This is going back around 8 years so my memories aren't perfect and the issues could have been fixed by now. IIRC my biggest gripes were:
1. There was no scroll-back so you could only see 1 terminal screen of output. If you scrolled back, the text would be wrong or truncated or something like that. We would work around this by launching a tmux session and using that for scroll-back, but tmux scroll-back over a remote connection is so much clunkier than scrolling back in your local terminal emulator. If I'm already launching a tmux session, then I don't need mosh's ability to recover from broken connections because I can re-attach to tmux.
2. I found the typing prediction to be more annoying than it is worth. It worked great in bash but interactive apps (like those based on ncurses), the app would often end up a distorted mess until the round-trip to the server got back the real/correct rendering from the server. If I'm waiting for a round-trip from the server then I'm not getting any benefit from mosh's predictive typing.
3. It was only a remote terminal replacement. If you're doing anything more over ssh (port forwards, x11 forwarding, etc) then you're going to run into all the normal ssh issues anyway.
4. You had to run a 3rd party script to kill old abandoned mosh sessions. This is a consequence of being able to re-attach after a broken connection: the mosh server has no idea if you're coming back or not, so it leaves your session there. At least with tmux I can list the active sessions, re-attach to them, see what they're doing, and destroy them at will. With mosh, you could only reconnect to an old session if you hadn't closed mosh on your laptop. If your laptop powered off (intentionally or not) then you could either play "guess which mosh pid to kill" or leave your mosh session running long enough for the mosh killer script to get it.
Also there was the issue of popularity/network effects. ssh is everywhere, mosh is not, and its hard to argue that mosh should be everywhere when I was getting only slight benefit from it. This isn't mosh's fault, any ssh competitor that isn't upstreamed into openssh is going to run into the same exact problem.
With my hypothetical ssh-over-QUIC I'm hoping for just plain-old ssh with all the normal features, just using QUIC instead of TCP. Ideally upstreaming into openssh. I know there are a bunch of projects out there attempting something like ssh-over-QUIC but I think it has to be upstreamed and a drop-in replacement to overcome the popularity moat regular ssh has built.
Great list that sparked my memory. The tmux scroll back issue was very frustrating for me, and ultimately what made me stop using it. What got me hooked on trying to use it - I do remember being amazed I could close my laptop, go home, and just reopen my laptop and immediately start typing away on my remote terminal without issue most of the time.
A few other issues I do remember running into (probably 10 years ago?):
1. Because it was udp, sometimes a captive portal, work, or hotel firewall blocked it.
2. It was also not installed by default anywhere, so sometimes sysadmins at work were unwilling to install it for me.
3. I sometimes ran into issues with color handling and utf8. I honestly don't remember if it was mosh or my terminal, though. It seemed to happen more often when I used mosh, but I could be misremembering.
1. There was no scroll-back so you could only see 1 terminal screen of output. If you scrolled back, the text would be wrong or truncated or something like that. We would work around this by launching a tmux session and using that for scroll-back, but tmux scroll-back over a remote connection is so much clunkier than scrolling back in your local terminal emulator. If I'm already launching a tmux session, then I don't need mosh's ability to recover from broken connections because I can re-attach to tmux.
2. I found the typing prediction to be more annoying than it is worth. It worked great in bash but interactive apps (like those based on ncurses), the app would often end up a distorted mess until the round-trip to the server got back the real/correct rendering from the server. If I'm waiting for a round-trip from the server then I'm not getting any benefit from mosh's predictive typing.
3. It was only a remote terminal replacement. If you're doing anything more over ssh (port forwards, x11 forwarding, etc) then you're going to run into all the normal ssh issues anyway.
4. You had to run a 3rd party script to kill old abandoned mosh sessions. This is a consequence of being able to re-attach after a broken connection: the mosh server has no idea if you're coming back or not, so it leaves your session there. At least with tmux I can list the active sessions, re-attach to them, see what they're doing, and destroy them at will. With mosh, you could only reconnect to an old session if you hadn't closed mosh on your laptop. If your laptop powered off (intentionally or not) then you could either play "guess which mosh pid to kill" or leave your mosh session running long enough for the mosh killer script to get it.
Also there was the issue of popularity/network effects. ssh is everywhere, mosh is not, and its hard to argue that mosh should be everywhere when I was getting only slight benefit from it. This isn't mosh's fault, any ssh competitor that isn't upstreamed into openssh is going to run into the same exact problem.
With my hypothetical ssh-over-QUIC I'm hoping for just plain-old ssh with all the normal features, just using QUIC instead of TCP. Ideally upstreaming into openssh. I know there are a bunch of projects out there attempting something like ssh-over-QUIC but I think it has to be upstreamed and a drop-in replacement to overcome the popularity moat regular ssh has built.