5.2. Attribute storage

The Write Attributes and Discover operations, as well as the Information Reporting interface, use a concept of Attributes that may be set for Resources, Object Instances or Objects.

See also

OMA LwM2M - Brief description chapter contains more information about LwM2M Attributes.

When the library needs to read or write these attributes, it calls resource_read_attrs or resource_write_attrs handlers, respectively, in case of Resources, or appropriately instance_read_default_attrs, instance_write_default_attrs, object_read_default_attrs or object_write_default_attrs, in case of Object Instances or Objects. Note that the library will automatically call the “default” handlers if there are some unset attributes on the more specific level.

5.2.1. Pre-implemented attribute storage subsystem

As the cases described above are very common and generic – and as such, usually implemented in exactly the same manner for all objects in the code base, the library includes a subsystem that implements all the attribute-related handlers in a generic and reusable way.

To use the Attribute Storage subsystem, make sure that the library is compiled with it enabled, which means that:

  • If building using CMake: the WITH_ATTR_STORAGE CMake option needs to be enabled, e.g. by specifying -DWITH_ATTR_STORAGE. Note that it is enabled by default.

  • If using an alternative build system: the ANJAY_WITH_ATTR_STORAGE macro needs to be defined in the anjay_config.h file. Note that this is true for all the sample configurations from the example_configs directory.

No additional steps are necessary - the Attribute Storage subsystem will provide implementations of attribute-related handlers in the original objects with its own implementation, unless some handlers are actually already implemented (set to anything else than NULL). For a detailed description on how does the handler replacement behave when only some of the handlers are implemented, refer to the documentation.

5.2.1.1. Persistence

To facilitate storing attribute values between executions of the program, the Attribute Storage subsystem contains a persistence code, that can be used to serialize and deserialize all the stored attributes to some kind of external memory.

To use the persistence code, you first need to include the appropriate header:

#include <anjay/attr_storage.h>

It contains declarations of the two functions that can be used to manage the persistent information:

avs_error_t anjay_attr_storage_persist(anjay_t *anjay,
                                       avs_stream_t *out_stream);

avs_error_t anjay_attr_storage_restore(anjay_t *anjay, avs_stream_t *in_stream);

The data are read or written to and from objects of the avs_stream_t type. Please refer to documentation of AVSystem Commons for information on what it is and how to create one.

For the simple case, the avs_stream_file_ family of functions may be useful.

Note

The persistence functions shall be called after registering all the LwM2M Objects in the Anjay object and fully loading the data model structure (i.e. instantiating all the Object Instances that are supposed to be instantiated). Otherwise, attributes stored for non-existent Objects or their Instances will be discarded.