Re: help with execnet

[Daniel Stender <debian@danielstender.com>]

On 27.12.2016 14:18, Daniel Stender wrote:
> On 27.12.2016 12:09, Tobias Hansen wrote:
>> Hi,
>>
>> I just want to add that execnet is marked for autoremoval on January 17.
>> Also, sagemath depends indirectly on execnet.
>> The bugs should be fixed before January 7 to avoid stuff getting removed
>> from testing.
>>
>> Best,
>> Tobias
> 
> For the current state, I've just test build again without any errors on amd64 incl. passing
> the build time tests [1].
> 
> Three FTBFS resp. failing tests have been reported so far [2,3,4], another one - also from DEP-8 -
> I've just added [5].
> 
> [2] is from another test build, [3] from the reproducible builds CI (where it currently doesn't appear),
> [4] and [5] from DEP-8 testing (where it never passed).
> 
> Thanks for comments,
> DS
> 
> [1] http://www.danielstender.com/uploads/execnet_1.4.1-3_amd64-2016-12-27T12:50:09Z.build
> 
> [2] https://bugs.debian.org/840823 (testing/test_gateway.py::TestBasicGateway::test_gateway_status_busy[thread-popen] FAILED)
> 
> [3] https://bugs.debian.org/846951 (testing/test_gateway.py::TestTracing::test_popen_stderr_tracing FAILED)
> 
> [4] https://bugs.debian.org/846952 (testing/test_channel.py::TestStringCoerce::test_2to3 FAILED)
> 
> [5] https://bugs.debian.org/849466 (testing/test_basics.py::test_stdouterrin_setnull[thread] FAILED)

... some progress, #849466 has been reported as an issue at Bitbucket, I've disabled/removed warnings for
the Python3 tests in DEP-8 and this vanishes (that could be restored if the issue is closed).

The other failure from DEP-8 (#846952 on Python 2.7) doesn't appear in a local DEP-8 autopkgtest [2]. This
tests gets skipped in a Python 2 enviroment, but deb/tests/control is all right. Strange. I've lowered the
severity (hardly RC).

#846951 is difficult to reproduce. It's from reproducible builds CI, but current it doesn't appear (nor any other
problem). I'm tending to lower this, too.

#840823 is confirmed, I have that in a clean new Debian testing on i386:
<cut>
________________________ TestBasicGateway.test_gateway_status_busy[thread-popen] ________________________

self = <test_gateway.TestBasicGateway instance at 0xb39811ec>
gw = <Gateway id='popen' receive-live, thread model, 0 active channels>

    def test_gateway_status_busy(self, gw):
        numchannels = gw.remote_status().numchannels
        ch1 = gw.remote_exec("channel.send(1); channel.receive()")
        ch2 = gw.remote_exec("channel.receive()")
        ch1.receive()
        status = gw.remote_status()
        assert status.numexecuting == 2  # number of active execution threads
        assert status.numchannels == numchannels + 2
        ch1.send(None)
        ch2.send(None)
        ch1.waitclose()
        ch2.waitclose()
        for i in range(10):
            status = gw.remote_status()
            if status.numexecuting == 0:
                break
        else:
>           pytest.fail("did not get correct remote status")
E           Failed: did not get correct remote status

testing/test_gateway.py:88: Failed
</cut>

DS

[1] https://bitbucket.org/hpk42/execnet/issues/31/test-failure-with-enabled-warnings

[2] http://www.danielstender.com/uploads/execnet-local-dep8.txt
 


-- 
4096R/DF5182C8
Debian Developer (stender@debian.org)
LPIC-1 (LPI000329859 64mz6f7kt4)
http://www.danielstender.com/