Am I alone in thinking that the syntax for nullable types is bizarre?
"function example(): string? {"
vs
"function example() : ?string {"
"function example(): string? {"
vs
"function example() : ?string {"
Because the alternative conflicts in case we ever introduce generics. Consider <?php new Foo<Bar?> ?>. This could either be a generic instantiation, or it could be the same as <?php ((new Foo) < Bar) ?> ?> (which is how it is currently interpreted -- this is valid PHP code!)
Of course, you could specify how this is to be interpreted and implement the necessary parser hacks to make it happen, but it's best to avoid the issue altogether by using ? as a prefix.
I assume that this is also the reason why Hack does the same.
Question mark never appeared directly after colon before this, so the second syntax is not ambiguous. The first one would be understood as part of a ternary operator, followed by choking on the start of a block.
Reusing :: for namespaces would complicate this. Is foo::bar a class constant or a namespaced constant? Do we autoload src/foo.php or src/foo/bar.php? If I do `new foobar::qux`, is foobar::qux a class, a class constant containing a class name, or a namespaced constant containing a class name? Do we autoload src/foobar.php or src/foobar/qux.php?
It's not like the lexer or parser could magically solve this. The way the language is designed, there is inherent ambiguity here.
In some other languages, there is a unified symbol table, so you don't have this problem. But PHP doesn't have that.
Even a blind php dev...
PHP actually once supported this kind of ambiguity (PHP/FI 2.0 ended PHP code with a single > yet also allowed if ($a > $b)), but this wasn't a good idea and resulted from a greatly suboptimal implementation.