Return of paid lost items will not reopen a closed transaction

[Kathy Lussier]

Evergreen version: tested in 2.12, but probably a bug in earlier releases

For libraries that do not prohibit negative balances, if the library has configured its settings to void a lost bill upon a lost item's return and a patron fully pays the lost fee for an item, when that item is returned, the transaction does not re-open at the time the negative balance is created.

As a result, the transaction will not be listed in the "Patrons with Negative Balances" interface and also will not be listed on the patron's current bills screen in the web client. The transaction will be listed in the xul client's current bill screen.

Here is the sequence of events that occurs:

1. The transaction moves to lost, generating a bill for the patron.
2. The patron fully pays the lost fee, which sets the xact_finish date as expected.
3. The lost item is returned, creating a negative balance on the patron's record. At this point, the xact_finish date should be reset to null since the transaction is open again.

Also making a note that in investigating this bug, I discovered that the web client current bills is not displaying transactions with an xact_finish date, but the xul client didn't seem to care about the presence of an xact_finish date. I don't know if the web client behavior is a problem, but I'm just pointing it out in case these two interfaces should behave the same way.

[Bill Erickson (berick)]

Created and marked bug #1676906 as a duplicate. It contains manual steps for creating a slightly more convoluted example. Of note, the code tries to clear the xact_finish value during checkin, but a later update to action.circulation clobbers the cleared value by replacing it with the original bogus value.

[Bill Erickson (berick)]

Fix pushed:

http://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/berick/lp1670407-lost-checkin-xact-stays-closed

This smartens-up the partial xact_finish handling code that happens at the tail end of the checkin process so that it doesn't clobber the xact_finish value.

To confirm, repeat Kathy's steps above and check that xact_finish is NULL on the circulation once done.

[Dan Wells (dbw2)]

I have tested this and it works fine. It is pushed to master through 2.11. Thanks, Bill!

However, Bill and I also talked this over briefly at the conference, and agreed that this code now has a belt, suspenders, and a rope tied around its waist. Probably time for some new pants! As such, I am going to open and work on a follow-up bug to refactor this particular area of the code to be a little more sensible and less duplicative.

[Dan Wells (dbw2)]

P.S. Also pushed a couple new tests for these bits.