I feel like most of the corporate world is gravitating toward web interfaces. Works great for me, since then a "thin client" can be as simple as a stripped-dow

I feel like most of the corporate world is gravitating toward web interfaces. Works great for me, since then a "thin client" can be as simple as a stripped-down Linux running a Qt5 or Electron app (or a browser in kiosk mode) that does nothing but access that web interface (or perhaps has a couple different tabs for different systems). For actual thin-client needs, it seems like Citrix XenDesktop is the go-to nowadays. Linux as a base OS works okay for this (there are Linux ICA clients). Linux also has RDP clients for Windows terminal servers (or hell, even Linux terminal servers with an RDP server; I typically do this for openSUSE servers in corporate environments where the people doing day-to-day maintenance are used to Windows and insist on a GUI). > How do I make sure you recover from crashes reliably? Just restart the app. > How do you secure the inputs, to avoid escapes from the application? With a stripped down environment the only such escapes are ones you configure yourself. X11 and the session thereof can be configured to have zero options besides starting up your chosen application. Even if you leave Ctrl-Alt-Backspace enabled, it'd just boot right back into that app. Ctrl-Alt-F(whatever) (if still enabled) should present a login prompt, so unless the password's compromised you're good to go there, too. Then it's just a matter of locking down the app itself, which again is straightforward if the session is configured to just restart the app if it closes/crashes. A custom application (say, a Qt5 app doing nothing except presenting a WebEngine widget and perhaps some support for hardware interfacing) offers fewer escape opportunities than, say, a generic browser in "kiosk mode".