Added support for Django 2.0 and py3.6 #26
Conversation
| runtests36-postgres auth_tests | ||
| runtests36-mysql auth_tests | ||
|
|
||
| runtests36-sqlite3 admin_widgets --selenium chrome --parallel 1 |
There was a problem hiding this comment.
There's no need to include admin_widgets here (that doesn't include all selenium tests). --selenium limits to the selenium tests.
There was a problem hiding this comment.
I don't want to execute every selenium test - just a few of them to validate that selenium testing works in general.
There was a problem hiding this comment.
Oops, didn't notice this was box-tests.sh.
| (vm) $ runtests27-sqlite3 admin_widgets --selenium chrome | ||
| (vm) $ runtests36-sqlite3 admin_widgets --selenium chrome --parallel 1 | ||
|
|
||
| The test suite will sometimes hang when running selenium tests in parallel mode. |
There was a problem hiding this comment.
I haven't seen this issue before but did reproduce it on the master branch just now. Did you see it on the 1.11 branch? I'm wondering if it's caused by a recent change in Django.
There was a problem hiding this comment.
No, I don't remember seeing this on the 1.11 branch. I'd planned to go back and test to see if we needed to open a ticket against Django.
Did you reproduce on master with this box, or another setup?
There was a problem hiding this comment.
Reproduced without the box. I'm suspecting django/django@bece837 might be the culprit but am running repeated tests on the commit before to see if reproduces there.
There was a problem hiding this comment.
If it helps, I found the tests hanging on the last test. I'm guessing it has something to do with the shutdown procedure.
| @@ -28,11 +21,17 @@ runtests35-sqlite3 auth_tests | |||
| runtests35-postgres auth_tests | |||
| runtests35-mysql auth_tests | |||
|
|
|||
| runtests27-sqlite3 admin_widgets --selenium chrome | |||
| runtests36-sqlite3-gis gis_tests | |||
There was a problem hiding this comment.
It may help to use the same terminology as Jenkins: spatialite, postgis, mysql_gis
There was a problem hiding this comment.
Good idea - being consistent in the naming will help if I'm promoting the similarities between the environments. It'll also force me to cleanup the current ansible statements that generate the aliases which aren't pretty at all.
No description provided.