1bb6e4509a
This is another attempt to fix selenium spec reliability. Without this change, when doing a `get` on a URL, we weren't waiting for the new page to actually load before doing our checks. This leads to stale element errors, etc. We do have to be sure to *not* wait for a new page load if the target URL is the same as the current URL, but with a different hash. Running the following command: ``` for i in $(seq 10); do echo "**** $i" bundle exec rspec -e "moving outcomes tree should move a learning outcome group via tree modal" spec/selenium/ done ``` would fail at least half the time without this patchset, and never fails with it (in my testing). Change-Id: I2d707f96d7713f69be865355d8a78bad1261d446 Reviewed-on: https://gerrit.instructure.com/53211 Tested-by: Jenkins Reviewed-by: Zach Wily <zach@instructure.com> Product-Review: Zach Wily <zach@instructure.com> QA-Review: Zach Wily <zach@instructure.com> |
||
|---|---|---|
| .. | ||
| config | ||
| kinesalite | ||
| seleniumff | ||
| Dockerfile | ||