We have notices that updates in git libraries require not only restart the jupyter kernel but also first change to a different code enviroment. This is really impractical as you loose everything that is in RAM and for a small bug in your library code you need to rerun all the previous workings.
We were hope that this would have been fixed with the new update but... :((((
Is any way to surpass this? Do we do somehting wrong?
You do not need to change the code environment but to unload/reload the notebook (not just restart the kernel).
In 7.0 we added a button directly in the notebook UI to reload it much more easily (just click it and you're good).
You'll still need to rerun previous cells however. This step is required for security reasons and we don't plan to change it. This is because the library code is stored in folders that are protected against arbitrary writes (this would lose isolation and traceability). Hence, the library code must be copied in a user-accessible location each time it is modified. Changing this behavior would introduce code execution flaws.