rdflib-3.2.1 test issues for python3

[idella]

tests for python3.2 ALL fail. Having looked into the processes, it appears to come down to a class or two, or some functions in a class or two, not having compatibility with python3.2.

from the build log;

Skipping implicit fixer: set_literal Skipping implicit fixer: ws_comma Can't parse docstring in build-3.2/lib/rdflib/graph.py line 49: ParseError: bad input: type=24, value='%', context=('', (49, 40)) Can't parse docstring in build-3.2/lib/rdflib/graph.py line 77: ParseError: bad input: type=24, value='%', context=('', (77, 11)) Can't parse docstring in build-3.2/lib/rdflib/graph.py line 102: ParseError: bad input: type=24, value='%', context=('', (102, 35)) Can't parse docstring in build-3.2/lib/rdflib/graph.py line 106: ParseError: bad input: type=24, value='%', context=('', (106, 35)) Can't parse docstring in build-3.2/lib/rdflib/graph.py line 110: ParseError: bad input: type=24, value='%', context=('', (110, 35)) Can't parse docstring in build-3.2/lib/rdflib/graph.py line 624: ParseError: bad input: type=24, value='%', context=('', (624, 11)) Can't parse docstring in build-3.2/lib/rdflib/resource.py line 77: ParseError: bad input: type=24, value='%', context=('\n\n\n\n\n\n$ Can't parse docstring in build-3.2/lib/rdflib/term.py line 527: ParseError: bad input: type=24, value='%', context=('\n\n\n\n\n\n\n\$ Can't parse docstring in build-3.2/lib/rdflib/term.py line 529: ParseError: bad input: type=24, value='%', context=('', (529, 28)) Can't parse docstring in build-3.2/lib/rdflib/plugins/parsers/notation3.py line 141: ParseError: bad input: type=24, value='%', cont$ Can't parse docstring in build-3.2/lib/rdflib/plugins/parsers/notation3.py line 144: ParseError: bad input: type=24, value='%', cont$ Can't parse docstring in build-3.2/lib/rdflib/plugins/parsers/notation3.py line 339: ParseError: bad input: type=24, value='%', cont$ warning: build_py: byte-compiling is disabled, skipping.

Source compiled.

• Testing of dev-python/rdflib-3.2.1 with CPython 3.2... version is 3.2 the PYTHONPATH is build-3.2/lib/ Running nose with: --attr= test rdflib --where=./ --with-doctest --doctest-extension=.doctest --doctest-tests EEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE

Every single tests bails out. Most commonly cited are term.py and graph.py.

This; class URIRef(Identifier): in rdflib-3.2.1/rdflib/term.py && from rdflib.term import URIRef, BNode, Literal, Variable in rdflib-3.2.1/rdflib/init.py are commonly cited.

It appears the instances of ParseError: bad input: I highlighted are taking place in the compiling and or prep of the build-3.2 folder, the conversion for python3 anyway.

Hmmmm no facility to attach the log.

running build running build_py creating build-3.2 creating build-3.2/lib creating build-3.2/lib/rdflib copying build/src/rdflib/namespace.py -> build-3.2/lib/rdflib copying build/src/rdflib/store.py -> build-3.2/lib/rdflib copying build/src/rdflib/serializer.py -> build-3.2/lib/rdflib

---

and so forth creating build-3.2/lib/rdflib/plugins/parsers/rdfa/transform copying build/src/rdflib/plugins/parsers/rdfa/transform/headabout.py -> build-3.2/lib/rdflib/plugins/parsers/rdfa/transform copying build/src/rdflib/plugins/parsers/rdfa/transform/init.py -> build-3.2/lib/rdflib/plugins/parsers/rdfa/transform creating build-3.2/lib/rdflib/plugins/serializers copying build/src/rdflib/plugins/serializers/nquads.py -> build-3.2/lib/rdflib/plugins/serializers copying build/src/rdflib/plugins/serializers/rdfxml.py -> build-3.2/lib/rdflib/plugins/serializers copying build/src/rdflib/plugins/serializers/trix.py -> build-3.2/lib/rdflib/plugins/serializers copying build/src/rdflib/plugins/serializers/init.py -> build-3.2/lib/rdflib/plugins/serializers copying build/src/rdflib/plugins/serializers/turtle.py -> build-3.2/lib/rdflib/plugins/serializers copying build/src/rdflib/plugins/serializers/nt.py -> build-3.2/lib/rdflib/plugins/serializers copying build/src/rdflib/plugins/serializers/xmlwriter.py -> build-3.2/lib/rdflib/plugins/serializers copying build/src/rdflib/plugins/serializers/n3.py -> build-3.2/lib/rdflib/plugins/serializers Fixing build-3.2/lib/rdflib/namespace.py build-3.2/lib/rdflib/store.py build-3.2/lib/rdflib/serializer.py build-3.2/lib/rdflib/excep$ Skipping implicit fixer: buffer Skipping implicit fixer: idioms Skipping implicit fixer: set_literal Skipping implicit fixer: ws_comma Fixing build-3.2/lib/rdflib/namespace.py build-3.2/lib/rdflib/store.py build-3.2/lib/rdflib/serializer.py build-3.2/lib/rdflib/excep$ Skipping implicit fixer: buffer Skipping implicit fixer: idioms Skipping implicit fixer: set_literal Skipping implicit fixer: ws_comma

goes south at this point which is where the Can't parse docstring errors begin

[takluyver]

It looks like you're running the tests on the unconverted source (--where=./).

The CI tests are passing, although the same parse errors show up the build logs. They run with --where=./build/src. https://jenkins.shiningpanda.com/rdflib/job/rdflib/PLATFORM=debian6,TOXENV=py32/lastBuild/console

[idella]

seems I spoke too soon again., At the same time you said use --where the gentoo dev found it.

The copy process from build/src -> build-3.2/lib/ it seems left the tests folder behind. This then leaves a src/build folder which has had 2to3 run thru it, and a build-3.2 folder without the test suite. Needless to say the gentoo default traffic director didn't know that, so it needed an extra prompt in the ebuild to set it straight.

[takluyver]

I think it's intentional that the test suite isn't copied to the lib/ directory - it's not part of the library, after all. I'll close this issue as I think everything's working as intended, but let us know if you disagree. Thanks for testing it and reporting the problem.