Nice one, thank you! Did you check if hash parameters are copied correctly?

Nice one, thank you! Did you check if hash parameters are copied correctly?

The hash part don't sent to server, we don't need to handle it.

And I think an error should be thrown when the pathname or query or something is too long, the current behavior is just ignoring the over-limit part, in some edge cases, this may cause some confusion.

We could compute the necessary length and always dynamically allocate it?

We could compute the necessary length and always dynamically allocate it?

That will cause some security problems, attacker can send a large payload to comsume all memory on server.
IMHO we would better have a configuable limit of header size and body size.

Usually, 2048 bytes header should enough for most case, the problem is url being truncated without any notice.

Fair enough. That's what we have right now though, which I guess it's reasonable.

Let's fix it if / when we run into real world problems?

Fair enough. That's what we have right now though, which I guess it's reasonable.

Let's fix it if / when we run into real world problems?

That's ok.