Syntax

YAML workflow files contain four parts, ensuring optimal configuration of all required settings.

name: <string>            # optional, defaults to "Unnamed Workflow"
logging: <object|string>  # optional
service: <object>         # optional
containers: [ ... ]       # optional list of container definitions
pipeline: [ ... ]         # ordered list of pipeline steps

logging

Per default, logging is be enabled at the INFO level and log_to_file as false, no logs are written to files as default. Logging accepts a logging level, a log_to_file option and a file_path used as logging path.

logging:
  level: DEBUG            # default "INFO"
  log_to_file: true       # default false
  file_path: logs/delorean.log

service

Per default service is configured with worker_threads : 4, io_threads: 4 and 200 batch_size. worker_threads opnes as many threads and sess

worker_threads

The worker_threads setting dictates how many concurrent Hibernate sessions process incoming inserts/reads/updates concurrently.

• Set this to a maximum of 2× the physical CPU core count on the host machine to prevent context-switching overhead.
• Higher thread counts increase connection pool usage and database lock contention, spiking end-to-end insertion latency instead of improving throughput.

io_threads

The io_threads setting determines the number of concurrent, non-blocking database sessions that execute geometry rendering on unlogged staging tables through inserts.

• Set this to a maximum of 1x the physical CPU core count allocated to the database hosting machine to prevent context-switching overhead.
• Databases that require their displayed geometry to be re-rendered should first be over-provisioned, with only a portion of their CPU being allocated to geometry rendering. A database hosted on a single CPU will be unavailable during the rendering process.

batch_size

The batch_size setting controls how many entities are held in the Hibernate Session state and flushed via JDBC batch statements before resetting session memory.

• Hibernate must hold all managed entity instances in the First-Level Cache (Session memory). For complex AIXM top-level features with deep, geometry-heavy child object graphs, setting batch_size too high causes memory pressure, excessive Garbage Collection pauses, or OutOfMemoryError prior to the final commit.

service:
  worker_threads: 8       # default 4
  io_threads: 4           # default 4
  batch_size: 500         # default 200

containers

A container comprises a name, a database connection, and an AIXM message. Multiple containers can be opened simultaneously to handle different or split datasets in different ways. A database-less container can be created that only operates with AIXM message file in memory.

containers:
  - container: main         # container name
    database:
      url: jdbc:postgresql://localhost:5432/aixm # database connection string jdbc:postgresql://<host>:<port>/<database name>
      username: delorean    # database user
      password: secret      # database password
      hbm2ddl: update       # Hibernate schema generation strategy (none, create)

pipeline

An ordered array of steps. Every step needs an action, certain action require a target, some additional parameters are also required.

pipeline:
  - action: startup          # action name
    target: main             # target container

  - action: unmarshal        # action name
    target: main             # target container
    path: data/input.xml     # additomal attribut path

  - action: persist
    target: main

  - action: shutdown
    target: main

startup

Initializes the target container's runtime state (Hibernate session, connection pool, etc.) so it can accept unmarshal/persist/extract calls.

Parameters

Key	Required	Description
domain-check	no	Boolean. If true, build postgresql domain in the database schema as part of startup. Defaults to false.

- action: startup
  target: main
  domain-check: true

shutdown

Gracefully releases the target container's database connections and runtime resources. No additional parameters.

- action: shutdown
  target: main

unmarshal

Loads an AIXM XML file from disk and loads it into the target container's in-memory model, accepts .xml and .zip extension (as long as it is zipped xml file).

Parameters

Key	Required	Description
path	yes	Path to the source AIXM XML file.
description	no	Free-text label recorded against this unmarshal operation.

- action: unmarshal
  target: main
  path: data/input.xml
  description: Initial AIP snapshot

marshal

Serializes the target container's current in-memory model back out to an AIXM XML file, accepts .xml and .zip extension and will automatically zip the serialized AIXM message.

Parameters

Key	Required	Description
path	yes	Destination path for the output XML file.

- action: marshal
  target: main
  path: out/result.xml

sax_inspection

Runs a SAX-based structural pass over the target container's loaded AIXM message and reports parsing-level issues. No additional parameters.

- action: sax_inspection
  target: main

temporality_inspection

Inspects the target container's AIXM message time-slice structure for temporal consistency issues (overlaps, gaps, ordering). No additional parameters.

- action: temporality_inspection
  target: main

inspect_persisted_message

Using the container database connection, inspects already persisted AIXM message. No additional parameters.

- action: inspect_persisted_message
  target: main

summary_inspection

Produces a summary inspection report (feature counts, type breakdown) for the target container AIXM message. No additional parameters.

- action: summary_inspection
  target: main

persist

Writes the target container's AIXM message into the database. No additional parameters.

- action: persist
  target: main

merge

Merges the target container's AIXM message into an persisted AIXM message using the target container database connections, matched by a field/value pair (e.g. merging by hjid).

This will add new timeslices to existing features, ignore old or duplicate timeslices, and add new features that have not yet been persisted in the existing message.

Parameters

Key	Required	Description
field	yes	Field name to match on (e.g. hjid).
value	yes	Value to match against. Integers are parsed as int, otherwise treated as text.

- action: merge
  target: main
  field: hjid
  value: 10432

diff

Computes a diff for the target container's AIXM message and unmarshalling the output to a AIXM file at path. This turn a BASELINE AIXM message containing all the features and timeslices into a PERMDELTA AIXM message by keeping only the changed features. The time range is assumed to have been defined previously, either by a predicate or a timeslice action.

Parameters

Key	Required	Description
path	yes	Path to the AIXM XML file to diff against.

- action: diff
  target: main
  path: data/previous.xml

extract

Extracts AIXM message from the target container matching a field/value pair, staging them for a subsequent marshal.

Parameters

Key	Required	Description
field	yes	Field name to match on.
value	yes	Value to match against. Integers are parsed as int, otherwise treated as text.

- action: extract
  target: main
  field: hjid
  value: 10432

predicate

Extracts AIXM message from the target container that are active past a given timestamp, filtered further by a field/value pair.

Parameters

Key	Required	Description
time	yes	ISO-8601 timestamp (e.g. 2026-09-01T12:00:00Z) defining the active time slice.
field	yes	Field name to match on.
value	yes	Value to match against. Integers are parsed as int, otherwise treated as text.

- action: predicate
  target: main
  time: "2026-09-01T12:00:00Z"
  field: hjid
  value: 10432

integrate

Integrates an external PERMDELTA AIXM message file into the target container's AIXM message. Merges a PERMDELTA AIXM message into a existing Baseline that converting partial timelice into full timelices. Can then be merged into the persisted AIXM messaege.

Parameters

Key	Required	Description
path	yes	Path to the AIXM XML file to integrate.

- action: integrate
  target: main
  path: data/patch.xml

render

Triggers rendering of geometry and materialized views for the target container. No additional parameters.

- action: render
  target: main

set_status

Sets a status flag on the target container.

Parameters

Key	Required	Description
status	yes	Status value to apply.

- action: set_status
  target: main
  status: ACTIVE

filter

set_context

Set a new registerd context as global context that converts gml:ID into UUID. This process allows multiple AIXM messages to coexist in the same database, mapping the same gml:ID to the same UUID using the same context.

Parameters

Key	Required	Description
description	yes	Description to set as the active context.

- action: set_context
  description: EAD cycle 2026-09

register_context

Set a existing registerd context as global context that converts gml:ID into UUID. This process allows multiple AIXM messages to coexist in the same database, mapping the same gml:ID to the same UUID using the same context.

Parameters

Key	Required	Description
salt	yes	Unique key/identifier for the context.
description	yes	Description associated with the context.

- action: register_context
  salt: cycle-2026-09
  description: EAD cycle 2026-09

clear_contexts

Clears all registered contexts from the processor. No additional parameters, no target.

- action: clear_contexts

un_set_active_context

Unsets the currently active context on the processor, without clearing the registered contexts. No additional parameters, no target.

- action: un_set_active_context

filter

Applies a filter to the target container's feature or time-slice model, narrowing what subsequent actions operate on.

Parameters

Key	Required	Description
type	yes	One of feature_indentifier, feature_type, timeslice_valid_time.
nullHandling	yes	How to treat missing/null data during evaluation (consumed by the base filter specification).
evaluationType	yes	Evaluation mode for the filter (consumed by the base filter specification).

Filter types

type value	Specification class	Applies to
feature_indentifier	FeatureIdentifierSpecification	Feature filter
feature_type	FeatureTypeSpecification	Feature filter
timeslice_valid_time	TimeSliceValidTimeSpecification	Time-slice filter

feature_indentifier

Matches features whose gml:identifier value is in a given list (case-insensitive).

Key	Required	Description
identifiers	yes	Array of feature identifier strings to match against.

- action: filter
  target: main
  type: feature_indentifier
  nullHandling: EXCLUDE
  evaluationType: INCLUDE
  identifiers:
    - EG-RWY-0001
    - EG-RWY-0002

feature_type

Matches features by Java class simple name is in a given list types.

Key	Required	Description
types	yes	Array of feature type names (matched against feature.getClass().getSimpleName(), lowercased).

- action: filter
  target: main
  type: feature_type
  nullHandling: EXCLUDE
  evaluationType: INCLUDE
  types:
    - RunwayElement

timeslice_valid_time

Matches time slices whose valid-time begin/end fall before, after, or at given instants. All fields are optional individually, but at least one is needed for the filter to do anything.

Key	Required	Description
beforeBegin	no	ISO-8601 instant — valid-time begin must be strictly before this.
afterBegin	no	ISO-8601 instant — valid-time begin must be strictly after this.
atBegin	no	ISO-8601 instant — valid-time begin must equal this.
beforeEnd	no	ISO-8601 instant — valid-time end must be strictly before this.
afterEnd	no	ISO-8601 instant — valid-time end must be strictly after this.
atEnd	no	ISO-8601 instant — valid-time end must equal this.

- action: filter
  target: main
  type: timeslice_valid_time
  nullHandling: EXCLUDE
  evaluationType: INCLUDE
  afterBegin: "2026-01-01T00:00:00Z"
  beforeEnd: "2026-12-31T23:59:59Z"

prune

Prunes data from one container into another, using the same filter-specification types and JSON shape as filter (see above) to select what gets pruned. Uses from/to instead of target.

Parameters

Key	Required	Description
from	yes	Name of the source container.
to	yes	Name of the destination container.
type	yes	One of feature_indentifier, feature_type, timeslice_valid_time — same filter types as filter.
(additional)	depends	Same per-type keys as filter (see above).

- action: prune
  from: staging
  to: main
  type: timeslice_valid_time
  nullHandling: EXCLUDE
  evaluationType: INCLUDE
  afterBegin: "2026-01-01T00:00:00Z"

clone

Clones one container's state into another. Uses from/to instead of target.

Parameters

Key	Required	Description
from	yes	Name of the container to clone into.
to	yes	Name of the container being cloned.

- action: clone
  from: backup
  to: main