Don't break socket connections for postgres#540
Don't break socket connections for postgres#540Archmonger merged 2 commits intoArchmonger:masterfrom
Conversation
|
This is odd. The test failures seem unrelated to the changes. |
|
It does seem to somehow be related to this PR's changes. Just rebased #536 and got no CI errors. |
Also, use dict.get() for default values.
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #540 +/- ##
==========================================
+ Coverage 91.40% 91.44% +0.03%
==========================================
Files 19 19
Lines 873 912 +39
Branches 157 154 -3
==========================================
+ Hits 798 834 +36
- Misses 40 43 +3
Partials 35 35 ☔ View full report in Codecov by Sentry. |
|
I got confused by the sqlite errors, but those seem to be expected/ok. I couldn't track down the problem completely, but there seems to be some modification of the At some point during the tests, |
|
Could be this LOC, not sure though |
|
Will go ahead and merge this since it's a step up from current behavior. If someone in the future needs |
Also, use dict.get() for default values.
Description
Fixes #490.
#521 prevents users from setting host to an empty string for socket connections. It also contradicts the Django docs referenced in #520 which clearly state that empty HOST for Postgres means local socket connections.
Checklist
Please update this checklist as you complete each item:
By submitting this pull request I agree that all contributions comply with this project's open source license(s).