Support Durable Objects (in-memory only, for testing) #15
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jasnell
approved these changes
Sep 23, 2022
…n-promise. The next commit will do the same for `makeWorker()`, so that services are constructed entirely synchronously, which is needed to simplify cross-linking service bindings.
Previously, WorkerService's constructor took a list of `Service`s implemeting subrequest channels, which implied that those services had to be constructed before the WorkerService, and so cycles were impossible. We now split into two stages, construction and linking. We construct all services first, then we link them. This also means we no longer have any service types that need to be constructed asynchronously, which improves error handling since all errors will be reported by the time we've fininished constructing+linking.
…ice>. We can make it so all the service objects it returns are long-lived. This fixes a segfault on shutdown because destroying an `Own<T>` that uses `NullDisposer` still requires the object is live if it is polymorphic, in order to `dynamic_cast<void*>` it. We can't easily ensure shutdown order here.
kentonv
force-pushed
the
kenton/actors
branch
from
c71ea23 to
d879226
Compare
a-robinson
approved these changes
Sep 25, 2022
| @@ -378,6 +485,15 @@ void WorkerdApiIsolate::compileGlobals( | |||
| value = lock.wrap(context, kj::mv(importedKey)); | |||
| } | |||
|
|
|||
| KJ_CASE_ONEOF(ns, Global::EphemeralActorNamespace) { | |||
| value = lock.wrap(context, jsg::alloc<api::ColoLocalActorNamespace>(ns.actorChannel)); | |||
There was a problem hiding this comment.
Is it intentional that this type is in the repo? I'm fairly surprised to see this type here at all, but especially without a name change to something like EphemeralActorNamespace.
There was a problem hiding this comment.
Yeah I think it'd make sense to rename it. Only people actually looking at the C++ code would see this name, and I'm not too worried about them learning that these are historically colo-local, but a rename would make the code clearer for sure.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This implementation of Durable Objects is not actually... "durable". Storage is kept in-memory (in
ActorCache) and blown away when the process restarts. But, this is good enough to use for local testing of DO-based applications.Eventually, we'd like to have an actual durable storage back-end in
workerd. The implementation we use for Cloudflare Workers, though, is deeply coupled to our network details and so wouldn't be very useful to release as-is.