Take into account the data queued in the sender

[lpinca]

This makes the bufferedAmount getter take into account the data queued in the sender.

Fixes #970.

cc: @alexions

[alexions]

I'd move this block of code to a separate function, like

makeBuffer (data) {
  var readOnly = true;

  if (data && !Buffer.isBuffer(data)) {
    if (data instanceof ArrayBuffer) {
      data = Buffer.from(data);
    } else if (ArrayBuffer.isView(data)) {
      data = viewToBuffer(data);
    } else {
      data = Buffer.from(data);
      readOnly = false;
    }
  }

  return [data, readOnly];
}

Usage:
[data, readOnly] = makeBuffer(data);

[lpinca]

I also don't like having this code duplication. The reason for not using a separate function is to avoid creating an array.

[alexions]

Agree. Array can make additional overhead... ok :) Speed is better than duplicates :)

[lpinca]

It should be benchmarked to see if it actually makes any difference :)

[lpinca]

Another option would be to add the readOnly flag to the returned buffer. For example:

const kReadOnly = Symbol('read-only');

function toBuffer(data) {
  var readOnly = true;

  if (!Buffer.isBuffer(data)) {
    if (data instanceof ArrayBuffer) {
      data = Buffer.from(data);
    } else if (ArrayBuffer.isView(data)) {
      data = viewToBuffer(data);
    } else {
      data = Buffer.from(data);
      readOnly = false;
    }
  }

  data[kReadOnly] = readOnly;
  return data;
}

but I'm still not happy with it as that changes the buffer hidden class.

[lpinca]

I would like to have at least one LGTM from the other @websockets/admin members before merging.

[alexions]

@lpinca How do you like this variant?
#975

[lpinca]

@alexions thanks but I'm -1 on that approach as that proposes to check the things twice (2x Buffer.isBuffer(), 2x ArrayBuffer.isView(), etc.), first to see if data is readOnly then to convert it to a buffer. I would rather keep the duplicated code.

[alexions]

ok. Agree that your changes will work faster.

[lpinca]

Before this change the whole if...elsechain was used in 2 functions, now in 3 so I think it's not so bad. I'm just waiting for the approval of another @websockets/admin member :)

[lpinca]

I'll merge this tomorrow if there are no objections.

[lpinca]

@alexions thanks!

[alexions]

@lpinca you're welcome!