-
#include "tensorstore/spec.h"
- class tensorstore::Spec;
Specifies the parameters necessary to open or create a
TensorStore
.Includes the driver identifier, driver parameters, optionally an
IndexTransform
andSchema
.Constructors¶
- Spec();
Constructs an invalid specification.
Methods¶
- bool valid() const;
Returns
true
if this is a valid spec.
- DimensionIndex rank() const;
Returns the rank of the TensorStore, or
dynamic_rank
if unknown.
- Result<Schema> schema() const;
Returns the effective schema, which includes all known information, propagated to the transformed space.
- Result<IndexDomain<>> domain() const;
Returns the effective domain, based on the schema constraints as well as any driver-specific constraints. If the domain cannot be determined, returns a null index domain.
- Result<ChunkLayout> chunk_layout() const;
Returns the effective chunk layout, which includes all known information, propagated to the transformed space.
- Result<SharedArray<const void>> fill_value() const;
Returns the effective fill value, propagated to the transformed space.
- Result<DimensionUnitsVector> dimension_units() const;
Returns the effective dimension units, propagated to the transformed space.
- kvstore::Spec kvstore() const;
Returns the associated key-value store used as the underlying storage. If unspecified or not applicable, returns a null (invalid) spec.
- const IndexTransform<>& transform() const;
Returns the transform applied on top of the driver.
- absl::Status Set(Option&&... option);
- absl::Status Set(SpecConvertOptions&& options);
Applies the specified options in place.
- absl::Status BindContext(const Context& context);
Binds any unbound context resources using the specified context. Any already-bound context resources remain unmodified.
- void UnbindContext();
Unbinds any bound context resources, replacing them with context resource specs that may be used to recreate the context resources. Any already-unbound context resources remain unmodified.
- void StripContext();
Replaces any context resources with a default context resource spec.
- ContextBindingState context_binding_state() const;
Indicates the context binding state of the spec.
- Result<IndexTransform<>> GetTransformForIndexingOperation() const;
Returns a transform that may be used for apply a DimExpression.
Friend functions¶
- friend auto ApplyIndexTransform(Expr&& expr, Spec spec);
Applies a function that operates on an IndexTransform to a Spec.
-
friend Result<Spec>
ApplyIndexTransform(IndexTransform<> transform, Spec spec); Applies an index transform to a Spec.
- friend bool operator==(const Spec& a, const Spec& b);
- friend bool operator!=(const Spec& a, const Spec& b);
Compares for equality via JSON representation, except that bound context resources are compared by identity.
Related Types¶
- enum class tensorstore::OpenMode;
Specifies the mode to use when opening a
TensorStore
.
- class tensorstore::MinimalSpec;
Indicates a minimal spec, i.e. missing information necessary to recreate.
- struct tensorstore::RecheckCacheOption;
Specifies time bound on cached data that may be used without revalidation.
- struct tensorstore::RecheckCachedData : public RecheckCacheOption;
Specifies time bound on cached array data (as opposed to metadata).
- struct tensorstore::RecheckCachedMetadata : public RecheckCacheOption;
Specifies time bound on cached metadata (as opposed to actual array data).
- struct tensorstore::RecheckCached : public RecheckCacheOption;
Specifies the same time bound for both cached array data and metadata.
- struct tensorstore::SpecOptions : public Schema;
Options for mutating
Spec
objects.
- struct tensorstore::SpecRequestOptions : public SpecOptions;
Options for requesting a
Spec
from an openTensorStore
.
- struct tensorstore::SpecConvertOptions : public SpecRequestOptions;
Options for converting an existing
Spec
.
- struct tensorstore::OpenOptions : public SpecOptions;
Options for opening a
Spec
.
- struct tensorstore::TransactionalOpenOptions : public OpenOptions;
Options for opening a
Spec
with optional transaction and optional batch.
Related Functions¶
-
Result<Spec>
tensorstore::Cast(const Spec& base_spec, DataType target_dtype); Returns a view of a
Spec
with a converted data type.