How many time do we have to repeat that to learn?
It’s possible to write scripts using some GNU Xnee bash functions to fake user input. To speed up things (don’t pre optimise!!!) I’ve tested launching one one instance of cnee and use a fifo or tail -f to write to that instance. The solution has been buggy to say the least.
A while ago, I got yabr (yet another bug report). This time I decided to be a bit bolder and perhaps remove old code. So right now when diving down into the report I remove all the fancy fifo and file/tail -f stuff and start a new cnee process for every xnee function the user calls. Guess what, it works better. Smoothly. …. and not consuming that much CPU.
So again, the kiss principe proves to be a good one.