Many years ago, we began scrapping the RTOS approach for some of our timing-critical applications. Our first applications were based on Solaris real-time extensions and served well us well for many years. Then came the era of shaky support for Intel Solaris and we moved many applications to Linux. There appeared to be no actual need for special real-time features with a good Linux distribution kernel and a GHz processors.
One of our most productive applications involves special test equipment computers that drive high speed serial ports. The original development of these systems had the "added benefit" of helping us to learn udev rules. The serial communication is very flexible and most of it is script driven. A current need to update computing platforms is driving script review.
There appears to be a bit of variety in outputs to the serial drivers. The echo of a string with redirection to the serial driver, cat of a file with redirection to the serial driver and dd of=/dev/.... are methods used at various places in the scripts. It is not at all clear what drove the choice of the output method at various points in the application. In addition, changing the method of output has not succeeded in breaking most of the applications.
Is there some fundamental difference between these methods of output to the driver that might trip us up as we simplify the scripts and port to new computer platforms?
Thanks in advance for any assistance.