This makes it impossible to distinguish between a missing header and a header that is present whose value is the empty string. It would be better if we were precise about this as Fetch and HTTP are too.
<rdar://problem/106944735>
String has a null state in WebKit. This should be distinguishable already. `string.isNull()` -> header is missing. `string.isEmpty() && !string.isNull()` -> header is present but has no value.
I suggest we mark this bug as invalid unless I missed something.
You could also use `httpHeaderFields().contains()`.
The bug is more that, apparently, some code paths are using isEmpty() while they should probably use isNull().
Ah, I didn't realize that! Let's close this as INVALID then. I'll follow-up with more detailed bugs as I come across things.