Nice news, metvas.
Here is my report. I made some modifications to solve the counters issue. It works, but trellis have a bug on the counters (the demo version have it too, so i didnt break it
so, even when you update the list of members from one or more groups, the counter is 0. This is a minor bug anyway and surely will be fixed on next release.
Because lack of time, I didnt go with the cache issue. Is this important for us? The cache is nice, you can undo some changes if you want. I can imagine just one scenario: the server clock have a problem, so, mysql will remove users who are not really expired. This can happen if the clock of the server is foward. The worst thing can happen is an authorized member who cant access to his account. Remember this is a change hard to trace.
If something like this happen, we want to undo the changes, fix the clock, and re-run the script to remove really-expired accounts. I am thinking in adding a report with date included, to be sended by mail to the admin. This will notice us of the problem quickly (hopefuly), but oviously not resolve the undo issue.
A possible solution: generate an intermediate group, with just the last accounts moved. In that way you can move back the unfairly moved users with some efficience. The members of this group needs to be moved in the next script execution to a definitive group, and the transitional group updated with the last expired accounts.
what do you think about this?