Closed thbar closed 2 years ago
@thbar , could you get this working with travis, et? Any new OS needs to have CI testing to ensure everything is working properly.
@gclough I will have a look at this, since my migration budget is soon to arrive :-)
Upgrade budget still not coming. Situation is complicated.
If you merge ANXS:master into your fork, you will get CI testing as per #511 in Github Actions. We have an Ubuntu 20 CI test running now.
I have refreshed my fork, closed the PR and reopened it to force a CI run.
The installation seems to work locally with a refreshed fork, and I get the correct version of PostGIS installed.
The CI fails though (at least on Debian 10):
TASK [postgresql : PostgreSQL | Reload all conf files] *************************
668
changed: [postgresql-11]
669
changed: [postgresql-12]
670
fatal: [postgresql-10]: FAILED! => {"changed": false, "msg": "Unable to start service postgresql: Job for postgresql.service failed because the control process exited with error code.\nSee \"systemctl status postgresql.service\" and \"journalctl -xe\" for details.\n"}
671
changed: [postgresql-13]
672
changed: [postgresql-9_6]
I cannot tell yet if this is a transient CI error, or something linked to the PR. I'll close and reopen again.
It was a transient error, good!
Something is still missing to properly test this: since the suite wasn't failing before the PR, this means the modified code path is not under test.
Testing this would require to install some version of PostGIS during the tests.
@gclough I am using this at the moment (in Vagrant instances), so far so good.
@gclough thanks for incorporating this!
This mimics #435 but for Ubuntu 20 instead of Ubuntu 18.
To be considered a WIP.
I believe it would ultimately be better to first fix #464, then make it easy to override that from the user side, but at least this can help some!