You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Since #337 it's possible to update the runtime version and metadata. However, that currently needs to be done manually by the user. The way I'd imagine it, is that one listens to runtime upgrade events and then calls the update_runtime function.
This could be offered as a subscription function by the client itself, such that not every user needs to self-implement this.
Open question: Since we'd introduce mulithreading here, is a lock required? If so, is that possible in a no_std environment?
The text was updated successfully, but these errors were encountered:
Hmm, I think finding a nice solution to the locking issue is the main challenge. Some thoughts:
We could build a separate RuntimeUpdateSubscriber which would take a Mutex<Api<..>>. Creating the Mutex is the responsibility of the user. Not very nice from a user perspective.
The thread could just set an atomic bool (part of core) something like needs_runtime_upgrade. Then all the other functions would have to check the flag before doing anything. Not very nice, but I think this should work without Mutex
async support could probably provide a nicer solution
so, let's first do the async support and then we look at this issue again? -> I adjusted the order in our backlog.
This feature is something, that has been asked for (see the comment by Paradox at our Kusama proposal), so it would be good to implement it within this proposal.
Since #337 it's possible to update the runtime version and metadata. However, that currently needs to be done manually by the user. The way I'd imagine it, is that one listens to runtime upgrade events and then calls the
update_runtime
function.This could be offered as a subscription function by the client itself, such that not every user needs to self-implement this.
Open question: Since we'd introduce mulithreading here, is a lock required? If so, is that possible in a no_std environment?
The text was updated successfully, but these errors were encountered: