The reason a setter cannot be async
is that an async
function returns a future, and a setter does not return anything. That makes it highly dangerous to make a setter async
because any error in the setter will become an uncaught asynchronous error (which may crash your program).
Also, being async
probably means that the operation will take some time, but there is no way for the caller to wait for the operation to complete. That introduces a risk of race conditions.
So, it's for your own protections.
If you need to do something asynchronous inside the setter anyway, perhaps log something after doing the actual setting, you have a few options.
The simplest is to just call an async
helper function:
set foo(Foo foo) {
_foo = foo;
_logSettingFoo(foo);
}
static void _logSettingFoo(Foo foo) async {
try {
var logger = await _getLogger();
await logger.log("set foo", foo);
logger.release(); // or whatever.
} catch (e) {
// report e somehow.
}
}
This makes it very clear that you are calling an async function where nobody's waiting for it to complete.
If you don't want to have a separate helper function, you can inline it:
set foo(Foo foo) {
_foo = foo;
void _logSettingFoo() async {
...
}
_logSettingFoo();
}
or even
set foo(Foo foo) {
_foo = foo;
() async {
...foo...
}();
}
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…