python3 code seems shockingly inefficient

[joernhees]

Comparing the typical travis run times of py2 and py3 tests it seems py3 is on avg. taking 2x as long.
This only partially seems to be due to the 2to3 runs and a wild guess of me is that it has to do with some items() -> list(items()) transformation.

It seems this becomes most apparent in this test:

Python3.4: 86 seconds

test.test_dawg.test_dawg(RDFTest(uri=rdflib.term.URIRef('http://www.w3.org/2009/sparql/docs/tests/data-sparql11/syntax-update-2/manifest#syntax-update-other-01'), name='syntax-update-other-01', comment=None, data=None, graphdata=None, action='file:///home/travis/build/RDFLib/rdflib/build/src/test/DAWG/data-sparql11/syntax-update-2/large-request-01.ru', result=None, syntax=True),): 86.3655s

Python2.7: 49 seconds

test.test_dawg.test_dawg(RDFTest(uri=rdflib.term.URIRef(u'http://www.w3.org/2009/sparql/docs/tests/data-sparql11/syntax-update-2/manifest#syntax-update-other-01'), name=u'syntax-update-other-01', comment=None, data=None, graphdata=None, action=u'file:///home/travis/build/RDFLib/rdflib/test/DAWG/data-sparql11/syntax-update-2/large-request-01.ru', result=None, syntax=True),): 49.0514s

Todos:

• check why test is so slow in general (profiler-runs)
• check if it can be decomposed

[gromgull]

Here: py2 test in 2m09s - py3 in 4-6 minutes :(

https://travis-ci.org/RDFLib/rdflib/builds/196356635

(ignoring that one test failed, it's unrelated)

[gromgull]

They are comparable now: https://travis-ci.org/RDFLib/rdflib/builds/447210434

[ashleysommer]

This was closed as resolved, (I don't know if it resolved itself over time, with updates in Python3, or a fix was made in rdflib). I'm changing milestone from 6.0.0 to 5.0.0 for inclusion in the 5.0.0 changelog.