Intermittent failure to set tuner output level (*OF2) due to Serial race condition

[cjritola]

In hardware_usbserial.py, the delay given by Python/Linux between opening a port and when it is available for writing is undefined. The current convention is to use delays, such as in the original source, as well as for other projects:
https://stackoverflow.com/questions/7266558/pyserial-buffer-wont-flush

... however, this still didn't work on my system, the *OF2 setting was not parsed by the RS-HFIQ except for rare situations by chance. Setting the delay longer improved the probability but doesn't really fix the underlying problem.

There appears to be a design flaw in Python or the way Linux interacts with serial where a handshake interaction may help things, such as getting the firmware version *W to confirm that the device is communicating before issuing state commands like *OF2, which don't have a confirmation back.

The odd thing is, strace shows the messages actually being sent out to serial at the OS level:

38110 ioctl(10, TCFLSH, TCIFLUSH)       = 0
38110 ioctl(10, TCFLSH, TCIFLUSH)       = 0
38110 ioctl(10, TCFLSH, TCOFLUSH)       = 0
38110 write(10, "*OF2\r", 5)            = 5
38110 select(14, [13], [10], [], NULL)  = 1 (out [10])
38110 ioctl(10, TCFLSH, TCOFLUSH)       = 0
38110 ioctl(10, TCFLSH, TCIFLUSH)       = 0
38110 ioctl(10, TCFLSH, TCOFLUSH)       = 0
38110 write(10, "*W\r", 3)              = 3

... perhaps write() doesn't truly wait for the message to be sent. Consequently when flushOutput is invoked, the message is removed from the queue?

Will be submitting a pull request for an improvement soon.