caja has, historically, been SPECTACULARLY crashtastic. Much <3 to everyone who's resolutely continued dragging it into a fit state over the years, as 19.10's seems to finally be able to manage basic tasks without just falling over on a daily basis. 
That said, the thing is still made of glass, and its "special" nature makes it both hard to work on and a major headache for users every time it DOES die. Between GNOME devs trying to use threads without any synchronisation, the memory leaks, 18.04 segfaulting when ejecting USB drives - it's got a pretty terrible track record, and every time anything goes wrong with it the only "recovery" option is Nuke It From Orbit.
Windows Explorer has the same problematic design: it's not "The FM", it's the whole desktop. But Windows did eventually at least get capable enough to run a new process for each FM window, massively reducing the impact of each problem. And I'm wondering if it's worth investing the time to do something similar with caja.
It's not at the top of my list, thanks to all the damage GTK3 has caused, but it's high enough up that I'm giving it some thought. It seemed likely that it would have been considered in the past, so I thought I'd ask if anyone had and what they turned up.
Obviously it wasn't worth it on balance, but if it was "just" a matter of "it would take more time than we can spare" rather than "it made my eyes bleed", I might have the chance to chip away at it a bit over the next couple of month, and any info that's already out there would be handy to have.
Thanks.