Message162385
| Author |
r.david.murray |
| Recipients |
michael.foord, r.david.murray |
| Date |
2012年06月06日.01:15:33 |
| SpamBayes Score |
-1.0 |
| Marked as misclassified |
Yes |
| Message-id |
<1338945336.0.0.232822972471.issue15007@psf.upfronthosting.co.za> |
| In-reply-to |
| Content |
Right, I'm not wanting to run discovery from the command line, I'm wanting to run the tests in the package by package name. In my mind, this is exactly parallel to specifying a module name and having unittest automatically discover the TestCase classes in it. We don't have unittest run 0 tests because discovery wasn't invoked when the module name was specified. Why should it be different for a test package? If boilerplate is required in __init__.py to make that happen that's OK, though to my mind not ideal.
Is there some different magic I can put into __init__.py that will result in the tests in the package being run such that the package name shows up in the report? Without that, specifying a package name on the unittest command line seems pretty useless. (I mean, to get it to do anything useful, you'd have to be putting all the TestCases in the __init__.py, and if you are doing that, why have a package?)
The issue about improving the name output was about making it copy and pasteable (something I would also very much like). The naming issue here is different, about how to get the package name to show up in the fully qualified test name.
I will open another bug for the _top_level_dir issue. |
|
History
|
|---|
| Date |
User |
Action |
Args |
| 2012年06月06日 01:15:36 | r.david.murray | set | recipients:
+ r.david.murray, michael.foord |
| 2012年06月06日 01:15:36 | r.david.murray | set | messageid: <1338945336.0.0.232822972471.issue15007@psf.upfronthosting.co.za> |
| 2012年06月06日 01:15:35 | r.david.murray | link | issue15007 messages |
| 2012年06月06日 01:15:33 | r.david.murray | create |
|