New statement code shape allows you to write more complex C# code statements, e.g. when you want to use conditionals like if-else when building messages. Previously, this would have required separating the condition out from the code to multiple steps, or creating a custom task. Now you can write short code statements that can e.g. assign variables or create complex objext

The Process details view has been updated to show more details of the Process instance itself, as well as executed task details. 

The maintenance jobs could delete step data for instances not yet marked for deletion when using Azure blob store

Sometimes the log data for a step inside an inner loop might not be shown in the UI, due to wrong index in the blob name.

Results from steps logged as blobs can be copied to clipboard. You can now also view the raw JSON directly in the UI.

Test process execution results can be viewed, an error is no longer thrown. 

Schedule trigger data format changed slightly in 5.2, and the automatic migration set the repeat delays wrong if the delay type was minutes or hours. Now the migration has been fixed, but unfortunately the delay setting for any Processes that have been migrated and saved, need to be manually fixed.

Importing tasks now correctly resolves package references from the configured external NuGet repositories, so  e.g. FRENDS Cobalt tasks can again be imported.

There was another problem with the schedule trigger data migration, this time with copied or rebuilt Processes. The data was not migrated correctly, potentially causing the copied or rebuilt Processes to use wrong repeat delays.
Please note that the wrongly migrated schedules will need to be corrected manually here as well.

Parameters for failed Tasks were not logged by default when using the new step logging 

Processes were incorrectly marked as TerminatedByUser when a Task threw an OperationCanceledException. This could happen e.g. with the Frends.Web.HttpRequest Task request timing out. 

Queue trigger could not connect to durable RabbitMQ queues. Now there is a "Queue is durable" parameter that can be set, so the AMQP listener can connect. Also the connections are now released correctly in case of these errors.
Please note that in order for the new setting to work, the FRENDS Agent will also need to be updated.

Opening the API deployment dialog failed if the API had tens of Processes and hundreds of versions. The faulty SQL query has now been fixed.

Default timezone was not being initialized correctly to new schedule triggers. The view showed the client local value, e.g. "(UTC+2) Helsinki" by default, but that value was not saved. This meant the Agent actually defaulted to the local time zone on the host, which e.g. for the cloud Agents meant UTC time.
Now the time zone is initialized correctly. Any existing triggers that did not have the value initialized should now show a descriptive message that the Agent local time zone is being used.

The UI could not fetch the log data for Processes built with 5.2.0 and 5.2.1 while they were still executing.

Agent may crash when using a Queue Trigger

Cannot call a Subprocess from inside of a Conditional Trigger

Process execution viewing efficiency problems

Cannot use characters '$', '_' or '.' with Agent usernames

Error handler not shown correctly for Process when using blob logging

Security improvements

Null coalescing operator (e.g. #var.thisMightBeNull ?? "empty") not working correctly in text expressions