The dependencies for a functional RTC are spread in various repositories (at least 6). Each of those repositories test their API and behavior in a classic unit test approach.
With the Jupyter RTC Test repository, we strive to ensure that the full stack is covered. Is also ships tools to measure the quality and perfomance of the RTC system, think to You can not scale what you can not measure...
Historically, there have been a few RTC failures reported, with quite low user traffic.
- [OPEN] Lab 4 editor corrupting simple CSV files when collaborative mode is on jupyterlab/jupyterlab#14715
- [OPEN] Incorrect file name shown on tab for markdown file jupyterlab/jupyterlab#14763
- [OPEN] CRITICAL - data loss and notebook corruption on a hosted Lab instance jupyterlab/jupyterlab#12154
- [OPEN] Non-empty notebooks appear as blank jupyterlab/jupyterlab#13930
- [OPEN] Damaged notebook (duplicate cells) in RTC-active hosted hub jupyterlab/jupyterlab#14031
- [OPEN] Development workflow with RTC jupyterlab/jupyterlab#14170
- [OPEN] Collaboration mode rolls back progress jupyterlab/jupyterlab#14224
- [OPEN] Collaboration rolls back progress jupyterlab/jupyterlab#14278
- [OPEN] Collaboration breaks notebooks jupyterlab/jupyterlab#14343
- [OPEN] Trusting notebook in RTC mode leads to File Changed dialog jupyterlab/jupyterlab#14347
- [OPEN] Inconsistent readonly status detection on RTC and after file status changes jupyterlab/jupyterlab#15040
We need now to make sure the system is robust enough at scale, and also identify their limits.