1.258

The ComponentMetadata class in the AuxiliaryContentLoader now includes a data_format field. This allows app scripts to also access the data format of components, which can be used for example for searching or filtering for components in the working pool version.

In multi-stage assembly, users can generate run outputs, import them as run output components, and then use them in subsequent runs. The dependencies between automatically submitted assembly jobs are now taken into consideration as follows:

• The client reports the multi-stage assembly job dependencies to the model server.
• The model server uses these dependencies when assigning jobs to the runner:
  • All job without dependencies (including leaf jobs) are delivered to the runner as usual.
  • Jobs with dependencies are only delivered once all jobs listed as their dependencies have been reported as finished by the runner.

Many solver input file structures require one file to reference other files. This feature provides a way in  Model to define how one file of an assembly is referenced in another, without hard-coding this in a template. A new component property called Reference ID was introduced for this. The workflow is as follows:

• The user first defines the Reference ID on the component (for example, “battery_reference”).
• This Reference ID can then be used in placeholders (e.g., in snippets or other components).

It is now possible to authenticate the LLM by adding the necessary details to the server configuration. The following configuration fields were added:

[ai]
oidc_client_id =
# OIDC client identifier used to obtain access tokens.

oidc_client_secret =
# OIDC client secret used to obtain access tokens.

oidc_idp_url =
# Token endpoint of the OIDC identity provider.

llm_api_url =
# Base URL of the LLM API endpoint

llm_virtual_key =
# Static virtual key associated with the LLM service.

llm_service_account_key =      ( To be implemented...) 
# Service account or master key used to generate short-lived virtual keys dynamically.

Validation rules:

• llm_api_url must always be configured.
• Exactly one of llm_virtual_key or llm_service_account_key must be provided.
• OIDC authentication is optional, but if any OIDC parameter is configured, all required OIDC fields must be present.

The ingestion process was improved so that component documents already present and up-to-date in the OpenSearch index are not re-indexed. The CLI command update_component_search_data no longer blindly re-indexes all components. Instead, only components that require an update are re-indexed, i.e., components that are new, have been modified since they were last indexed, or are otherwise missing or out of date in the OpenSearch index.

This fix applies to runs that do not contain any attributes specific to  Model such as the run UUID. For example, disciplines that are only used with  Result and not with  Model. Previously, an error occurred when executing the add-on “Open with External App” on tests/runs without the attributes LoCo Parameters.RUN_UUID and LoCo Parameters.RUN_NAME. This has been fixed.

Previously, when copying and pasting groups or pool versions with multiple mounted sub-pools (different trees), the paste tab only displayed one of the sub-pools. After applying the paste, the tab did not close automatically and an error was logged. The paste tab now shows all relevant sub-pools, and it closes correctly after applying the changes.

This fix relates to the Preview area. In version 1.248.3 on Red Hat 8, mvz attachments were not displayed correctly. This has been fixed.