Fix membership test fail for Dec 31

[demeritcowboy]

Overview

Tests are failing on Dec 31

Before

    CRM_Member_Form_MembershipTest.testSubmit with data set #0
    CRM_Member_Form_MembershipTest.testSubmit with data set #1

Failed asserting that two strings are equal.
--- Expected
+++ Actual
@@ @@
-'2022-12-31'
+'2023-12-31'

After

Technical Details

I'm not sure what the real-life expectation is. Here it's for 2 terms, but let's say it's 1 term. If I join a fixed membership on Dec 31 that ends on Dec 31, does it end immediately, or should I expect a full year? What about Dec 30 - do I only get 1 day?

Comments

[civibot]

(Standard links)

• If this is your first pull-request for CiviCRM, please browse CONTRIBUTING.md for information about the development and testing processes.
• If you are reviewing this pull-request, you may wish to consult the test sites and the Review Standards (long template, short template).

[totten]

I did an equivalent patch locally - so I basically agree with this. :)

I'm not sure what the real-life expectation is. Here it's for 2 terms, but let's say it's 1 term. If I join a fixed membership on Dec 31 that ends on Dec 31, does it end immediately, or should I expect a full year? What about Dec 30 - do I only get 1 day?

Right, any fixed-period policy (if strictly enforced) will have some yucky days (like 12/31) where paying gives the subscriber a terrible deal (eg 1 day of membership).

If you look at the membership type used for this unit-test (AnnualFixed), it declares:

      'fixed_period_start_day' => '101',
      'fixed_period_rollover_day' => '1231',

The key thing here appears to be the fixed_period_rollover_day.

Based on some r-run with different current-dates and different rollover-days, fixed_period_rollover_day defines a sort of hidden grace-period. So if you sign-up on 12/31 for 2 years of membership, you will get 1 day + 2 years (~729 days) of membership rather than 1 day + 1 year (~365 days).

[demeritcowboy]

Thanks @totten