Arkindex 1.12.2

The workers list page has been redesigned to make workers easier to discover and manage. The details of a single worker are now on a fully separate page, and the additional space reclaimed for the list allows us to show the worker description and some statistics about each worker.

As worker types are more emphasized with this new design, we have also added a new management page to help instance administrators edit existing worker types and assign them custom colors. Worker types cannot be created manually, as they are intended to only be created when importing a worker version.

Existing template processes can now be edited just like Workers or Dataset processes. Their selected worker versions, model versions, worker configurations and dependencies can be modified.

A process that uses a template process will not be updated when the template process gets updated.

The CloneProcess API endpoint introduced in the previous release is now available through Actions  Clone process when viewing the status of a process.

Clicking this new action will create a new process with the same element filters or selected datasets, and with the same worker versions, model versions and worker configurations.

The S3 bucket import now directly displays the worker configuration defined by the S3 import system worker. This allows the worker to define its own import options and to make developing additional S3 import features faster.

The worker included along with this release uses this new system to provide a new option to filter files by a list of suffixes, for example .jpg and .png.

The following API endpoints have been migrated to the new error format:

The documentation for each of those API endpoints contains a detailed description of the structure of this new format.

To upgrade a development instance, follow this documentation.

To upgrade a production instance, you need to:

The main changes impacting developers and system administrators are detailed below.