[REF] Fix utf8mb4 test in APIv4 and re-enable the altering of databas…

[seamuslee001]

…e in the v3 version of the test and fix handling for first name field

Overview

This is a follow on to #20905 which improves the APIv3 version of the test to do the faking of the database to test the re-writing of the query. It also properly fixes the handling for the v3 unit test.

Before

APIv4 unit test fails

After

V4 and V3 unit tests pass in a similar manner

ping @demeritcowboy @eileenmcnaughton

[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).

[eileenmcnaughton]

@seamuslee001 so my conclusion is we can't do this alter - I ended up ripping it out because I hit so many weird situations in various parts of the suite. I think in the end it's better to just let this test actually test non-utf8 on those matrix machines that don't support utf8mb4 rather than try to mock it

[eileenmcnaughton]

Also worth noting - this used to not cause problems because \CRM_Core_BAO_SchemaHandler::databaseSupportsUTF8MB4() used to always incorrectly return FALSE

[seamuslee001]

So this is the exact same alter that is happening in the v4 version of this test

[eileenmcnaughton]

@seamuslee001 yeah - which I think we should remove to get rid of bizarre fails like #20984

[seamuslee001]

@eileenmcnaughton that is only failing because the Civi::statics cache hasn't been reset after the alter. This value https://github.com/civicrm/civicrm-core/blob/master/CRM/Core/BAO/SchemaHandler.php#L922 with the current test setup will still return TRUE even tho table has now changed to not support emojis hence my reseting of the civi statics in the apiv4 test which is all that is needed to fix it

[eileenmcnaughton]

The apiv4 suite should probably unset Civi::statics like the main suite does -

civicrm-core/tests/phpunit/CiviTest/CiviUnitTestCase.php

Line 384 in c3ea4bd

\Civi::$statics = [];

or we will keep getting weirdness

[eileenmcnaughton]

Yep - I remember this part of the process

api_v3_SyntaxConformanceTest::testCreateSingleValueAlter with data set #2 ('ActionSchedule')
api result array comparison failed checking if custom_6 was correctly updated
Array
(
    [update-params] => Array
        (
            [id] => 40
            [custom_6] => defaultValue
        )

    [update-result] => Array
        (

I can't recall the underlying issue for those fails - - but it is to do with the full text indexes in some way

[eileenmcnaughton]

@seamuslee001 if you can win this fight then go ahead & merge :-)

[eileenmcnaughton]

Yeah - this is about the point I gave up & decided that we should only test lack-of-support-for-utf8mb4 on databases that don't support utf8mb4 :-)

Stacktrace
api_v3_SyntaxConformanceTest::testCustomDataGet with data set #0 ('Acl')
incorrect count returned from Acl getcount
Failed asserting that 0 matches expected 1.

/home/jenkins/bknix-dfl/build/core-21001-6c1u3/web/sites/all/modules/civicrm/Civi/Test/Api3TestTrait.php:199
/home/jenkins/bknix-dfl/build/core-21001-6c1u3/web/sites/all/modules/civicrm/tests/phpunit/api/v3/SyntaxConformanceTest.php:917
/home/jenkins/bknix-dfl/build/core-21001-6c1u3/web/sites/all/modules/civicrm/tests/phpunit/CiviTest/CiviUnitTestCase.php:258
/home/jenkins/bknix-dfl/extern/phpunit8/phpunit8.phar:671

[eileenmcnaughton]

woot - sheer determination - take that jenkins

[demeritcowboy]

well done