Fwd: Jython 2.7 rc2 soft launch

Oops, forgot to reply-all.

Since I wrote this last email, the issue reported by Roland Walter has been
resolved. The fix - ensuring JYTHON_HOME is not pointing to an old release
and not having the old Jython on PATH - should be documented in the release
notes, since that will likely trip other people once they try to upgrade to
2.7 from 2.5.

- Jim


---------- Forwarded message ----------
From: Jim Baker <[email protected]>
Date: Mon, Apr 6, 2015 at 8:13 AM
Subject: Re: [Jython-dev] Jython 2.7 rc2 soft launch
To: Jeff Allen <[email protected]>


On Sun, Apr 5, 2015 at 9:30 AM, Jeff Allen <[email protected]> wrote:

> It installed ok on Windows 7 64-bit, from jython-installer-2.7-rc2,
> running with Java 1.8.0_40.
>
> Running regrtest -e (this time on Java 1.7.0_60), I get 22 failures. Not
> to detract from the hard work over the last 4 months or so, which has
> been great, the result feels a little rough for a release candidate. How
> is it on Linux (sorry, my KVM is playing up) and Mac?
>

Although occasionally flaky, most of the time we get 100% tests passing on
Ubuntu 14.10 and OX 10.10.2. Flakiness is seen on those platforms in
test_jython_initializer, test_socket and other networking tests,
test_tarfile. At this point, it seems to be mostly due to a problem in the
tests than what is being tested.

I would like to have converged to zero failures for Windows, but this feels
like probably a few more months of work at least. At this point, the
release candidate seems to be solid, but we did have one report from Roland
Walter that the jython.exe launcher is failing on Windows:
http://bugs.jython.org/msg9755 (http://bugs.jython.org/issue2297) when it
attempts to launch java itself in a subprocess. To be further investigated.


> Is 22 failures within expectations/comfort?


Yes, and working to either fix the underlying failure; fix the test itself;
or remove it entirely. Hopefully we can make good progress on this for 2.7.1


> Shouldn't all failures at
> least be "expected" failures at this stage, with a bug report tracking
> them?
>

That's a good idea - we should open up a bug per test, or at least group of
tests. It would be a good way to kick off 2.7.1 dev work as well.


> My bottom line is:
>
> 354 tests OK.
>
> 13 tests skipped:
>
>      test__osx_support test_commands test_curses test_dbm test_lib2to3
>
>      test_pickle test_pipes test_sax test_smtpnet test_socketserver
>
>      test_subprocess test_urllib2net test_urllibnet
>
> 7 skips unexpected:
>
>      test__osx_support test_commands test_dbm test_lib2to3 test_pickle
>
>      test_pipes test_sax
>
> 22 tests failed:
>
>      test___all__ test_bat_jy test_chdir test_classpathimporter
>
>      test_file2k test_httpservers test_import test_inspect
>
>      test_jython_initializer test_netrc test_py_compile test_runpy
>
>      test_select test_shutil test_socket test_ssl test_sys test_sys_jy
>
>      test_tarfile test_urllib2 test_zipfile test_zipimport_jy
>
> 22 fails unexpected:
>
>      test___all__ test_bat_jy test_chdir test_classpathimporter
>
>      test_file2k test_httpservers test_import test_inspect
>
>      test_jython_initializer test_netrc test_py_compile test_runpy
>
>      test_select test_shutil test_socket test_ssl test_sys test_sys_jy
>
>      test_tarfile test_urllib2 test_zipfile test_zipimport_jy
>
>
> Of these, the following 19 are hard failures (repeatable at the prompt
> with jython -m ...):
> test___all__
> test_bat_jy    (passé, but is there a test_exe_jy to take its place?)
>

I should have removed test_bat_jy from testing. The test itself should be
renamed to something like test_jython_launcher and then tested for both
jython.py and jython.exe, given that it seems reasonable.


> test_chdir
>

Should be fixed, per your comments on http://bugs.jython.org/issue2307

test_classpathimporter
> test_file2k
> test_httpservers    (may be security s/w interference)
> test_import
> test_jython_initializer
>

Flaky test


> test_netrc
> test_py_compile
> test_runpy    (repeatable, but an unlink() error suggesting poor hygiene)
> test_shutil
> test_ssl
>

Flaky


> test_sys
> test_sys_jy
> test_tarfile    (some unlink(), some file mode)
> test_urllib2
> test_zipfile    (all unlink())
> test_zipimport_jy
> ---
>
> test_inspect fails in test_findsource_binary under regrtest, which is
> skipped when it runs at the prompt. Odd.
>

Agreed about that


>
> test_gc is very noisy ("gc: The PyObject-subclass blah-blah should
> either implement Traverseproc or ...") but doesn't actually fail.
>

This is new with Stefan's work on supporting visibility of gc. I expect to
see this to get quieter.


>
> You are skipping test_subprocess, but that actually passes for me on
> Windows (takes 10 minutes to run, tho'.)
>

Running test_subprocess requires enabling the subprocess resource with
--use. This is enabled by default with ant regrtest (or ant test).



-- 
- Jim

jim.baker@{colorado.edu|python.org|rackspace.com|zyasoft.com}
twitter.com/jimbaker
github.com/jimbaker
bitbucket.com/jimbaker

------------------------------------------------------------------------------
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