I made a version that worked okay with most network failures and browser failures. Not all, it’s impossible and I had zero motivation for that since I wanted to use WebRTC instead. I was going crazy.
I did quit and my old coworkers blame this video player, and they are right. But it was also because of the company strategy.
The items could be extremely customized and that became very difficult, including things like calculating the cart price.
It also felt like the most fragile system I wrote, but to my knowledge it's still running about six years later.
The majority of my web work, though, was not very technically challenging.
Earlier I would be reluctant to say a problem is "hard," because many of the best problems are not so much difficult as time-consuming.