Suggestion for another bind()

[g40]

Persuant to #923. I added the following code to my fork of webview to check the C++ side of things is all OK.

  // another overload of bind() with user data
  // using statement means we can overload cleanly
  using sync_binding_arg_t = std::function<std::string(std::string, void*)>;
  // invoked with user data
  void bind(const std::string &name, sync_binding_arg_t fn, void* arg) {
    auto wrapper = [this, fn](const std::string &seq, const std::string &req,
                              void * arg) { resolve(seq, 0, fn(req,arg)); };
    bind(name, wrapper, arg);
  }

Example use therefor:

// invoked with instance data, so JS -> C++ -> JS
std::string callout2(const std::string& arg, void* data)
{
	webview::webview* pwv = reinterpret_cast<webview::webview*>(data);
	std::string ret = std::string("{ \"status\": true, \"data\": \"JSONpacket\" }");
	std::printf("%s (%p) => %s\n", arg.c_str(), pwv, ret.c_str());
	// JSON status message
	return ret;
}

// example usage
webview::webview w(true, nullptr);
w.bind("callout2", callout2, &w);

This works, with the promise completing as expected:

callit:  callout2 ƒ () {
        var seq = RPC.nextSeq++;
        var promise = new Promise(function(resolve, reject) {
          RPC[seq] = {
            resolve: resolve,
            reject: reject,
          };
     …
index.html:61 ret =>  Promise {<pending>}
index.html:36 success:  {status: true, data: 'JSONpacket'}

If interested I can submit a PR.

Thanks for listening.