User tests: Successful: Unsuccessful:
Pull Request for Issue #26387 .
Don't perform update queries where they are not needed.
Make sure that the queries performed are correct
Perform a clean install
Check the db table and you will see that modules have no start publishing date
Export the database
apply the pr
Perform a clean install
Check the db table and you will see that modules have a start publishing date
Export the database
Compare the two database exports
The only differences between the two exports will be sessions, admin password and the timestamps.
There should be no difference for the userid
There should be a difference with the modules publish_up date which should be the time of the install and not 0000
Less queries performed that are not needed reduces the installation time etc
Status | New | ⇒ | Pending |
Category | ⇒ | Installation |
obviously - but as it wasn't that obvious I updated the instructions
Code review looks good to me, can do real test tonight after work (if still necessary then).
Just as a fair warning, this change probably breaks “fixing” timestamps and user IDs in the localise.sql
or custom.sql
file if you use those to seed tables with data like the sample data SQL files did. The timestamps are less of an issue, but the user IDs potentially leaves you in a spot where things look broken in the backend.
also the script is still wrong in its existing format for the use cases you described
Status | Pending | ⇒ | Closed |
Closed_Date | 0000-00-00 00:00:00 | ⇒ | 2019-09-24 12:53:23 |
Closed_By | ⇒ | brianteeman | |
Labels |
Added:
?
|
Just got pointed at this by Richard. I'm happy to remove the date code like done here in favour of people using GET_DATE
in their sql. but agree with michael the created by/modified by user needs to remain
@brianteeman Is there maybe a step "Apply the changes of this PR" (or similar) missing in the testing instructions between the 1st export and the 2nd clean install?