Re: [foreman-dev] [RFC] Proposal: make foreman STI to work even when an inherited class is not found in Ruby.

Wow this is nice! I did not see this coming... LZ On Sun, May 7, 2017 at 3:58 PM, Shimon Shtein wrote: > > OK, it wasn't exactly my intention, but I see the discussion moves towards > general plugin install/uninstall procedure. > > I have discovered a nice feature that

Re: [foreman-dev] [RFC] Proposal: make foreman STI to work even when an inherited class is not found in Ruby.

OK, it wasn't exactly my intention, but I see the discussion moves towards general plugin install/uninstall procedure. I have discovered a nice feature that enables us to isolate plugin migrations, and run them on per-plugin manner: Apparently we can run: rake db:migrate SCOPE=my_plugin rake

Re: [foreman-dev] [RFC] Proposal: make foreman STI to work even when an inherited class is not found in Ruby.

I agree with Tomas, it's more cleaner to remove all the data right away. Therefore I suggest that we check for these kind of objects during initialization and if such an class is (not) found, then we can throw an error like "Run foreman-rake plugin:clean" to delete orhpaned records. I am for (A)

Re: [foreman-dev] [RFC] Proposal: make foreman STI to work even when an inherited class is not found in Ruby.

Thanks for bringing this up. I think the right approach is to have plugin uninstallation or rather clean up rake tasks. The clean up should IMHO delete all data that would case problems after code removal. It shouldn't try to revert the schema changes, but destroy all data that can cause

Re: [foreman-dev] [RFC] Proposal: make foreman STI to work even when an inherited class is not found in Ruby.

This issue is tightly coupled with plugin uninstallation and I think we should solve the two problems together. At the moment there's no standard plugin uninstallation procedure that would take care of cleaning up the system. This is something each plugin should provide. One thing (imho the less