For the Travis CI integration, it runs all tests except the test that requires the RST Discourse Treebank data which is skipped. This integration is defined in .travis.yml.
It also adds Atlassian Bamboo CI integration so that all the tests (including the one that Travis skips) are run on an ETS internal bamboo instance. This is required since we cannot make the RST Discourse Treebank data publicly available (it's something that must be purchased from the Linguistic Data Consortium). The coverage report is also computed in Atlassian Bamboo rather than Travis CI. The integration is defined in ci/run_tests.sh.
Replaces conda_requirements.txt with environment.yaml which now uses the new crfpp package from our public ets conda channel and also installs python-zpar via pip which is more reliable.
Removes requirements.txt and updates setup.py to make pip install -e . disregard requirements. This is because there's no crfpp PyPI package (yet).
This PR closes #32 .
.travis.yml.ci/run_tests.sh.conda_requirements.txtwithenvironment.yamlwhich now uses the newcrfpppackage from our publicetsconda channel and also installspython-zparviapipwhich is more reliable.requirements.txtand updatessetup.pyto makepip install -e .disregard requirements. This is because there's nocrfppPyPI package (yet)..gitignorefile.