Testing with my PHP 8.1, I can't replicate this. But I do have a bunch of PHP 8.1 fixes in place which are still awaiting approval.

Apart from the pre-existing 'Statues' typo, looks good to me...

The issue is in a change which went in with PHP 8.1.22, and is being reverted in PHP 8.1.23.
@aklapper - as PHP 8.2.9 isn't released yet, I assume you compiled from source? If you 'git pull' and recompile, the latest 8.2.9-dev should work. The fix was committed Aug 2 21:08:52.

Raised PHP bug https://github.com/php/php-src/issues/11871

In T15586#12570, @Ekubischta wrote:

We do need the null check - but only for the unit test when the diff is not associated with a repository.
For live use, getSourceControlBaseRevision() will always return a string.

• Just checked the getSourceControlBaseRevision() functions

Yeah, we can use isset() and strlen() in place of phutil_nonempty_string(), and I'm happy to do it.

In D25364#10593, @avivey wrote:

$edit_mode has a true value for 'create' or 'edit', and false (null) for read. The strlen() is superfluous.

maybe add that as a comment in there...

Request review for the second part of the change!

$edit_mode has a true value for 'create' or 'edit', and false (null) for read. The strlen() is superfluous.

Explicitly check for null instead of using phutil_nonempty_string()

Tested valerio's suggestion -
$is_start = $message !== null;

Do you need to be able to replicate it? The fault does seem rather obvious in the code -

Andre already has a diff already out for this...

Changing to the (bool)strlen() to phutil_nonempty_string() is the simplest fix...

D25296 already out for this

Have replicated the fault.

Fixable using coalesce

I agree that Phorge should not fall over because of a deprecation warning.