Hello all!
I've been hard at work refactoring this extension and adding new features for a v2. Some of the often request changes included are: adding polls to posts, and having multiple polls per discussion/poll.
The operation that converts the poll association from discussions to posts is destructive in the case that fof/polls cannot find an associated 'first post'. I don't believe this is an issue, since in those cases the poll shouldn't be currently appearing in the forum anyways (aside from the badge).
However, just to be safe, I would prefer that others (with a lot more "old data") test these changes. Just to be safe that everything seems to work as it should migrating to this new version.
🛑 DISCLAIMER: THIS IS NOT BACKWARDS COMPATIBLE. DO NOT TEST IN PRODUCTION. BACKUP DATA 🛑
That out of the way, there's one PR to test: FriendsOfFlarum/polls73. This PR contains potentially destructive behavior that is unwanted - I want to verify that it only deletes what it should, and nothing else unintentionally. I also have a 2nd PR that builds off this first one, but that one doesn't require testing to avoid unintentional data loss.
If you know how to clone extensions & test their PRs already - great, feel free to do that. If not, you can test these PRs by running composer require fof/polls:dev-ds/move-polls-to-posts
.
With a DB backup ready to restore (!), running php flarum migrate
will perform the migrations that have a chance to delete polls. You'll know if it does if the logs contain a line with a count of how many polls were deleted - and their IDs. Knowing their IDs, you can restore the backup, go back to the latest v1 version of fof/polls
, and check whether the poll showed up in its discussion. If it didn't, then there's no issue. If there is an issue, please provide info on what the discussions whose polls were deleted when they shouldn't have been.
Note: the purpose of testing this is to verify that migrating from v1 to v2 works as intended. Testing this PR in a new, blank installation won't be nearly as helpful.