Fix upgrade failures from zero value trxn_date

[agh1]

Overview

Some longstanding CiviCRM installations would have upgrade failures going to 4.7.19 or higher with the database error,

Incorrect datetime value: '0000-00-00 00:00:00' for column 'trxn_date'

This change fixes those values to be NULL prior to the query that causes the problem.

Before

Certain sites with legacy data (canceled event registrations, potentially among other things, from around 2013) have trxn_date values of 0000-00-00 00:00:00 in the civicrm_financial_trxn table. This is not an allowable value for a datetime field, and an upgrade step on the way to 4.7.19 will cause the error,

ALTER TABLE civicrm_financial_trxn CHANGE pan_truncation pan_truncation VARCHAR( 4 ) NULL DEFAULT NULL COMMENT 'Last 4 digits of credit card' [nativecode=1292 ** Incorrect datetime value: '0000-00-00 00:00:00' for column 'trxn_date' at row XXX]

After

Before that query executes, all 0000-00-00 00:00:00 values in civicrm_financial_trxn.trxn_date are set to NULL. Meanwhile, those values seem to be always also occur in the receive_date field of the corresponding civicrm_contribution row. The upgrade now sets those values to NULL as well.

Technical Details

This shouldn't affect most users.

Comments

@KarinG @JoeMurray @eileenmcnaughton I suspect this is the legacy of some bug fixed in 4.3 or 4.4, but I couldn't easily find it.

Also, while I believe NULL is an improvement over 0000-00-00 00:00:00 since it's at least an acceptable value, those rows are now the only ones with null dates. Theoretically, this could cause problems in pages or functions that make use of the dates.

[eileenmcnaughton]

test this please

[seamuslee001]

test this please

[tschuettler]

@agh1 That may aswell been just some different server configuration, where those values had been permitted and used instead of NULL or the current time:
https://dev.mysql.com/doc/refman/5.7/en/sql-mode.html#sql-mode-strict

https://dev.mysql.com/doc/refman/5.7/en/server-system-variables.html#sysvar_explicit_defaults_for_timestamp

[KarinG]

@agh1 - agreed - no data (NULL) is better than wrong data;

[eileenmcnaughton]

I feel like this is a bit of an obscure scenario but those columns do ALLOW NULL and I feel this is obscure enough to defer to the person who has done the research. Merging