Re: Jython 2.7 rc3 soft launch
Jim Baker <[email protected]>
| Newsgroups | gmane.comp.lang.jython.devel |
|---|---|
| Message-ID | <CAOhO=aM3p9XTv9TKVbSnW2ap7SBit7zeg81OE3noYi2mLwUw=A@mail.gmail.com> |
On Wed, Apr 22, 2015 at 12:58 AM, Jeff Allen <[email protected]> wrote: > It wasn't a stop the launch message, just reporting what I saw. I'm happy > to go with others' judgement. > My feeling is that we are complete for 2.7.0, but we will not know until users start trying this release out on their own code, first with a final RC3 and then the final release itself. Right now, we are running any number of tests code that attempts to push various limits, often to ensure that CPython will not crash with a segfault. But on Jython, this is hard - if not impossible! - to do. (I have seen only one crash FWIW in the last two years, in Clamp's use of Java classes, through JNI, for writing jars. So nothing Jython could have prevented in any event.) So IMHO having a test report back a failure that could be wrapped better as a Python exception is OK. > > I am *instinctively* a bit uncomfortable that anything go out with failing > or skipped tests. But my instincts have been shaped by the sort of subjects > where just one counterexample proves you wrong, *not* by much experience > launching software products. I'm interested in how this level of test > conformance maps to user experience. I think Jython is pretty good now and > that we won't be buried in bug reports. > We still have 229 open bugs... so it's just a question of having even more ;) There will always be more for any successful project with some level of complexity, which Jython obviously exceeds by some a couple of orders of magnitude or so. (Well at least without formal verification methods...) > I'd like to see testing give us a clearer result. At the moment it always > fails and you have to work out what kind of failure to know where you > stand. On the one hand I think defects should be visible and painful; on > the other I'd like to have a test that only shouts when something > unexpected happens. (And I'd like a build bot to do the shouting.) Now I > understand regrtest better (since the weekend!), and which parts we added, > I think it offers a way of saying "everything is as expected (but don't > forget these problems)", vs. "you just broke something", with only the > latter being shouted. It's nearly there. > Agreed! A major goal of 2.7.1 should be make our tools work better for us, whether that's migrating to https://github.com/jython; moving to Maven-driven bhttps://github.com/jythonuilds (but still using Ant for driving pieces like Antlr); and having CI with Travis, including on all PRs. - Jim ------------------------------------------------------------------------------ BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT Develop your own process in accordance with the BPMN 2 standard Learn Process modeling best practices with Bonita BPM through live exercises http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_ source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF _______________________________________________ Jython-dev mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/jython-dev