-
Notifications
You must be signed in to change notification settings - Fork 96
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Improve header serialization and propagation #124
Improve header serialization and propagation #124
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks great, thank you, @jeffschoner-stripe!
One note — do you think it's feasible to support the legacy behavior while adding a deprecation message?
You can probably switch on a class name, however it might not be worth it as I don't think there are lots of people who are dependent on it. Let me ask in Slack
I've added another commit for this back-compat. Feel free to merge it, or we can remove it if there's no need to support that. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Perfect, thank you @jeffschoner-stripe !
* Pass config to the error handler instead of using the global config * Fix example tests * Make the config property in the workflow context publicly readable * [Fix] Retryer GRPC error lookup (coinbase#109) * Fix issue with GRPC error lookup in Retryer * Rename spec file for retryer to contain _spec * [Feature] Add id and domain to workflow context's metadata (coinbase#110) * Start syncing id and domain on workflow context metadata * Fixed tests Co-authored-by: DeRauk Gibble <[email protected]> * Explicit docker-compose project name (coinbase#114) * Add YARD documentation for Temporal::Client (coinbase#113) * Add YARD documentation for Temporal::Client * Add yard gem * Fix @option tag * Typo fix * Add signal arguments to start_workflow (support for signal_with_start) (coinbase#112) * Add signal arguments to start_workflow (to support signal_with_start) * Move signal arguments to the options hash * PR feedback * Fix merge error * Extend #wait_for to take multiple futures and a condition block (coinbase#111) * Differentiate TARGET_WILDCARD and WILDCARD, allow comparison with EventTarget objects (coinbase#118) * Turn off schedule_to_start activity timeout by default (coinbase#119) * Separate options from keyword args in #start_workflow (coinbase#117) * Separate options from keyword args in #start_workflow * fixup! Separate options from keyword args in #start_workflow * Surface additional workflow metadata on workflow context (coinbase#120) * Refactor metadata generation * Make task queue available on workflow metadata, add example test * Expose workflow start time metadata * Add memos (coinbase#121) * Add describe_namespace (coinbase#122) * Add describe_namespace * Feedback * Improve header serialization and propagation (coinbase#124) * [Fix] Non-started activity cancellation (coinbase#125) * Fix event target map entry for ACTIVITY_CANCELED event * Fix cancellation of a non-started activity * fixup! Fix event target map entry for ACTIVITY_CANCELED event Co-authored-by: DeRauk Gibble <[email protected]> Co-authored-by: DeRauk Gibble <[email protected]> Co-authored-by: Anthony Dmitriyev <[email protected]> Co-authored-by: nagl-stripe <[email protected]> Co-authored-by: jeffschoner-stripe <[email protected]> Co-authored-by: Drew Hoskins <[email protected]>
…kins_schedule Dynamic Workflows
Summary
This contains two improvements to headers, both of which change the behavior of workflow headers:
First, headers are now by default propagated to the next run of a workflow on continue-as-new. This can be overridden by passing
options: {headers: overridden_headers}
toworkflow.continue_as_new
if desired.Second, to set headers, you no longer need to use the generated proto classes. Instead, you can simply pass an ordinary hash, which will be automatically serialized into the proper proto. This makes the use of headers more ergonomic and eliminates the need to use the lower-level protobuf abstraction in code that uses temporal-ruby. However, this is a breaking change. If you are currently passing protobuf objects in for headers, you will need to instead pass a hash.
Testing
The continue-as-new integration test will now check that headers are propagated to the next run, as it already did with memos. The metadata workflow test now passes headers in the new way.
Existing unit tests have been updated to work with passing an ordinary hash for headers: