- The desired capabilities array should be omitted if it is empty.
- The producer now verifies that the input message follows the expected structure.
- Fix the definition of the earliest event position.
- Support Geographic replication to enable recovery in case of geographic disasters.
- The
offset
property of event data is now typed as a string instead of as a number.
- Adds support for connecting to the development emulator. The connection string for the development emulator should have the
";UseDevelopmentEmulator=true"
slug.
- The minimum value of timeout for all operations is no longer 60 seconds. The user can now set the timeout to lower values if needed. The default timeout value is still 60 seconds.
- Improve event prefetching to not overload the internal queue.
- NodeJS v18 is now the minimum version supported. Check out the LTS versions of Node.js for more information on NodeJS support timelines. And check out the Microsoft Support Policy for more information on Microsoft support timelines.
- Fixes a memory leak #25426 in the
EventHubBufferedProducerClient
when there is no activity for a longer interval after sending some events. #26748
- Fix a regression of missing
getToken
calls when renewing tokens.
- Adds an option to set an identifier for Event Hubs clients.
- Adds a property on each Event Hub client that returns the identifier of the client.
- Use Rhea's prefetch window to prefetch events from the service. This improves the performance of the receiver by reducing the number of round trips to the service. The default prefetch window is 3 *
maxBatchSize
events. This can be configured by setting theprefetchCount
option on thesubscribe
method onEventHubConsumerClient
.
- Fix a memory leak issue of linked child abort signals not being released.
- upgrade dependency
rhea-promise
version to^3.0.0
.
- Fixing a bug where events were not always received in order [#23993].
- The receiver no longer attempts to build batches of
maxMessageCount
size, instead, it returns batches as soon as they are received from the service, up tomaxMessageCount
.
- The
enableIdempotentPartitions
flag has been renamed toenableIdempotentRetries
MessageWithMetadata
has been renamed toMessageContent
.MessageContent
'sbody
has been renamed todata
.MessageAdapter
'sconsumeMessage
andproduceMessage
have been renamed toconsume
andproduce
.
- The Uint8Array payload was being stringified first before it gets sent which caused the receiver to treat it as an object instead of a Uint8Array. This is now fixed and Uint8Array is being treated the same as a Buffer.
- The hashing algorithm used to map partition keys to IDs in the buffered producer is no longer sensitive to the endianness of the local machine Issue #21190.
- Updated our
@azure/core-tracing
dependency to the latest version (1.0.0). There are no changes from the previous beta; however, please see below for changes from the previous minor version:- Notable changes include Removal of
@opentelemetry/api
as a transitive dependency and ensuring that the active context is properly propagated. - Customers who would like to continue using OpenTelemetry driven tracing should visit our OpenTelemetry Instrumentation package for instructions.
- Notable changes include Removal of
createEventDataAdapter
is updated so thatconsumeMessage
returns the original binary payload instead of the Buffer representation of it.
- Updated our
@azure/core-tracing
dependency to the latest version (1.0.0-preview.14)- Notable changes include Removal of
@opentelemetry/api
as a transitive dependency and ensuring that the active context is properly propagated. - Customers who would like to continue using OpenTelemetry driven tracing should visit our OpenTelemetry Instrumentation package for instructions.
- Notable changes include Removal of
-
A new function
createEventDataAdapter
is exported that can convert anEventData
to a simple message withbody
andcontentType
fields. This adapter can be used with the Avro encoder in@azure/schema-registry-avro
starting from version 1.0.0-beta.6 to createEventData
messages with Avro-encoded bodies. -
When publishing events to Event Hubs, timeouts or other transient failures may introduce ambiguity into the understanding of whether a batch of events was received by the service. To assist in this scenario, the option to publish events idempotently has been added to
EventHubBufferedProducerClient
. The functionality can be enabled by setting theenableIdempotentPartitions
client option totrue
. If enabled, retries during publishing will attempt to avoid duplication with a minor cost to throughput. Duplicates are still possible but the chance of them occurring is much lower when idempotent retries are enabled. PR #20156
- Added
EventHubBufferedProducerClient
with functionality to manage batching, concurrency, and sending of events implicitly. This abstracts the complexity away from applications regarding publishing events in an optimal fashion. See issue #17699 for more details. - Added
skipParsingBodyAsJson
optional parameter toEventHubConsumerClient.subscribe
method. When set totrue
it will disable the client from runningJSON.parse()
on the message body when receiving the message. Not applicable if the message was sent with AMQP body typevalue
orsequence
.
- Prevent empty spans from being created when tracing is disabled. Fixes issue #14063
- Updated to use the version
1.0.0-preview.13
of the@azure/core-tracing
dependency.
-
With the dropping of support for Node.js versions that are no longer in LTS, the dependency on
@types/node
has been updated to version 12. Read our support policy for more details. -
Updated our internal core package dependencies to their latest versions in order to add support for Opentelemetry 1.0.0 which is compatible with the latest versions of our other client libraries.
-
Changed TS compilation target to ES2017 in order to produce smaller bundles and use more native platform features
-
Adds the
contentType
,correlationId
, andmessageId
AMQP properties as top-level fields onEventData
andReceivedEventData
. -
Enable encoding the body of a message to the 'value' or 'sequence' sections (via AmqpAnnotatedMessage.bodyType). Using this encoding is not required but does allow you to take advantage of native AMQP serialization for supported primitives or sequences.
More information about the AMQP message body type can be found in the AMQP specification: link
- Fixes issue #13500 where a
TypeError: Cannot read property '_process' of undefined
could be thrown in rare cases.
-
Fixes a race condition that would cause connection recovery to sometimes fail if a consumer or producer was closed at the same time a connection was disconnected.
-
Fixes issue #14606 where the
EventHubConsumerClient
could call subscribe'sprocessError
callback with a "Too much pending tasks" error. This could occur if the consumer was unable to connect to the service for an extended period of time. -
Fixes issue #15002 where in rare cases an unexpected
TypeError
could be thrown fromEventHubProducerClient.sendBatch
when the connection was disconnected while sending events was in progress.
- Updates the methods on the
CheckpointStore
interface to accept an optionaloptions
parameter that can be used to pass in anabortSignal
andtracingOptions
.
- Allows passing
NamedKeyCredential
andSASCredential
as the credential type toEventHubConsumerClient
andEventHubProducerClient
. These credential types support rotation via theirupdate
methods and are an alternative to using theSharedAccessKeyName/SharedAccessKey
orSharedAccessSignature
properties in a connection string.
- Tracing options for
EventDataBatch.tryAdd
now match the shape ofOperationOptions
.
-
Adds the
customEndpointAddress
field toEventHubClientOptions
. This allows for specifying a custom endpoint to use when communicating with the Event Hubs service, which is useful when your network does not allow communicating to the standard Event Hubs endpoint. Resolves #12901. -
A helper method
parseEventHubConnectionString
has been added which validates and parses a given connection string for Azure Event Hubs. Resolves #11894 -
Re-exports
RetryMode
for use when setting theRetryOptions.mode
field inEventHubConsumerClientOptions
orEventHubClientOptions
. Resolves #13166. -
Updates documentation for
EventData
to call out that thebody
field must be converted to a byte array orBuffer
when cross-language compatibility while receiving events is required.
- Fixes issue #12278
where the
processEvents
handler could ignore themaxWaitTimeInSeconds
parameter after a disconnection.
- Adds
loadBalancingOptions
to theEventHubConsumerClient
to add control around how aggressively the client claims partitions while load balancing. (PR 9706). - Support using the SharedAccessSignature from the connection string. (PR 10951).
- Fixes issue #9704 where events could be skipped while receiving messages. Previously this could occur when a retryable error was encountered and retries were exhausted while receiving a batch of events.
- Fixes issue #10298
where spans had inconsistent
peer.address
attributes by removing the scheme (i.e.sb://
) from EventHubpeer.address
span attributes - Addresses #10276: adds
message_bus.destination
andpeer.address
attributes toAzure.EventHubs.message
spans. (PR 10389)
- Adds
loadBalancingOptions
to theEventHubConsumerClient
to add control around how aggressively the client claims partitions while load balancing. (PR 9706)
- Fixes issue #9289
where calling
await subscription.close()
inside of a subscription'sprocessError
handler would cause the subscription to deadlock. - Fixes issue #9083
where calling
EventHubConsumerClient.close()
would not stop any actively runningSubscriptions
. - Fixes issue #8598
where the EventHubConsumerClient would remain open in the background beyond
when
subscription.close()
was called. This would prevent the process from exiting until themaxWaitTimeInSeconds
(default 60) was reached. - Updated to use the latest version of the
@azure/core-amqp
package. This update fixes issue #9287 where some failed operations would delay the process exiting.
- Fixes issue #8584
where attempting to create AMQP links when the AMQP connection was in the
process of closing resulted in a
TypeError
in an uncaught exception. (PR 8884) - Fixes reconnection issues by creating a new connection object rather than re-using the existing one. (PR 8884)
Tracing functionality is still in preview status and the APIs may have breaking changes between patch and minor updates.
-
Updates the
peer.address
attribute on "Azure.EventHubs.send" spans to refer to the fully qualified namespace instead of the endpoint. Fixes #7109 -
Adds a new attribute -
enqueuedTime
- to the links on "Azure.EventHubs.process" spans.enqueuedTime
maps to the enqueuedTimeUtc field from received events, represented as Unix epoch time in milliseconds. Address #7112
- Updates the
EventHubProducerClient.sendBatch
API to accept an array of events. (PR #8622) - Adds compatibility with TypeScript versions 3.1 through 3.6+. (PR #8654)
- Improves the performance of the
EventDataBatch.tryAdd
method. (PR #8637)
- Addresses issue #7801 by moving known AMQP message properties to received events'
systemProperties
. (PR #7973)
- Fixes an issue that caused the
getPartitionIds
,getEventHubProperties
, andgetPartitionProperties
methods to throw an error when run against an Event Hub in Azure Stack due to missing security token on the outgoing request. (PR #7463)
- Fixed a potential issue with deadlock where greedy consumers could starve out other consumers, preventing us from properly balancing.
- Fixed an issue where calling
subscription.close()
immediately after callingsubscribe
would cause events to still be read. - Updated to use the latest version of the
@azure/core-amqp
package which in turn uses the latest version of therhea
package. This update improves support for bundling this library. (Pull Request)
- This release marks the general availability of the
@azure/event-hubs
package. - Fixed potential issues with claims being mismanaged when subscriptions terminate.
- Improved reporting of errors that occur when attempting to claim partitions from CheckpointStores.
- Updated to use the latest version of the
@azure/core-amqp
package. This update allows the SDK to detect when a connection has gone idle for 60 seconds and attempt to reconnect.
- Starting event positions are now passed in the
options
to thesubscribe()
method instead of using theprocessInitialize()
callback. - If no position is passed and no checkpoints are available for the partition,
the
subscribe()
method will start receiving events that are queued after the method invocation. This is different from the last preview, where events were received from the beginning of the partition. - The
MessagingError
class is updated to have thecode
property instead ofname
to contain the error type that the user can use to differentiate errors that can occur during communication with the service. Thename
property of this class will always have the value "MessagingError" and will not change based on the error type. - System errors around network issues like ENOTFOUND, ECONNREFUSED will retain their
code
value even after getting converted to aMessagingError
object and being passed to the user.
- Improves load-balancing capabilities to reduce the frequency that partitions are claimed by other running
instances of
EventHubConsumerClient.subscribe
after all partitions are being read. (PR #6294) - Updated to use OpenTelemetry 0.2 via
@azure/core-tracing
- CheckpointStore and consumer group are now passed to the EventHubConsumerClient constructor rather than being passed to subscribe().
- Updated to use the latest version of the
@azure/core-tracing
&@azure/core-amqp
package.
EventHubsClient
has been split into two separate clients: EventHubProducerClient
and EventHubConsumerClient
The EventHubConsumerClient
provides several overloads for subscribe
which all take event handlers rather than
requiring an EventProcessor
. There are no longer any methods that directly return ReceivedEventData
- all
receiving is done via event handlers.
The EventHubProducerClient
has standardized on only providing sending via sendBatch
.
Construction of both objects is the same as it was for the previous client.
- Updated to use the latest version of the
@azure/core-tracing
package.
- Current implementation of the Partition Manager takes the event hub name, consumer group name and partition id to ensure uniqueness for the checkpoint and ownership.
Since the same event hub name and consumer group name can exist in another namespace, we added
fullyQualifiedNamespace
as well to ensure uniqueness. (PR #5153) - Adds preview capabilities for
@azure/core-tracing
. (PR #5207
-
Removed the
createFromIotHubConnectionString
method fromEventHubClient
. (PR #5311). Instead, pass an Event Hubs-compatible connection string when instantiating anEventHubClient
to read properties or events from an IoT Hub.Previously:
const client = await EventHubClient.createFromIotHubConnectionString(iotConnectionString);
Current:
const client = new EventHubClient(iotEventHubsCompatibleConnectionString);
-
The
PartitionContext
properties have been merged into thePartitionProcessor
class. ThePartitionProcessor
now exposes anupdateCheckpoint
method that can be called to checkpoint events.
- Adds load-balancing capabilities to
EventProcessor
.EventProcesor
will use the data fromPartitionManager
to regulate how many partitions it should read from. (PR #4839). - Adds
lastEnqueuedEventInfo
property onEventHubConsumer
. When the consumer is created withtrackLastEnqueuedEventInfo
set totrue
, thelastEnqueuedEventInfo
field is updated everytime a message is received and provides details about the last enqueued event in the partition theEventHubConsumer
is reading from. (PR #5036) - Received event data will now expose
systemProperties
for message annotations set by the service. (PR #5008). - Improved error messages when constructing an
EventHubClient
with an invalid connection string and Event Hub name combo. (PR #4899). - Adds client-side type-checking for
EventPosition
static helper methods. (PR #5052).
- The
PartitionProcessor
interface is now a class with default implementations ofinitialize
,close
,processEvents
, andprocessError
. (PR #4994).- Users should extend the
PartitionProcessor
class and override any of the methods that need custom logic. - All 4 methods now accept
PartitionContext
as the last parameter.PartitionContext
contains information about the partition being processed, as well as theupdateCheckpoint
method that can be used to persist a checkpoint.
- Users should extend the
- The
EventProcessor
constructor was changed to no longer accept aPartitionProcessorFactory
function that returns aPartitionProcessor
. Instead, users should extend thePartitionProcessor
class and pass the class (not an instance of the class) to theEventProcessor
constructor. (PR #4994).
- The sender is refactored to avoid the warning around too may listeners being attached which would occur before if too many send requests were in flight at the same time from the same sender.
- The receiver is refactored to allow the same underlying AMQP link to be shared between streaming and batching mode. This results in seamless transition between the three different receive methods on the
EventHubConsumer
- All time related entites have been updated to use milli seconds as the unit of time for consistency.
- New error
InsufficientCreditError
is introduced for the scenario where rhea is unable to send events due to its internal buffer being full. This is a transient error and so is treated as retryable. - The error
OperationTimeoutError
was previously mistakenly classified as an AMQP error which is now corrected. Since this can also be a transient error, it is treated as retryable.
- Added method
createBatch()
on theEventHubProducer
to create anEventDataBatch
that can then be used to add events until the maximum size is reached.- This batch object can then be used in the
send()
method to send all the added events to Event Hubs. - This allows publishers to build batches without the possibility of encountering the error around the message size exceeding the supported limit when sending events.
- It also allows publishers with bandwidth concerns to control the size of each batch published.
- This batch object can then be used in the
- Introduced a new class
EventProcessor
which replaces the older concept of Event Processor Host.- This early preview is intended to allow users to test the new design using a single instance of
EventProcessor
. The ability to store checkpoints to a durable store will be added in future updates
- This early preview is intended to allow users to test the new design using a single instance of
- The properties on the
RetryOptions
interface have been renamed for ease of use. - New property
timeoutInMs
onRetryOptions
to configure the time to wait before declaring an attempt to have failed withOperationTimeoutError
error which is retryable. - New properties
mode
andmaxRetryDelayInMs
onRetryOptions
to configure the exponential retry mode that is now supported
Version 5.0.0-preview.1 is a preview of our efforts to create a client library that is user friendly and idiomatic to the Javascript ecosystem. The reasons for most of the changes in this update can be found in the Azure SDK Design Guidelines for TypeScript. For more information, please visit https://aka.ms/azsdk/releases/july2019preview
- Creating an instance of
EventHubClient
is now done using constructor overloads instead of static helpers.- If you previously used the
createFromTokenProvider
static helper to provide your own custom token provider, you will now need to update the provider to follow the newTokenCredential
interface instead. - If you previously used the
@azure/ms-rest-nodeauth
library to provide AAD credentials, you will now need to use the new @azure/identity library instead.
- If you previously used the
- The send methods are moved from the
EventHubClient
class to the newEventHubProducer
class.- Use the
createProducer()
function on theEventHubClient
to create an instance of aEventHubProducer
. - Each producer represents a dedicated AMQP sender link to Azure Event Hubs.
- The
EventData
type used for the data being sent only supports abody
for the content being sent and aproperties
bag to hold any custom metadata you want to send. The properties corresponding to a received event are removed from this type and a separate type ReceivedEventData is used for received events.
- Use the
- The receive methods are moved from the
EventHubClient
class to the newEventHubConsumer
class.- Use the
createConsumer()
function on theEventHubClient
to create an instance of aEventHubConsumer
. - Each consumer represents a dedicated AMQP receiver link to Azure Event Hubs based
on the flavor of receive function being used i.e
receiveBatch()
that receives events in a batch vsreceive()
that provides a streaming receiver. - The static methods
EventPosition.fromStart()
andEventPosition.fromEnd()
are renamed toEventPosition.earliest()
andEventPosition.latest()
respectively.
- Use the
- Inspecting Event Hub
- The methods
getHubRuntimeInformation()
andgetPartitionInformation()
on theEventHubClient
are renamed togetProperties()
andgetPartitionProperties()
respectively. Please refer to the return types of these functions to ensure you are using the right property names.
- The methods
- You can now configure retry options that are used to govern retry attempts when a retryable error occurs. These can be
set when creating the
EventHubClient
,EventHubProducer
andEventHubConsumer
- You can now pass an abort signal to any of the async operations. This signal can be used to cancel such operations. Use the package @azure/abort-controller to create such abort signals.
- An async iterator is now available to receive events after you create an instance of
EventHubConsumer
. Use the functiongetEventIterator()
on the consumer to get aAsyncIterableIterator
which you can then use in a loop or use it'snext()
function to receive events.
- Refer to the
API reference documentation
to get an overview of the entire API surface. - Refer to our samples to understand the usage of the new APIs.
- Added support for WebSockets. WebSockets enable Event Hubs to work over an HTTP proxy and in environments where the standard AMQP port 5671 is blocked. Refer to the websockets sample to see how to use WebSockets.
@types/async-lock
has been moved to being a dependency from a dev-dependency. This fixes the bug 3240
- If you have been using the
createFromAadTokenCredentials
function to create an instance of theEventHubClient
, you will now need to use the @azure/ms-rest-nodeauth library instead of ms-rest-azure library to create the credentials that are needed by thecreateFromAadTokenCredentials
function. - Typescript: Replaceimport * from "ms-rest-azure";
withimport * from "@azure/ms-rest-nodeauth";
- Javascript: Replacerequire("ms-rest-azure")
withrequire("@azure/ms-rest-nodeauth")
- If you have been passing a non string value in the
partitionKey
property on the message when sending it using theEventHubClient
, an error will now be thrown. This property only supports string values.
- A network connection lost error is now treated as retryable error. A new error with name
ConnectionLostError
is introduced for this scenario which you can see if you enable the logs. - When recovering from an error that caused the underlying AMQP connection to get disconnected, rhea reconnects all the older AMQP links on the connection resulting in the below 2 errors in the logs. We now clear rhea's internal map to avoid such reconnections. We already have code in place to create new AMQP links to resume send/receive operations. - InvalidOperationError: A link to connection '.....' $cbs node has already been opened. - UnauthorizedError: Unauthorized access. 'Listen' claim(s) are required to perform this operation.
- Enabled the
esModuleInterop
compilerOption in thetsconfig.json
file for this library to be compliant with the best practices.
- Use
isItselfClosed()
instead ofisClosed()
in rhea to correctly determine if the sdk initiated close on receiver/sender. This ensures that on connection issues like the ECONNRESET error, the receivers get re-connected properly thus fixing the bug 174
- Only set
message_id
while sending the message, when provided by caller PR.
- export
EventHubConnectionConfig
from the library.
- Moved
lib/amqp-common
to"@azure/amqp-common"
package and took a dependency on it. - Moved
lib/rhea-promise
to"rhea-promise"
package and took a dependency on it. - Fixed issues where the private instance of
rhea receiver or sender
were undefined when*_open
and*_close
events happened instantaneously.
- update the version of ms-rest-azure to "2.5.9"
EventPosition.fromSequenceNumber()
accepts0
as a valid argument.client.receive()
andclient.receiveBatch()
accept partitionId as astring | number
.- User's error handler in
client.receive()
will only be notified if the user did not close the receiver and the error is not retryable.
client.getPartitionInformation()
should works as expected when partitionId is of typenumber | string
.
- Stable version of the libray.
- Added support to provide custom user-agent string that will be appended to the default user agent string.
- Updated examples and content in README.md
- Fixed issue
- Added error handlers to the $management sender/receiver links.
- Added error handlers to the amqp session of the $management and $cbs sender/receiver links.
- Exported
AadTokenProvider
andSasTokenProvider
from the client.
- Improved logging statements to the connection context.
- Added timeout to promisifed creation/closing of rhea sender, receiver, session, connection.
- Fixed a bug in the EventData deserialization logic by checking for
!= undefined
check rather than the!
check. - While handling disconnects we retry for 150 times at an interval of 15 seconds as long the error is
retryable
.
- Improved log statements.
- Documented different mechanisms of getting the debug logs in README.
- Minimum dependency on
"rhea": "^0.2.18"
. - Fixed bugs in recovery logic
- Added support to recover from session close for sender and receiver
- Added a new property
isConnecting
that provides information whether a linkEntity is currently in the process of establishing itself. - Using
is_closed()
method of sender, receiver and session in rhea to determine whether the sdk initiated the close. - MessagingError is retryable by default.
- Added support to translate node.js
SystemError
into AmqpError. - Added a new static method
createFromTokenProvider()
on the EventHubClient where customers can provide their own TokenProvider.
- Improved log statements
- Updated README.md
- Updated dependency rhea to "^0.2.16" instead of github dependency.
- Added support to handle disconnects and link timeout errors.
- Fixed client examples link in README.
- Updated issue templates
- Improvised the example structure
- Moved the common stuff to
amqp-common
and addedConnection
,Session
,Sender
,Receiver
objects torhea-promise
. - Improved tsconfig.json and tslint.json config files.
- Added
import "mocha"
to all the test files, inorder to get rid of red squiggles in vscode. - Replaced crypto with jssha which is browser compatible
- Minor doc fixes and sample updates.
- Add a listener for the disconnected event after opening the connection.
- Fixed the partitionkey issue while sending events. #73.
- Bumped the minimum dependency on rhea to 0.2.13. This gives us type definitions for rhea.
- rpc.open() returns the connection object. This makes it easy to extract common functionality to a separate library.
- Added support to create EventHubClient from an IotHub connectionstring. The following can be done
const client = await EventHubClient.createFromIotHubConnectionString(
process.env.IOTHUB_CONNECTION_STRING
);
- Internal design changes:
- ManagementClient also does cbs auth before making the management request.
- EventHubSender, EventHubReceiver, ManagementClient inherit from a base class ClientEntity.
- Moved opening the connection to CbSClient as that is the first thing that should happen after opening the connection. This reduces calls to
rpc.open()
all over the sdk and puts them at one place in theinit()
method on the CbsClient.
- Added functionality to encode/decode the messages sent and received.
- Created an options object in the
client.createFromConnectionString()
and theEventHubClient
constructor. This is a breaking change. However moving to an options object design reduces the chances of breaking changes in the future. This options object will: - have the existing optional
tokenProvider
property - and a new an optional property named
dataTransformer
. You can provide your own transformer. If not provided then we will use the DefaultDataTransformer. This should be applicable for majority of the scenarios and will ensure that messages are interoperable between different Azure services. It fixes issue #60.
- Added missing dependency for
uuid
package and nit fixes in the README.md
- Changing
client.receiveOnMessage()
toclient.receive()
as that is a better naming convention and is in sync with other language sdks.
- Previously we were depending on amqp10 package for the amqp protocol. Moving forward we will be depending on rhea.
- The public facing API of this library has major breaking changes from the previous version 0.0.8. Please take a look at the Readme and the examples directory for detailed samples.
- Removed the need to say
client.open.then()
. First call to create a sender, receiver or get metadata about the hub or partition will establish the AMQP connection. - Added support to authenticate via Service Principal credentials, MSITokenCredentials, DeviceTokenCredentials.
- This should make it easy for customers to login once using the above mentioned credentials,
- Create the EventHubs infrastructure on the Azure management/control plane programmatically using (azure-arm-eventhubs) package over HTTPS prtocol.
- Use the same credentials to send and receive messages to the EventHub using this library over AMQP protocol.
- This should make it easy for customers to login once using the above mentioned credentials,
- Provided a promise based API to create senders/receivers off the
EventHubClient
. - Added capability to send multiple messages by batching them together.
- Added capability to receive predefined number of messages for a specified amount of time. Note that this method will receive all the messages and return an array of EventData objects.
- Added capability to create an epoch receiver.
- Simplified the mechanism to specify the
EventPosition
from which to receive messages from the EventHub. - Added proper TypeScript type definitions to the library that improves the intellisense experience for our customers.
- Fixed a race condition within the AMQP redirection code when using an IoT Hub connection string.
- Disabled auto-retry of AMQP connections in amqp10 since the current client is not built to handle them and fails when retrying.
- Pulled changes for #14 and #20/#21.
- Special thanks to @kurtb and @ali92hm for their contributions!
- Added support for message properties in the EventData structure.