[Stackless] Stackless 2.5 branch updated for Python 2.5.1c1

Richard Tew richard.m.tew at gmail.com
Sat Apr 14 17:13:33 CEST 2007


On 4/14/07, Hieu Hoang <hieu.d.hoang at gmail.com> wrote:
> On *nix there's a "make test" target for Python, it runs the test suit
> twice. Both trunk and 25-maint revision 54826 pass the test:

Ah, well I haven't touched trunk.  That could do with a merge as well,
but for now, I am ignoring it.  But if people want to be bleeding
edge, I could update it.

> About stackless part, stupid me didn't test 25-maint before svn
> update, so I don't know which fails due to the merge, sorry. How
> should I go back to an older rev?

This should be the official Stackless 2.5 release code:

  http://svn.python.org/projects/stackless/tags/python-2.5/

> The stackless tests:
> on linux 2.6.20-generic on amd64, libc6 2.5-0ubuntu14
> RuntimeError: cannot execute invalid frame with 'run_cframe': frame
> had a C state that can't be restored.
> RuntimeError: cannot execute invalid frame with
> 'gen_iternext_callback': frame had a C state that can't be restored.
> RuntimeError: cannot execute invalid frame with 'run_cframe': frame
> had a C state that can't be restored.
>     self.assertEquals(t.recursion_depth, 0)
> AssertionError: -1 != 0

This is not pretty :-)  It can be relied upon that if there are flaws
in the way Stackless does what it does, then they will be identical
and reproducible on all platforms.  Especially when seen as failures
in the unit tests.

So my initial naive assumption is that there must be something about
your platform which is influencing the way Stackless is going about
its business.  If you get similar or the same errors when running
these unit tests with the tagged 2.5 release of Stackless, then I
think we can assume this is the case.  It is known to be stable on all
supported platforms and to have no unit test failures.

Some more results from other platforms should hopefully confirm that
the problems are isolated to yours.  We'll see what we can do after we
get that feedback.

Thanks,
Richard.

_______________________________________________
Stackless mailing list
Stackless at stackless.com
http://www.stackless.com/mailman/listinfo/stackless



More information about the Stackless mailing list