[RFC] possible rstream simplification

[aktau]

@tarruda some time ago I asked a question about the possibility of alloc_cb(); alloc_cb(); read_cb(); read_cb() occurring on Windows: https://groups.google.com/forum/#!topic/libuv/OD38PeGeVgQ

This answer is enlightening:

E.g. this could happen: 

> > alloc_cb(handle1); 
> > alloc_cb(handle2); 
> > read_cb(handle1); 
> > read_cb(handle2); 

But this isn't: 

> > alloc_cb(handle1); 
> > alloc_cb(handle1); 
> > read_cb(handle1); 
> > read_cb(handle1); 

While reading rstreamc., I see:

// Called by libuv to allocate memory for reading.
static void alloc_cb(uv_handle_t *handle, size_t suggested, uv_buf_t *buf)
{
  RStream *rstream = handle_get_rstream(handle);

  if (rstream->reading) {
    buf->len = 0;
    return;
  }

  buf->len = rstream->buffer_size - rstream->wpos;
  buf->base = rstream->buffer + rstream->wpos;

  // Avoid `alloc_cb`, `alloc_cb` sequences on windows
  rstream->reading = true;
}

It seems like the rstream->reading thing is not necessary, even on Windows. The reason being that each stream has a 1-to-1 correspondance with a libuv handle, and alloc_cb(handle1); alloc_cb(handle1); can never happen.

There would be a problem if we were reusing the same rstream->buffer for all streams though. As far as I can see, we don't though.

[tarruda]

I added this guard is because @saghul mentioned(or I misunderstood) that sequential calls to alloc_cb might happen for the same stream on windows.

[aktau]

I added this guard is because @saghul mentioned(or I misunderstood) that sequential calls to alloc_cb might happen for the same stream on windows.

Do you remember where he said it? I'd like to know for sure too.

[saghul]

I added this guard is because @saghul mentioned(or I misunderstood) that sequential calls to alloc_cb might happen for the same stream on windows.

Not for the same stream, just for different streams.

[aktau]

Not for the same stream, just for different streams.

Perfect, then we can remove a couple of lines. I'll do it in my PR close to the commit that increases the buffer size. (and thanks @saghul!)