if there had been a change of policy since then. FWIW, I think this is fine. I would not expect a lot of value added in a special test suite for the py2exe'd bzr. Thanks ~M