We're following what Unicode specifies here. Excerpt from SpecialCasing:

# Preserve canonical equivalence for I with dot. Turkic is handled below.

0130; 0069 0307; 0130; 0130; # LATIN CAPITAL LETTER I WITH DOT ABOVE

0069 0307 is an "i" followed by "combining dot above". The fact that the dot is rendered over the next character seems like a bug, as combining marks generally apply to the previous character.

Not sure what to do here, I can definitely see how the current behavior is unwanted if you're working with Turkish text and I'm not sure why it's specified to behave the way it does.

You can fix this creating a simple function of turkish chars if you need to use this in  7.3 version.


<?php
$var = "MOZAİK";
function TurkishFix($inputText) {
    $search  = array('ç', 'Ç', 'ğ', 'Ğ', 'ı', 'İ', 'ö', 'Ö', 'ş', 'Ş', 'ü', 'Ü');
    $replace = array('c', 'C', 'g', 'G', 'i', 'I', 'o', 'O', 's', 'S', 'u', 'U');
    $outputText=str_replace($search, $replace, $inputText);
    return $outputText;
}
$string = TurkishFix($var);
echo mb_strtolower($string);


Check here : https://3v4l.org/sYr8m

> […] and I'm not sure why it's specified to behave the way it
> does.

I assume that is to not loose information (e.g. for a later
mb_strtoupper(): <https://3v4l.org/gA2aH>).

> The fact that the dot is rendered over the next character seems
> like a bug, […]

Indeed, but not a bug in ext/mbstring, but rather in Firefox (and
maybe other browsers/renderers).  It's displayed as expected in
Chrome.