Hello, all.

I've been putting together a little protocol that uses QDataStreams and QIODevices. I am expecting to use it on QTcpSockets but I want to keep things at the QIODevice level.

At the recieve end I was doing the following:
Qt Code: 
  1. if (false == d_stream.device()->waitForReadyRead(-1)) {
  2.     // Report error.
  3. }
  4. while (d_stream.device()->bytesAvailable() < sizeof(Header)) {
  5.     // Spin. This shouldn't take long, the header is small, so
  6.     // once any data is available, it probably all will be.
  7. }
To copy to clipboard, switch view to plain text mode 

This would block on waitForReadyRead if all the data had already arrived, which was surprising.

I have changed this now to
Qt Code: 
  1. /* (1) */while (d_stream.device()->bytesAvailable() < sizeof(Header)) {
  2.     /* (2) */ if (false == d_stream.device()->waitForReadyRead(-1)) {
  3.         // Report error
  4.     }
  5. }
To copy to clipboard, switch view to plain text mode 

This seems to work, and at a glance it makes a certain sense, but isn't there now a race condition? If all the data arrives between (1) and (2) then (2) will block as it did in my first implementation.

This code is supposed to be in a high-reliability, high-frequency application. "Unlikely" isn't good enough.

Am I using the API wrong or is this a hole in Qt?

Thanks for reading.