Stop using == '' with regard to PHP 8

[leofeyer]

See https://3v4l.org/XfAXs

[m-vo]

For reference, this is the RFC: https://wiki.php.net/rfc/string_to_number_comparison

If I get this correctly many of the comparisons in this PR are actually comparing strings where == imo is still correct (even for '0'). I don't think we should transform them into another representation with type juggling involved but rather make them strict or leave them unchanged.

What about case and in_array (without 3rd parameter) statements that also use non-strict comparisons?

[ausi]

many of the comparisons in this PR are actually comparing strings

Most of these checks are boolen checks like empty(). I think they are more correct in most cases because an empty array() and the string '0' is mostly considered to be false in our codebase I think.

[m-vo]

Well I'm thinking about all the "boolean" values or numbers with an empty state from the db that we get via dbal/models. Things like $objComments->addReply (→ '1', '') or $objParent->start (→ '', '0', '123', ...) that are actually strings and are only interpreted as boolean/integer. They either won't contain 0 (int) or 0 would mean something different.

I'd rather see them written like this:

$objComments->addReply === '1'
$objComments->addReply === ''

$objParent->start !== '' && $objParent->start > $time

[ausi]

What if someone sets $objComments->addReply = true; before the comparison? The model can come from the registry so this is possible in many cases I think.

[m-vo]

Ah, you got a point there.

[m-vo]

What about in_array? https://3v4l.org/TqmZG

Do we have any relevant cases of this as well in the codebase?

[leofeyer]

We have one in_array(0, case, which however does not cause problems. To be sure, we have still adjusted it in 984fc19.