This project adheres to Semantic Versioning.
Nothing yet!
- The
advancedCallMsoption had no effect unless passed to TimeoutExtender directly, which no sane person would do. - The
noExtensionsAfterSecsoption defaulted to zero, which led to messages never being auto-extended unless a custom value was passed.
- The
advancedCallMsoption to tweak how far in advance of a message's VisibilityTimeout that a ChangeVisibilityTimeout call will be made
- A possible scenario where someone could set the visibilityTimeoutSecs ultra low and cause timers to be set for negative milliseconds
- The option
noExtensionsAfterSecsnow defaults to seconds instead of milliseconds
- The
autoExtendFailevent, fired when the TimeoutExtender fails to change the VisibilityTimeout of a message due to the message no longer existing or being expired. This event may encompass more errors in the future.
- TimeoutExtender no longer continually tries to extend messages that SQS has reported as previously deleted or expired. Instead, messages are removed from the list on this error, and a
autoExtendFailevent is emitted. - TimeoutExtender no longer tries to extend the timeout beyond AWS's maximum message lifetime.
- Message now passes an instance of itself to the handled call, fixing cases where the message would not be emitted with the "handled" event.
- TimeoutExtender now passes seconds to
changeMessageVisibilityinstead of milliseconds. AWS didn't like that.
- The
deletedevent, fired when a message has been confirmed to be successfully deleted - The
delQueuedevent, fired when a message deletion has been queued up for the next batch - The
handledevent, fired when a message has been handled by any means (kept, deleted, or released) - The
releasedevent, fired when a message has been confirmed to be successfully released back for immediate availability - The TimeoutExtender feature, which can automatically extend the VisibilityTimeout of any message that has not yet been handled. See README for details!
- The
autoExtendTimeoutandnoExtensionsAfterSecsoptions to support the TimeoutExtender feature - The
getQueueVisibilityTimeoutfunction. Intuitively, this gets the queue's VisibilityTimeout. opts.SQSnow supports an instance of the SQS client class in addition to a constructor (cwhenderson20)
- BREAKING: The
deleteMessagemethod now requires a Message object to be passed to it; not just the receiptHandle - BREAKING: The
releaseMessagemethod now requires a Message object to be passed to it; not just the receiptHandle - BREAKING: The
handledMessagemethod now requires a Message object to be passed to it - The
start()function now returns a Promise that resolves when the poller has started
- You can now pass an SQS constructor from your own aws-sdk version instead of using the one packaged with Squiss (cwhenderson20)
- The visibilityTimeout of a received message can now be changed with
squiss.changeMessageVisibilityandmessage.changeVisibility(tothandras) - A message can now be returned to the queue and made immediately available with
message.release(tothandras) - Messages can also be returned to the queue with immediate availablility through
squiss.releaseMessage
- Deleting messages by handle now works. The Message ID used will be the position of the message in the delete queue, 1-indexed. (#13)
- Tests now use dirty-chai to work around Chai's cardinal sin of allowing nonexistent assertions to be called without error
- Tests now use chai-as-promised to assert promise results
- The
gotMessagesevent, which fires when Squiss retrieves a new batch of messages from SQS - The
maxInFlightevent, which fires when Squiss stops requesting new messages due to hitting the maxInFlight cap - The
abortedevent, which fires when stop() is called during an active SQS receive message request - Documentation for the already-existing
drainedevent - Documentation for the already-existing
sqsproperty - New option:
activePollIntervalMs, allowing SQS requests to be spaced out when the queue has messages - New option:
idlePollIntervalMs, allowing SQS requests to be spaced out when the queue is empty getQueueUrlmethod to retrieve the configured queue's URL, even if only the name was provided to the constructorcreateQueuemethod to create the configured queuedeleteQueuemethod to delete the configured queuesendMessagemethod to send a message to the configured queuesendMessagesmethod to send an array of messages of any size (within reason) to the configured queuedeleteMessagenow accepts a ReceiptHandle string in lieu of a Message object, allowing messages to be deleted later without caching the full message itself.
- Dropped support for Node 0.12. For Node 0.12 support, consider compiling with an ES6 transpiler such as Babel, or using version 0.7.
- Switched codebase to native ES6, updating to newest TechnologyAdvice style guide
stop()now aborts any ongoing receiveMessage request by default. It also accepts asoftargument (boolean) to soft-stop the poller without the abort functionality.opts.visibilityTimeouthas been renamed toopts.visibilityTimeoutSecsfor consistency.
- pollRetryMs has been added to automatically restart the poller after a failed call to get new messages (dannyrscott)
- maxInFlight can now be set to 0 to uncap message flow (dannyrscott)
- In low-volume use cases, the delete timer would only fire once, and all other deletes depended on filling the queue (dannyrscott)
- Messages failed to stream appropriately when maxInFlight === receiveBatchSize (dannyrscott)
- Specifying unwrapSns now provides message.topicArn and message.topicName.
- Specifying unwrapSns now provides the SNS subject in the message.subject property (dannyrscott)
- An SQS request that ends with no messages now causes
queueEmptyto be emitted with no arguments (dannyrscott)
- Squiss no longer crashes when no messages are found. Apparently we should have said "high volume only" instead of just "high volume".
- opts.correctQueueUrl added to support dynamic testing environments.
- opts.msgFormat is now opts.bodyFormat to avoid confusion.
- Initial release