For Eclair, it was once just some configuration parameters + remote Core node that were causing a series of force closings. Like 5 simultaneously.
So yes, I fully agree. But maybe it is just a protocol flaw so developers of major clients couldnt' converge to reliable solutions.