Deprecations

This page describes deprecated code which has been removed in Gaffer 2 and how to migrate to better equivalents. Each heading for a section below refers to a classname from uk.gov.gchq.gaffer where there have been changes or where that class has been removed entirely. The section headings link to the code on GitHub for that class (as of the Gaffer 1.21.1 release).

Deprecations impacting the serialisers used in schemas are listed first, followed by changes to Seed Matching and changes to Traits. Other deprecations are then listed in alphabetical order.

Serialisers

Migrating away from deprecated Serialisers

Various deprecated serialisers have been removed completely (details below). If any of these are being used in an existing schema, a new graph and schema will need to be created (see below for replacement serialisers to use) and data from existing graphs migrated. Data will need to be migrated (export and reimport) from graphs using deprecated serialisers before upgrading to Gaffer v2.

It is essential to migrate data stored using deprecated serialisers. Simply replacing these serialisers is not enough because this will prevent existing data from being read and potentially put the backing store into a corrupted state.

Preservation of ordering

When using an ordered store (such as Accumulo), all serialisers used on vertices must preserve order. As such, compactRaw serialisers (which do not preserve order) cannot be used on vertices in ordered stores.

However, when preserving order is not required, such as for properties, CompactRaw serialisers are the most effective solution and should always be used. Using an ordered serialiser on a property would reduce performance without providing any benefit. See the schemas documentation for more detail.

Removed Serialisers

serialisation.implementation.raw.RawDateSerialiser and serialisation.DateSerialiser

Use uk.gov.gchq.gaffer.serialisation.implementation.ordered.OrderedDateSerialiser instead - note that this will preserve order. Neither of these replacement serialisers implement .deserialiseString(String), instead use new Date(Long.parseLong(String)) in place of this.

serialisation.implementation.raw.RawDoubleSerialiser and serialisation.DoubleSerialiser

Use uk.gov.gchq.gaffer.serialisation.implementation.ordered.OrderedDoubleSerialiser instead - note that this will preserve order. Neither of these replacement serialisers implement .deserialiseString(String), instead use Double.parseDouble(String) in place of this.

serialisation.implementation.raw.RawFloatSerialiser and serialisation.FloatSerialiser

Use uk.gov.gchq.gaffer.serialisation.implementation.ordered.OrderedFloatSerialiser instead - note that this will preserve order.

serialisation.IntegerSerialiser and serialisation.implementation.raw.RawIntegerSerialiser

Use uk.gov.gchq.gaffer.serialisation.implementation.ordered.OrderedIntegerSerialiser instead if you need order preserved (e.g. vertex types). If object ordering definitely does not need to be preserved (e.g. only property types), uk.gov.gchq.gaffer.serialisation.implementation.raw.CompactRawIntegerSerialiser should be used instead. Neither of these replacement serialisers implement .deserialiseString(String), instead use Integer.parseInt(String) in place of this.

serialisation.LongSerialiser and serialisation.implementation.raw.RawLongSerialiser

Use uk.gov.gchq.gaffer.serialisation.implementation.ordered.OrderedLongSerialiser instead if you need order preserved (e.g. vertex types). If object ordering definitely does not need to be preserved (e.g. only property types), uk.gov.gchq.gaffer.serialisation.implementation.raw.CompactRawLongSerialiser could also be used instead.

Changed Serialisers

serialisation.ToBytesSerialiser and serialisation.ToBytesViaStringDeserialiser

In both serialisers, the method deserialise(byte[]) has been marked as deprecated. It cannot be deleted as it is needed to implement the Serialiser interface. It is recommended for speed/performance to use the other implementation with an offset and a length: deserialise(byte[], int, int).

Removal of Seed Matching

operation.SeedMatching

SeedMatching has been removed from Gaffer. This was previously used in get operations, like GetElements, to select whether you wanted your results to contain only Elements that are the same type as the seed, or both Edges and Entities. For more info, see the Gaffer 1.X docs page on SeedMatching. As described in the Gaffer 1.X docs, SeedMatching can be replaced with a View. The default behaviour in Gaffer is the same as if you used seed_matching="RELATED", so if this is the case, there is no migration required. However, if you used seed_matching="EQUAL", you will need to migrate to a View.

SeedMatching migration with EdgeSeeds

Where SeedMatching was used to only get back Edges from EdgeSeeds

JavaJSONPython

final GetElements getEdgesWithSeedMatching = new GetElements.Builder()
        .input(new EdgeSeed("source", "dest", true))
        .seedMatching(SeedMatching.SeedMatchingType.EQUAL)
        .build();

{
    "class" : "GetElements",
    "input" : [ {
        "class" : "EdgeSeed",
        "source" : "source",
        "destination" : "dest",
        "matchedVertex" : "SOURCE",
        "directedType" : "DIRECTED"
    } ],
    "seedMatching" : "EQUAL"
}

g.GetElements(
    input=[
        g.EdgeSeed(
            source="source",
            destination="dest",
            directed_type="DIRECTED",
            matched_vertex="SOURCE"
        )
    ],
    seed_matching="EQUAL"
)

You should instead specify that in a View

JavaJSONPython

final GetElements getEdgesWithoutSeedMatching = new GetElements.Builder()
        .input(new EdgeSeed("source", "dest", true))
        .view(new View.Builder()
                .edge("relevantEdgeGroup")
                .build())
        .build();

{
    "class" : "GetElements",
    "input" : [ {
        "class" : "EdgeSeed",
        "source" : "source",
        "destination" : "dest",
        "matchedVertex" : "SOURCE",
        "directedType" : "DIRECTED"
    } ],
    "view" : {
        "edges" : {
            "relevantEdgeGroup" : { }
        }
    }
}

g.GetElements(
    input=[
        g.EdgeSeed(
            source="source",
            destination="dest",
            directed_type="DIRECTED",
            matched_vertex="SOURCE"
        )
    ],
    view=g.View(
        edges=[g.ElementDefinition(group="relevantEdgeGroup")]
    )
)

SeedMatching migration with EntitySeeds

Where SeedMatching was used to only get back Entities from EntitySeeds

JavaJSONPython

final GetElements getEntitiesWithSeedMatching = new GetElements.Builder()
        .input(new EntitySeed("vertex"))
        .seedMatching(SeedMatching.SeedMatchingType.EQUAL)
        .build();

{
    "class" : "GetElements",
    "input" : [ {
        "class" : "EntitySeed",
        "vertex" : "vertex"
    } ],
    "seedMatching" : "EQUAL"
}

g.GetElements(
    input=[
        g.EntitySeed(
            vertex="vertex"
        )
    ],
    seed_matching="EQUAL"
)

You should instead specify that in a View

JavaJSONPython

final GetElements getEntitiesWithoutSeedMatching = new GetElements.Builder()
        .input(new EntitySeed("vertex"))
        .view(new View.Builder()
                .entity("relevantEntityGroup")
                .build())
        .build();

{
    "class" : "GetElements",
    "input" : [ {
        "class" : "EntitySeed",
        "vertex" : "vertex"
    } ],
    "view" : {
        "entities" : {
            "relevantEntityGroup" : { }
        }
    }
}

g.GetElements(
    input=[
        g.EntitySeed(
            vertex="vertex"
        )
    ],
    view=g.View(
        entities=[g.ElementDefinition(group="relevantEntityGroup")]
    )
)

SeedMatching migration with EdgeSeeds and EntitySeeds

Where SeedMatching was used to only get back only Edges from the provided EdgeSeeds and only Entities from the provided EntitySeeds

JavaJSONPython

final GetElements getBothWithSeedMatching = new GetElements.Builder()
        .input(new EntitySeed("vertex"), new EdgeSeed("source", "dest", true))
        .seedMatching(SeedMatching.SeedMatchingType.EQUAL)
        .build();

{
    "class" : "GetElements",
    "input" : [ {
        "class" : "EntitySeed",
        "vertex" : "vertex"
    }, {
        "class" : "EdgeSeed",
        "source" : "source",
        "destination" : "dest",
        "matchedVertex" : "SOURCE",
        "directedType" : "DIRECTED"
    } ],
    "seedMatching" : "EQUAL"
}

g.GetElements(
    input=[
        g.EntitySeed(
            vertex="vertex"
        ),
        g.EdgeSeed(
            source="source",
            destination="dest",
            directed_type="DIRECTED",
            matched_vertex="SOURCE"
        )
    ],
    seed_matching="EQUAL"
)

You will instead need to perform multiple Operations and combine the results. To perform the above operation, you would have to combine both previous examples.

Changes to Store Traits

store.Store

• The method getTraits() has been removed. Use Store.execute(Operation, Context) with the GetTraits operation instead.
• The method hasTrait(StoreTrait) has been removed. Use Store.execute(Operation, Context) with the HasTrait operation instead.

federatedstore.FederatedGraphStorage

• The method getTraits(GetTraits, Context) has been removed. Use Store.execute(Operation, Context) with the GetTraits operation instead.

federatedstore.FederatedStore

• The methods getTraits() and getTraits(GetTraits, Context) have been removed. Use Store.execute(Operation, Context) with the GetTraits operation instead.

Changes to Schemas

Deprecated methods in store.schema.TypeDefinition and store.schema.Schema have been removed (see below sections).

Elements Schema

Specifying a property name to use as a time stamp is now set under config.

Previously set at the top level:

{
  "entities": {
    ...
  },
  "edges": {
    ...
  },
  "timestampProperty": "timestamp"
}

Now set in config:

{
  "entities": {
    ...
  },
  "edges": {
    ...
  },
  "config": {
    "timestampProperty": "timestamp"
  }
}

Types Schema

Specifying a serialiser class using deprecated serialiserClass is no longer possible. Instead, use class in serialiser. Deprecated vertexSerialiserClass has also been removed.

Old, deprecated, and now removed approach:

{
  "types": {
    "example.map": {
      "description": "Map type description",
      "class": "java.util.LinkedHashMap",
      "serialiserClass": "uk.gov.gchq.gaffer.serialisation.implementation.MapSerialiser"
    }
  }
}

Now set in serialiser:

{
  "types": {
    "example.map": {
      "description": "Map type description",
      "class": "java.util.LinkedHashMap",
      "serialiser": {
        "class": "uk.gov.gchq.gaffer.serialisation.implementation.MapSerialiser"
      }
    }
  }
}

All other Deprecations

accumulostore.AccumuloProperties

• The TABLE setting/variable plus the methods getTable() and setTable(String) have been removed. For getTable(), uk.gov.gchq.gaffer.accumulostore.getTableName() could be used instead.
• A graphId should be supplied instead of setting TABLE directly.

accumulostore.MockAccumuloStore

• This class has been removed.
• For in memory graphs, use uk.gov.gchq.gaffer.mapstore.MapStore instead.
• For Accumulo specific store tests, use uk.gov.gchq.gaffer.accumulostore.MiniAccumuloStore instead.

commonutil.TestTypes

• This class has been removed.
• Use the equivalent TestTypes class in the store module uk.gov.gchq.gaffer.store.TestTypes instead.

commonutil.CommonConstants

• This class has been removed as it was redundant.
• For the UTF-8 constant use StandardCharsets.UTF_8.name() from built in Java libraries.
• The above also applies to the ISO_8859_1 constant from this class.
• This also allows for more robust error handing, this commit is an example of a change implementing this.

data.elementdefinition.view.NamedViewDetail

• The method hasWriteAccess(String userId, Set<String> opAuths, String adminAuth) has been removed.
• Use hasWriteAccess(User user, String adminAuth) instead.

data.elementdefinition.view.ViewElementDefinition

• The method setAggregator(ElementAggregator aggregator) has been removed.
• A ViewElementDefinition should be constructed using the builder uk.gov.gchq.gaffer.data.elementdefinition.view.ViewElementDefinition.Builder instead.

Example

Where setAggregator was used previously

final ViewElementDefinition elementDef = new ViewElementDefinition();
elementDef.setAggregator(myElementAggregator);

You should now use the Builder

final ViewElementDefinition elementDef = new ViewElementDefinition.Builder()
        .aggregator(myElementAggregator)
        .build();

federatedstore.FederatedAccess

• The method isAddingUser(User) has been removed.
• Use hasReadAccess(User user, String adminAuth)/hasWriteAccess(User user, String adminAuth) instead.

federatedstore.FederatedGraphStorage

• The methods getAllIdsAsAdmin(), getAllGraphAndAccessAsAdmin(List<String>) and changeGraphAccessAsAdmin(String, FederatedAccess) have all been removed.
• The method remove(String graphId) has been removed. The following can be used instead:
  • remove(String graphId, User user)
  • remove(String graphId, User user, String adminAuth)
  • remove(String graphId, Predicate<Entry<FederatedAccess, Set<Graph>>> entryPredicateForGraphRemoval)

federatedstore.FederatedStore

• The method updateOperationForGraph(Operation, Graph) has been removed. Use FederatedStoreUtil.updateOperationForGraph(Operation, Graph) instead.
• The method addGraphs(Set<String> graphAuths, String addingUserId, GraphSerialisable... graphs) has been removed. The following can be used instead:
  • addGraphs(Set<String> graphAuths, String addingUserId, boolean isPublic, GraphSerialisable... graphs)
  • addGraphs(Set<String> graphAuths, String addingUserId, boolean isPublic, boolean disabledByDefault, GraphSerialisable... graphs)
  • addGraphs(Set<String> graphAuths, String addingUserId, boolean isPublic, boolean disabledByDefault, AccessPredicate readAccessPredicate, AccessPredicate writeAccessPredicate, GraphSerialisable... graphs)
  • addGraphs(FederatedAccess access, GraphSerialisable... graphs)

federatedstore.operation.RemoveGraph

• The method Builder.setGraphId(String graphId) has been removed.
• Use Builder.graphId(String graphId) which has identical behaviour instead.

Example

Where Builder.setGraphId was used previously

final RemoveGraph removeGraphOp = new RemoveGraph.Builder()
        .setGraphId("myGraph")
        .build();

You should now use Builder.graphId

final RemoveGraph removeGraphOp = new RemoveGraph.Builder()
        .graphId("myGraph")
        .build();

graph.Graph

• The methods Builder.graphId, Builder.library, Builder.view, Builder.addHook, Builder.addHooks have all been removed in all forms.
• Instead of using these methods, use .config() to set the graphConfig.

Example

Where the graph config was added using the Graph.Builder before

final Graph myGraph = new Graph.Builder()
        .graphId("myGraph")
        .library(myGraphLibrary)
        .view(myView)
        .addHook(customHook)
        .addSchema(mySchema)
        .storeProperties(storeProperties)
        .build();

You should now use the GraphConfig.Builder

final Graph myGraph = new Graph.Builder()
        .config(new GraphConfig.Builder()
                .graphId("myGraph")
                .library(myGraphLibrary)
                .view(myView)
                .addHook(customHook)
                .build())
        .addSchema(mySchema) // (1)!
        .storeProperties(storeProperties) // (2)!
        .build();

1. Schemas are not part of the GraphConfig
2. StoreProperties are not part of the GraphConfig

hdfs.operation.MapReduce

• The methods getNumReduceTasks() and setNumReduceTasks(Integer) have been removed.
• Gaffer’s operations that inherit MapReduce did not make use of numReduceTasks, either setting it to a constant number in the JobFactory or using Accumulo to automatically set the number (recommended for performance) and using min/max to keep it within a range. Therefore, numReduceTasks, getNumReduceTasks and setNumReduceTasks have been removed from this interface.

hdfs.operation.AddElementsFromHdfs

• The methods getNumReduceTasks() and setNumReduceTasks(Integer) have been removed.
• The number of reduce tasks should not be set. By default the number of reduce tasks should match the number of tablets. Use minimum and maximum reduce tasks to specify boundaries for the number of reduce tasks.

hdfs.operation.SampleDataForSplitPoints

• The methods getNumReduceTasks() and setNumReduceTasks(Integer) have been removed.
• These methods were not required as NumReduceTasks was always set to 1 in any case.

jobtracker.JobDetail

• The constructors which took userId as a String have been removed.
• Instead, a User (uk.gov.gchq.gaffer.user.User) should be used in its place. See the Builder for User.
• getUserId and setUserId have also been removed. For getting the UserId, getUser().getUserId() can be used instead. See the Javadoc for User.

jsonserialisation.JSONSerialiser

• The method update(String jsonSerialiserClass, String jsonSerialiserModules) has been removed.
• Use update(String jsonSerialiserClass, String jsonSerialiserModules, Boolean strictJson) instead. Passing strictJson as null will result in the same behaviour.

operation.Operation

• The method asOperationChain(Operation operation) has been removed.
• Use OperationChain.wrap with the Operation instead.

operation.impl.GetWalks

• The method Builder.operation has been removed.
• Use the vararg method Builder.addOperations instead.

operation.impl.SplitStore

• This class has been removed.
• It is replaced by SplitStoreFromFile which is identical except in name.

operation.impl.join.methods.JoinFunction

• The method join(Iterable keys, String keyName, String matchingValuesName, Match match, Boolean flatten) which was not implemented has been removed.

rest.SystemProperty

• GRAPH_ID, GRAPH_HOOKS_PATH, GRAPH_LIBRARY_PATH and GRAPH_LIBRARY_CONFIG have been removed.
• These config options have been removed in favour of providing a graphConfig JSON and using GRAPH_CONFIG_PATH instead.

rest.service.v2.example.ExamplesFactory

• This class has been removed.
• It is replaced by uk.gov.gchq.gaffer.rest.factory.ExamplesFactory, which can be used instead.

store.StoreProperties

• StoreProperties ID (gaffer.store.id) and related methods (getId(), setId(String)) have been removed.
• The ID of the store properties is instead directly set in the GraphLibrary when adding the StoreProperties with GraphLibrary.add(String graphId, String schemaId, Schema schema, String propertiesId, StoreProperties properties).
• See the Javadoc for GraphLibrary for more detail.
• If you aren't using a GraphLibrary, this change shouldn't affect you as store properties ID is only used in GraphLibrary.

store.Context

• The private constructor Context(User user, Map<String, Object> config, String jobId) has been removed; along with the jobId(String) method.
• Use Context(User user, Map<String, Object> config) instead. This does not support supplying the Job ID, this will be set automatically. To get the Job ID use .getJobId().

store.schema.TypeDefinition

• The method getSerialiserClass() has been removed. Instead, use getSerialiser() with .getClass() and related methods.
• The method setSerialiserClass(String) has been removed. Instead, set the Serialiser directly using setSerialiser(Serialiser).

store.schema.Schema

• Schema ID (gaffer.store.id) and related methods have been removed. The ID is now defined in GraphLibrary when adding the schema.
• timestampProperty and related methods have been removed. Instead, this is specified by setting "config": {"timestampProperty": "timestamp"} (where "timestamp" is the property name to use as a time stamp) in the Schema. See this example schema for more info.
• The method getVertexSerialiserClass() has been removed. It can be replaced by calling vertexSerialiser.getClass() and converting the result as appropriate, e.g. getVertexSerialiserClass() used SimpleClassNameIdResolver.getSimpleClassName(vertexSerialiser.getClass()).

store.library.GraphLibrary

• The method addSchema(Schema schema) has been removed. Use addSchema(String id, Schema schema) instead.
• The method addProperties(StoreProperties properties) has been removed. Use addProperties(String id, StoreProperties properties) instead.
• Both of these now require the ID to be supplied.

store.operation.OperationChainValidator

• The method validateViews(Operation op, ValidationResult validationResult, Schema schemaNotUsed, Store store) has been removed. Use validateViews(Operation op, User user, Store store, ValidationResult validationResult) instead, passing user as null will result in the same behaviour.
• The method validateComparables(Operation op, ValidationResult validationResult, Schema schemaNotUsed, Store store) has been removed. Use validateComparables(Operation op, User user, Store store, ValidationResult validationResult) instead, passing user as null will result in the same behaviour.

store.operation.handler.named.cache.NamedViewCache

• The method deleteNamedView(String name) has been removed. Use deleteNamedView(String name, User user) instead, passing user as null will result in the same behaviour.
• The method getNamedView(String name) has been removed. Use getNamedView(String name, User user) instead.
• The method getAllNamedViews() has been removed. Use getAllNamedViews(User user) instead.

types.IntegerFreqMap

• This class has been removed.
• Use uk.gov.gchq.gaffer.types.FreqMap instead, this is identical except for using Long rather than Integer.

types.function.IntegerFreqMapAggregator

• This class has been removed.
• Use uk.gov.gchq.gaffer.types.function.FreqMapAggregator instead.

serialisation.IntegerFreqMapSerialiser

• This class has been removed.
• Use uk.gov.gchq.gaffer.serialisation.FreqMapSerialiser instead.