Yeah, that is the problem. I knew it only affected the new importer in the test
suite, but thought it might just confuse if I drew the distinction at that
point.
Anyway, I have a fix which I am testing, but in doing so it threw up some
strange things in itself related to output filters. I'm not sure yet if this
was caused by something I was doing wrong in my test harness, or whether I have
uncovered a new separate problem. I would have got to it last night, but had to
spend the night removing a BotNet off my web site and tightening up
the security of the site. :-(
Jim Gallacher wrote ..
> I just noticed that when it hangs, the Apache process pegs the CPU at 98%.
>
> When the test initially hung I just killed the test.py, but didn't think
> to actually stop Apache. It took me a little while to figure out why my
> computer was *so* sluggish this morning.
>
> Jim
>
> Jim Gallacher wrote:
> > I just ran the tests with r427382 Apache 2.0.55 mpm-prefork.
> >
> > It *does not* hang with the traditional importer.
> >
> > It *does* hang when I enable the new importer in the tests with
> > PythonOption('mod_python.future.importer *')
> >
> > Jim
> >
> >
> > Graham Dumpleton wrote:
> >> FYI. The change I made for memory leak described in:
> >>
> >> https://issues.apache.org/jira/browse/MODPYTHON-181
> >>
> >> has caused a problem with other changes I made for:
> >>
> >> https://issues.apache.org/jira/browse/MODPYTHON-155
> >>
> >> This is causing test suite to hang on the tests:
> >>
> >> test_req_handler
> >> test_accesshandler_add_handler_to_empty_hl
> >>
> >> The specific problem is caused by code which uses req.add_handler().
> >> If you are using 3.3 from trunk, as long as you aren't using that method
> >> you should be okay.
> >>
> >> I'll try and work out an alternate way of doing what I needed. :-(
> >>
> >> Graham
> >>
> >
> >
