Consumed OData Service Requirements

Last modified: August 20, 2024

The OData implementation in Mendix does not support all features in the OData specification, nor do we have any plans to provide a full implementation of the entire specification. The supported capabilities of OData are focused on providing a simple and productive way to share data between Mendix apps. Although OData external entities may also work for third-party OData APIs, their use is not tested nor guaranteed. Please validate upfront that the functionality provided in Mendix matches the requirements of your third-party APIs. If external entities do not work with your OData APIs, the advised alternative is to use REST functionality.

Introduction

This document describes the requirements for an OData service that is going to be consumed. These requirements are not further verified at runtime and expected to hold. If these requirements are not met, errors may result.

Requirements for a Consumed OData Service

The requirements for a consumed OData service used in a Mendix app are the following:

The OData service must be either an OData v3 service returning Atom XML, or an OData v4 service returning either Atom XML or JSON
It should support queries on the OData feed, including $filter, $orderby, $top, $skip, $expand, and $count (or $inlinecount)

Requirements on the Service Entities and Attributes

This section describes the features of a consumed OData service that are supported in Mendix apps. These features are checked before an external entity is used in the domain model.

Entities

Vocabulary annotations can be used in a service to indicate features that are not supported. The following vocabulary annotations are recognized for entity sets:

Countable – an entity set marked as Countable="false" prevents the user from adding the entity to the app
Filterable – an entity set marked as Filterable="false" sets all properties as non-filterable
Sortable – an entity set marked as Sortable="false" sets all properties as non-sortable
Marking an entity set as Filterable="false" and Sortable="false" sets all properties as non-filterable and non-sortable; marking properties with the NonFilterableProperties annotation or the NonSortableProperties annotation sets specific attributes as non-filterable or non-sortable
Insertable – an entity marked as Insertable="true" will make the entity creatable in the domain model. That means, for example, that you can model pages that create new objects, and that you can use the entity in the Create Object activity. For insertable entities, the annotations NonInsertableProperties and NonInsertableNavigationProperties list the (navigation) properties that cannot be passed to the service in the Send External Object activity.
Updatable – an entity marked as Updatable="true" with DeltaUpdateSupported="true" and UpdateMethod="2" will make the entity updatable in the domain model. That means, for example, that you can model pages that change attributes values and associated objects, and that you can use the entity in the Change Object activity. For updatable entities, the annotations NonUpdatableProperties and NonUpdatableNavigationProperties list the (navigation)properties that cannot be updated.
Deletable - an entity marked as Deletable="true" can be used in the Delete External Object activity.

The Insertable and Deletable functionalities were introduced in Studio Pro 9.12.0.

An entity can only be used when it is accessible through an entity set.

Furthermore, an entity can only be used if it is uniquely identifiable with a key. The key can consist of one or more properties, as long as the following conditions are met:

The properties cannot be nullable (so they must have isNullable="false" specified).
Only the following types are allowed: Byte, SByte, Int16, Int32, Int64, Boolean, Decimal, Single, Double, and String.
If the type of a key property is String, it must have a limited (maximum) length specified. This is because not all databases support indexes on strings of unlimited length. It is sufficient if a MaxLength is specified in the contract. However, if a MaxLength is not specified in the contract, and you know that the string is limited in length, you can still use the entity by specifying the maximum length of the attribute in the domain model.

This feature of using entities with keys that do not have a maximum length specified in the contract applies to version 9.3.0 and above. In previous versions of Studio Pro, you must change the contract to ensure that MaxLength is specified.

The list above for supported key fields does not include Date or DateTime data types.

Attributes

Attributes marked as FC_KeepInContent=false cannot be used.

The most commonly used attribute types can be used in your app. The types of the attributes in your app will be based on the types of the attributes in the OData metadata, as given in the following table:

OData Type	Mendix Type
Binary	Binary (but see 3.4)
Boolean	Boolean ^[1]
Byte, SByte, Int16, Int32	Integer
DateTime, DateTimeOffset, Time	Date/time
Decimal, Double, Single	Decimal ^[2]
Enumeration	Enumeration
Int64	Long
String, Guid	String
(Other)	(Ignored)

Support for enumerations was introduced in Studio Pro 9.24.0.

When the OData endpoint contains operations, these are not imported in the consumed OData service. You can use a Call REST service activity to call these operations.

^[1] In Mendix, Booleans cannot be null. If the service returns null, the value will be false in Mendix.
^[2] Decimal values outside of the range of a Mendix decimal are currently not supported. If the service returns a value outside of the range, there will be an error.

Generalizations

The consumed OData service does not support importing generalizations and specializations. This means that the Published OData service contract from the originating app will show specializations as discrete entities which will include the attributes of the generalization along with the attributes of the specialized entity.

This means that when you are consuming a Mendix OData endpoint, it is not necessary to consume both a generalization and its specialization. The specialization will now be an entity with all the attributes and associations of the generalization.

Associations to the generalizations with other exposed entities in the published OData service will be included for the now discrete “specialized” entities.

When a generalization and a specialized entity are exposed in the same service. Only the association for the generalization will be visible when both entities are consumed. The now discrete specialization will have the inherited association. A possible work-around for this is to publish a service with the specializations without the generalization. Alternatively, the association for the generalization should not be published, allowing for the inherited association in the specialization to be preserved.

Binary Attributes

The binary data format is supported in the form of media entities. When a media entity is dragged into the domain model, a corresponding external entity is created. The entity will have a contents attribute with the binary data.

Currently, the binary data can only be accessed by Java actions.

Associations

An OData v3 association can only be used if it has two ends.

An OData v4 navigation property can only be used as an association if it has a partner.

When you publish a self-referencing association, you can only publish one side of it. This means that you cannot use the association when you consume the resource as an external entity.

Enumerations

Enumeration types that have one or more members with a name that is not a valid enumeration value name are not supported.