Merge lp:~coreygoldberg/autopilot/tox-and-packaging into lp:autopilot
Status: | Merged |
---|---|
Approved by: | Martin Pitt |
Approved revision: | 328 |
Merged at revision: | 327 |
Proposed branch: | lp:~coreygoldberg/autopilot/tox-and-packaging |
Merge into: | lp:autopilot |
Diff against target: |
95 lines (+24/-15) 6 files modified
.bzrignore (+3/-0) MANIFEST.in (+2/-1) bin/autopilot (+1/-4) docs/faq/contribute.rst (+1/-1) setup.py (+4/-9) tox.ini (+13/-0) |
To merge this branch: | bzr merge lp:~coreygoldberg/autopilot/tox-and-packaging |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Martin Pitt (community) | Approve | ||
Thomi Richards (community) | Approve | ||
PS Jenkins bot | continuous-integration | Approve | |
Review via email: mp+186034@code.launchpad.net |
Commit message
add tox config and packaging fixes
Description of the change
this branch adds tox support and fixes some packaging issues.
The result is that you can now build/sdist/test against supported Python versions (2.7/3.3) by running 'tox' in the root dir of the source.
changes:
=== modified file '.bzrignore'
added more files to ignore list
=== modified file 'MANIFEST.in'
added sources to MANIFEST.in so they are included in package build.
lttng_module extension and debian changelog are both referenced in setup.py, and therefore need to exist.
=== modified file 'bin/autopilot'
script was using a non-Py3 compatible sorted().
changed from using cmp to a sort key.
=== modified file 'docs/faq/
FAQ was using wrong path for autopilot script when running from source branch.
=== added file 'tox.ini'
config for python-tox
Thanks for the setup.py and the sorted() fixes!
18 +recursive-include debian changelog
Upstream releases generally shouldn't ship packaging information, so debian/ should not be included. But if you do it, why only debian/changelog? Most upstream projects these days don't ship a changelog as you can always get it from VCS, or they autogenerate it from "bzr log" or similar.
64 +commands = tests.unit
65 + autopilot run autopilot.
That runs only the unit tests; could we make this a bit more clever to run the full test suite if there's a $DISPLAY? When porting to py3 I found that a lot of issues only turn up with the integration tests.