Add string type declarations to global API functions.

[mfb]

Overview

As of PHP 7 we can add string type declarations.

Before

These functions have only type hints for the first two params.

After

The string params are also enforced via type declarations.

[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]

Yup, makes sense to me. 👍

(Tangent: One might be reserved about filling-in signatures generally - e.g. one might say array when a better type woud be Traverseable. But this change makes a lot of sense -- seems manifest that civicrm_api() needs the $entity and $action to be either string or readily-convertible-to-string.)

[mfb]

The v3 API explicitly checks if params is an array, not just iterable. The v4 API might work with iterable though? I didn't try this

[mfb]

Looks like we can't use iterable as a type declaration until PHP 7.1 is the minimum version.

[eileenmcnaughton]

test this please

[totten]

For the array constraint - this PR doesn't change that, so I'm +1 on the PR anyway.

The weird r-test thing is.... the test timed out in both previous runs for this patch (30630, 30609).

ok 1840 - CRM_Utils_API_MatchOptionTest::testCreateMatch_one with data set #1 ('match-mandatory')
ok 1841 - CRM_Utils_API_MatchOptionTest::testCreateMatch_many with data set #0 ('match')
ok 1842 - CRM_Utils_API_MatchOptionTest::testCreateMatch_many with data set #1 ('match-mandatory')
not ok 1843 - Error: CRM_Utils_API_MatchOptionTest::testReplaceMatch_Email
not ok 1844 - Error: CRM_Utils_API_MatchOptionTest::testReplaceMatch_Address
Build timed out (after 60 minutes). Marking the build as aborted.
Terminated
Build was aborted

It feels like this might be some weird/quirky thing specific to this change. I can't see any other/similar timeouts among recent PR tests. The timeout happened in the same spot in both runs. (Background: The timeout policy is this: if the build-job doesn't produce any console output for 60m, then it aborts. So in both cases, it spent 60m with no output after testReplaceMatch_Address.)

Let's give it one more test-run to rule out some fluke in the execution...

[totten]

jenkins, test this please

[mfb]

Various tests assume civicrm_api() $entity param is nullable, so rather than rewriting the tests and/or internal API code, I decided to leave this assumption intact for now.

[eileenmcnaughton]

We should probably rip out those tests now. However, I'm OK to merge this now it's passing