Conversation
- Fix shed_diff so its exit code actually is non-zero if there are differences. - By default - don't report differences for tool shed populates fields (toolshed, changeset_revision). This can be disabled with --raw. - Tests for ensure --raw and smart differencing are working and exit codes are fixed. - Update shed_app test app to inject the tool shed populated fields into respective arguments.
+1 |
can you just |
There was a problem hiding this comment.
It is used inside of planemo.shed.
|
+1 |
That would really increase the runtime of those tests - it takes well less than a second to start the server and run those tests. I also like mocking out the tool shed interface because it really crystalizes exactly the API planemo is consuming. There is one functional test that actually exercises a real tool shed (https://github.com/galaxyproject/planemo/blob/master/tests/test_shed.py) and certainly it would be cool if that would automatically start a tool shed and stuff but I have not gotten around to it yet. It is a little bit harder than just |
|
That all sounds fair. Definitely harkens back to your point about "reason for travis and a reason for buildbot". If you'd like long running tests, the IUC jenkins bot is underutilised! :) |
|
Thanks for the offer - it isn't so much about CI though. I definitely want that tool shed test running on Travis - another minute is not my concern. I think my other comment was about Galaxy and that minute would drawfs in comparision to the hours it would take to run Galaxy's functional tests. What I love is |
|
Gotcha, sounds good. Didn't realise that shed tests ran that quickly. Made a offhand assumption that it'd be 15-30m to really test all the shed related functionality and corner cases. (Especially if testing against shed@master and shed@dev) |
shed_diffso its exit code actually is non-zero if there are differences.toolshed,changeset_revision). This can be disabled with--raw.--rawand smart differencing are working and exit codes are fixed.Implements #141.