The Descartes Labs Platform is designed to answer some of the world’s most complex and pressing geospatial analytics questions. Our customers use the platform to build algorithms and models that transform their businesses quickly, efficiently, and cost-effectively.

By giving data scientists and their line-of-business colleagues the best geospatial data and modeling tools in one package, we help turn AI into a core competency.

Data science teams can use our scaling infrastructure to design models faster than ever, using our massive data archive or their own.

Please visit https://descarteslabs.com for more information about the Descartes Labs Platform and to request access.

The descarteslabs python package, available at https://pypi.org/project/descarteslabs/, provides client-side access to the Descartes Labs Platform for our customers. You must be a registered customer with access to our Descartes Labs Platform before you can make use of this package with our platform.

The documentation for the latest release can be found at https://docs.descarteslabs.com. For any issues please request Customer Support at https://support.descarteslabs.com.

Bugfix only

• The descarteslabs client CLI script generated by the installation process was broken. Now it works!

A very minor release with some obscure bug fixes.

• The descarteslabs client CLI has had an overhaul. Gone is the obsolete support for the Raster client, and added is support for querying Catalog Products, Bands, and Blobs and managing sharing for the same.
• Minor fixes to the authorization flow on login.

• Add testing of Blobs.

• Corrected regular expressions used to parse the memory argument to the Function constructor.
• Improved documentation of the the cpus and memory arguments to the Function constructor.

• Fixed a bug in seldom-used code to clear client state causing an import failure.

• Fixed a bug in Table.visualize() which could cause Authorization (401) errors when rendering tiles into an ipyleaflet.Map.

• Bumped some minimum dependency version constraints to avoid security vulnerabilities.

• Fixed a bug in Table.visualize() that was causing it to fail.

Due to a number of breaking changes, the version has been bumped to 3.0.0. However, the vast majority of typical use patterns in typical user code will not require changes. Please review the specifics below.

• The tags attributes on Catalog objects can now contain up to 32 elements, each up to 1000 characters long. But why would you even want to go there?
• Breaking Change: Derived bands, never supported in the AWS environment and catalog products, have been removed.
• The new Blob.delete_many method may be used to delete large numbers of blobs efficiently.
• The Blob.get_or_create method didn't allow supplying storage_type, namespace, or name parameters. Now it works as expected, either returning a saved Blob from the Catalog, or an unsaved blob that you can use to upload and save its data.
• Image methods ndarray and download no longer pass the image's default geocontext geometry as a cutline. This is to avoid problems when trying to raster a complete single image in its native CRS and resolution where imperfect geometries (due to a simplistic projection to EPSG:4326) can cause some boundary pixels to be masked. When passing in an explicit GeoContext to these methods, consider whether any cutline geometry is required or not, to avoid these issues.

• Function and Job objects now have a new environment attribute which can be used to define environment variables for the jobs when they are run.
• Breaking Change: The Function.map method previously had no bound on how many jobs could be created at one time. This led to operational problems with very large numbers of jobs. Now it submits jobs in batches (up to 1000 jobs per batch) to avoid request timeouts, and is more robust on retryable errors so that duplicate jobs are not submitted accidently. There is still no bound on how many jobs you may create with a single call to Function.map. Additionally, since it is possible that some jobs may be successfully submitted, and others not, the return value, while still behaving as a list of Jobs, is now a JobBulkCreateResult object which has a is_success and an error property which can be used to determine if all submissions were successful, what errors may have occurred, and what jobs have actually been created. Only if the first batch fails hard will the method raise an exception.
• The Job.statistics member is now typed as a JobStatistics object.
• The efficiency of deleting many jobs at once has been significantly improved using Function.delete and Function.delete_jobs. It is still possible to encounter request timeouts with very large numbers of jobs; workarounds are now documented in the API documentation for the Function.delete_jobs method.
• The ComputeClient.check_credentials method has been added, so that the client can determine if valid user credentials have already been registered with the Compute service.

• The Vector client library, previously available as the descarteslabs-vector package on PyPI, has now been integrated into the Descartes Labs Python Client (this package). It should no longer be installed separately.
• Visualization support (ipyleaflet.Map) is enabled when ipyleaflet is available. It is not installed by default, but can be installed manually, or by installing the descarteslabs python client with the viz extra (e.g. pip install descarteslabs[viz]). Note that in order to be compatible with jupyterlab notebooks, the visualize() method no longer returns the layer, it just adds it to the supplied map.
• The Vector package now has a VectorClient API client, with the usual support for get_default_client() and set_default_client(). Most constructors and methods now accept an optional client= parameter if you need to use something other than the default client.
• Configuration is now accomplished using the standard descarteslabs.config package. In particular, the vector_url setting is used to specify the default Vector host. The VECTOR_API_HOST environment variable is no longer consulted.
• Vector client methods now raise standard descarteslabs.exceptions Exception classes rather than the descarteslabs.vector.vector_exceptions classes of the old client.
• The is_spatial= parameter previously accepted by many methods and functions is now deprecated and ignored. It is not required because existing type information always determines if an operation is spatial or not. Warnings will be generated if it is used.
• Be advised that feature upload and download (query) do not currently support or impose any limits, and thus allow operations so large and slow that timeouts or other failures may occur. A future version will implement limits and batching, so that large operations can be supported reliably. Until then, the user may wish to implement their own batching were possible to avoid encountering network limits and timeouts.

• The old client version v1.12.1 is reaching end of life and will longer be supported as of February 2024. You can expect the version to stop working at any point after that as legacy backend support is turned off.
• Breaking Change: The deprecated Scenes client API has been removed.
• Breaking Change: The deprecated Metadata client API has been removed.
• The minimum required version of urllib3 has been bumped to 1.26.18 to address a security vulnerability.
• The minimum required version of shapely has been bumped to 2.0.0 to address thread safety issues.
• Python 3.7, formerly deprecated, is no longer supported.
• Python 3.12 is not yet officially supported due to the lack of support from blosc. However, if you are able to provide a functional blosc on your own, then 3.12 should work.
• Urllib3 2.X is now supported.
• Geopandas, Pydantic, and PyArrow have been added as core dependencies to support the Vector client.
• For those users of the clear_client_state function (not common), the bands cache for the Catalog client is now cleared also.

• Function.delete_jobs was failing to implement the delete_results parameter, so job result blobs were not being deleted. This has been fixed.
• Add delete_results parameter to Function.delete for consistency.
• Job.statistics field added which contains statistics (CPU, memory, and network utilization) for the job. This can be used to determine the minimal resources necessary for the Function after some representative runs.

• Filtering on datetime attributes (such as Function.created_at) didn't previously work with anything but datetime instances. Now it also handles iso format strings and unix timestamps (int or float).

• Following our lifecycle policy, client versions v1.11.0 and earlier are no longer supported. They may cease to work with the Platform at any time.

• The Catalog Blob class now has a get_data() method which can be used to retrieve the blob data directly given the id, without having to first retrieve the Blob metadata.

• Breaking Change The status values for Function and Job objects have changed, to provide a better experience managing the flow of jobs. Please see the updated Compute guide for a full explanation. Because of the required changes to the back end, older clients (i.e. v2.0.3) are supported in a best effort manner. Upgrading to this new client release is strongly advised for all users of the Compute service.

• Breaking Change The base images for Compute have been put on a diet. They are now themselves built from "slim" Python images, and they no longer include the wide variety of extra Python packages that were formerly included (e.g. TensorFlow, SciKit Learn, PyTorch). This has reduced the base image size by an order of magnitude, making function build times and job startup overhead commensurately faster. Any functions which require such additional packages can add them in as needed via the requirements= parameter. While doing so will increase image size, it will generally still be much smaller and faster than the prior "Everything and the kitchen sink" approach. Existing Functions with older images will continue to work as always, but any newly minted `Function`` using the new client will be using one of the new slim images.

• Base images are now available for Python3.10 and Python3.11, in addition to Python3.8 and Python3.9.

• Job results and logs are now integrated with Catalog Storage, so that results and logs can be searched and retrieved directly using the Catalog client as well as using the methods in the Compute client. Results are organized under storage_type=StorageType.COMPUTE, while logs are organized under storage_type=StorageType.LOGS.

• The new ComputeResult class can be used to wrap results from a Function, allowing the user to specify additional attributes for the result which will be stored in the Catalog Blob metadata for the result. This allows the function to specify properties such as geometry, description, expires, extra_attributes, writers and readers for the result Blob. The use of ComputeResult is not required.

• A Job can now be assigned arbitrary tags (strings), and searched based on them.

• A Job can now be retried on errors, and jobs track error reasons, exit codes, and execution counts.

• Function and Job objects can now be filtered by class attributes (ex. Job.search().filter(Job.status == JobStatus.PENDING).collect()).

• The Job.cancel() method can now be used to cancel the execution of a job which is currently pending or running. Pending jobs will immediately transition to JobStatus.CANCELED status, while running jobs will pass through JobStatus.CANCEL (waiting for the cancelation to be signaled to the execution engine), JobStatus.CANCELING (waiting for the execution to terminate), and JobStatus.CANCELED (once the job is no longer executing). Cancelation of running jobs is not guaranteed; a job may terminate successfully, or with a failure or timeout, before it can be canceled.

• The Job.result() method will raise an exception if the job does not have a status of JobStatus.SUCCESS. If Job.result() yields an None value, this means that there was no result (i.e. the execution returned a None).

• The Job.result_blob() method will return the Catalog Storage Blob holding the result, if any.

• The Job.delete() method will delete any job logs, but will not delete the job result unless the delete_results parameter is supplied.

• The Function object now has attributes namespace and owner.

• The Function.wait_for_completion() and new Function.as_completed() methods provide a richer set of functionality for waiting on and handling job completion.

• The Function.build_log() method now returns the log contents as a string, rather than printing the log contents.

• The Job.log() method now returns the log contents as a list of strings, rather than printing the log contents. Because logs can be unbounded in size, there's also a new Job.iter_log() method which returns an iterator over the log lines.

• The requirements= parameter to Function objects now supports more pip magic, allowing the use of special pip controls such as -f. Also parsing of package versions has been loosened to allow some more unusual version designators.

• Changes to the Function.map() method, with the parameter name change of iterargs changed to kwargs (the old name is still honored but deprecated), corrected documentation, and enhancements to support more general iterators and mappings, allowing for a more functional programming style.

• The compute package was restructured to make all the useful and relevant classes available at the top level.

• Property filters can now be deserialized as well as serialized.

• Allow deletion of Function objects.
  • Deleting a Function will deleted all associated Jobs.
• Allow deletion of Job objects.
  • Deleting a Job will delete all associated resources (logs, results, etc).
• Added attribute filter to Function and Job objects.
  • Attributes marked filterable=True can be used to filter objects on the compute backend api.
  • Minor optimization to Job.iter_results which now uses backend filters to load successful jobs.
• Function bundling has been enhanced.
  • New include_modules and include_data parameters allow for multiple other modules, non-code data files, etc to be added to the code bundle.
  • The requirements parameter has been improved to allow a user to pass a path to their own requirements.txt file instead of a list of strings.

• Allow data type int32 in geotiff downloads.
• BlobCollection now importable from descarteslabs.catalog.

• Added API documentation for dynamic compute and vector

• Due to recent changes in urllib3, rastering operations were failing to retry certain errors which ought to be retried, causing more failures to propagate to the user than was desirable. This is now fixed.

(Release notes from all the 2.0.0 release candidates are summarized here for completeness.)

• Deprecated support for Python 3.7 (will end of life in July).
• Added support for Python 3.10 and Python 3.11
• AWS-only client. For the time being, the AWS client can be used to communicate with the legacy GCP platform (e.g. DESCARTESLABS_ENV=gcp-production) but only supports those services that are supported on AWS (catalog and scenes). This support may break at any point in the future, so it is strictly transitional.

• Removed many dependencies no longer required due to the removal of GCP-only features.
• Added support for Shapely 2.X. Note that user code may also be affected by breaking changes in Shapely 2.X. Use of Shapely 1.8 is still supported.
• Updated requirements to avoid urllib3>=2.0.0 which breaks all kinds of things.

• Major overhaul of the internals of the config process. To support other clients using namespaced packages within the descarteslabs package, the top level has been cleaned up, and most all the real code is down inside descarteslabs.core. End users should never have to import anything from descarteslabs.core. No more magic packages means that pylint will work well with code using descarteslabs.
• Configuration no longer depends upon the authorized user.

• Added support for data storage. The Blob class provides mechanism to upload, index, share, and retrieve arbitrary byte sequences (e.g. files). Blobs can be searched by namespace and name, geospatial coordinates (points, polygons, etc.), and tags. Blobs can be downloaded to a local file, or retrieved directly as a Python bytes object. Blobs support the same sharing mechanisms as Products, with owners, writers, and readers attributes.
• Added support to Property for prefix filtering.
• The default geocontext for image objects no longer specifies a resolution but rather a shape, to ensure that default rastering preserves the original data and alignment (i.e. no warping of the source image).
• As with resolution, you can now pass a crs parameter to the rastering methods (e.g. Image.ndarray, ImageCollection.stack, etc.) to override the crs of the default geocontext.
• A bug in the code handling the default context for image collections when working with a product with a CRS based on degrees rather than meters has been fixed. Resolutions should always be specified in the units used by the CRS.

• Added support for managed batch compute under the compute module.

• Fixed a bug in the handling of small blocks (less than 512 x 512) that caused rasterio to generate bad download files (the desired image block would appear as a smaller sub-block rather than filling the resulting raster).

• The defaulting of align_pixels has changed slightly for the AOI class. Previously it always defaulted to True. Now the default is True if resolution is set, False otherwise. This ensures that when specifying a shape and a bounds rather than a resolution,the shape is actually honored.
• When assigning a resolution to an AOI, any existing shape attribute is automatically unset, since the two attributes are mutually exclusive.
• The validation of bounds for a geographic CRS has been slightly modified to account for some of the irregularities of whole-globe image products, correcting unintended failures in the past.
• Fixed problem handling MultiPolygon and GeometryCollection when using Shapely 2.0.

• Loosen up the restrictions on the allowed alphabet for Blob names. Now almost any printable character is accepted save for newlines and commas.
• Added new storage types for Blobs: StorageType.COMPUTE (for Compute job results) and StorageType.DYNCOMP (for saved dynamic-compute operations).

• Added testing of the client.

• The defaulting of the namespace value for Blobs has changed slightly. If no namespace is specified, it will default to <org>:<hash> with the user's org name and unique user hash. Otherwise, any other value, as before, will be prefixed with the user's org name if it isn't already so.
• Blob.get no longer requires a full id. Alternatively, you can give it a name and optionally a namespace and a storage_type, and it will retrieve the Blob.
• Fixed a bug causing summaries of Blob searches to fail.

• Function.map and Function.rerun now save the created Jobs before returning.
• Job.get return values fixed, and removed an extraneous debug print.

• Updated requirements to avoid urllib3>=2.0.0 which break all kinds of things.

• The defaulting of align_pixels has changed slightly for the AOI class. Previously it always defaulted to True. Now the default is True if resolution is set, False otherwise. This ensures that when specifying a shape and a bounds rather than a resolution,the shape is actually honored.
• When assigning a resolution to an AOI, any existing shape attribute is automatically unset, since the two attributes are mutually exclusive.
• The validation of bounds for a geographic CRS has been slightly modified to account for some of the irregularities of whole-globe image products, correcting unintended failures in the past.

• The default geocontext for image objects no longer specifies a resolution but rather a shape, to ensure that default rastering preserves the original data and alignment (i.e. no warping of the source image).
• The Blob.upload and Blob.upload_data methods now return self, so they can be used in a fluent style.
• As with resolution, you can now pass a crs parameter to the rastering methods (e.g. Image.ndarray, ImageCollection.stack, etc.) to override the crs of the default geocontext.

• A bevy of fixes to the client.

• Added support for data storage. The Blob class provides mechanism to upload, index, share, and retrieve arbitrary byte sequences (e.g. files). Blobs can be searched by namespace and name, geospatial coordinates (points, polygons, etc.), and tags. Blobs can be downloaded to a local file, or retrieved directly as a Python bytes object. Blobs support the same sharing mechanisms as Products, with owners, writers, and readers attributes.
• Added support to Property for prefix filtering.

• Added method to update user credentials for a Function.
• Added methods to retrieve build and job logs.

• Added support for Shapely=2.X.

• This is an internal-only release. There is as of yet no updated documentation. However, the user-facing client APIs remain fully compatible with v1.12.1.

• Added support for managed batch compute under the compute module.

• Removed the check on the Auth for configuration, since it is all AWS all the time.

• Fixed a bug in the handling of small blocks (less than 512 x 512) that caused rasterio to generate bad download files (the desired image block would appear as a smaller sub-block rather than filling the resulting raster).

• This is an internal-only release. There is as of yet no updated documentation. However, the user-facing client APIs remain fully compatible with v1.12.1.

• Deprecated support for Python 3.7 (will end of life in July).
• Added support for Python 3.10 and Python 3.11
• AWS-only client. For the time being, the AWS client can be used to communicate with the legacy GCP platform (e.g. DESCARTESLABS_ENV=gcp-production) but only supports those services that are supported on AWS (catalog and scenes). This support may break at any point in the future, so it is strictly transitional.

• Removed many dependencies no longer required due to the removal of GCP-only features.

• Major overhaul of the internals of the config process. To prepare for supporting other clients using namespaced packages within the descarteslabs package, the top level has been cleaned up, and most all the real code is down inside descarteslabs.core. However end users should never have to import anything from descarteslabs.core. No more magic packages means that pylint will work well with code using descarteslabs.
• GCP environments only support catalog and scenes. All other GCP-only features have been removed.

• A bug in the code handling the default context for image collections when working with a product with a CRS based on degrees rather than meters has been fixed. Resolutions should always be specified in the units used by the CRS.

• Fixed a bug causing descarteslabs.workflows.map.geocontext() to fail with an import error. This problem also affected the autoscaling feature of workflows map layers.

• Fixed a bug causing downloads of single-band images to fail when utilizing rasterio.

• Catalog V2 is now fully supported on the AWS platform, including user ingest.
• Catalog V2 has been enhanced to provide substantially all the functionality of the Scenes API. The Image class now includes methods such as ndarray and download. A new ImageCollection class has been added, mirroring SceneCollection. The various Search objects now support a new collect method which will return appropriate Collection types (e.g. ProductCollection, BandCollection, and of course ImageCollection). Please see the updated Catalog V2 guide and API documentation for more details.
• Previously, the internal implementation of the physical_range attribute on various band types was inconsistent with that of data_range and display_range. It has now been made consistent, which means it will either not be set, or will contain a 2-tuple of float values. It is no longer possible to explicitly set it to None.
• Access permissions for bands and images are now managed directly by the product. The readers, writers, and owners attributes have been removed from all the *Band classes as well as the Image class. Also the Product.update_related_objects_permissions and Product.get_update_permissions_status methods have been removed as these are no longer necessary or supported.
• All searches for bands (other than derived bands) and images must specify one or more product ids in the filtering. This requirement can be met by using the bands() and images() methods of a product to limit the search to that product, or through a filter(properties.product_id==...) clause on the search.
• Products have a new product_tier attribute, which can only be set or modified by privileged users.
• The Image.upload_ndarray will now accept either an ndarray or a list of ndarrays, allowing multiple files per image. The band definitions for the product must correspond to the order and properties of the multiple ndarrays.

• With the addition of the Scenes functionality to Catalog V2, you are strongly encouraged to migrate your Scenes-based code to use Catalog V2 instead. Scenes will be deprecated in a future release. Some examples of migrating from Scenes to Catalog V2 are included in the Catalog V2 guide. In the meantime the Scenes API has been completely reimplemented to use Catalog V2 under the hood. From a user perspective, existing code using the Scenes API should continue to function as normal, with the exception of a few differences around some little-used dark corners of the API.
• The Scenes search_bands now enforces the use of a non-empty products= parameter value. This was previously documented but not enforced.

• With the addition of the Scenes functionality to Catalog V2, you are strongly encouraged to migrate your Metadata-based code to use Catalog V2 instead. Metadata will be deprecated in a future release.
• As with Catalog and Scenes, one or more products must now be specified when searching for bands or images.

• The Raster client API now requires a bands= parameter for all rastering operations, such as raster, ndarray and stack. It no longer defaults to all bands defined on the product.

• An off-by-1/2-pixel problem was identified in the coordinate transforms underlying DLTile.rowcol_to_latlon and DLTile.latlon_to_rowcol. The problem has been corrected, and you can expect to see slight differences in the results of these two methods.

• All the REST client types, such as Metadata and Raster, now support get_default_client() and set_default_client() instances. This functionality was previously limited to the Catalog V2 CatalogClient. Whenever such a client is required, the client libraries use get_default_client() rather than using the default constructor. This makes it easy to comprehensively redirect the library to use a specially configured client when necessary.

• The GeoContext types that originally were part of the Scenes package are now available in the new descarteslabs.geo package, with no dependencies on Scenes. This is the preferred location from which to import these classes.

• The descarteslabs.utils package, added in the previous release for the AWS client only, now exists in the GCP client as well, and is the preferred location to pick up the DotDict and DotList classes, the display and save_image functions, and the Properties class for property filtering in Catalog V2.
• The display method now has added support for multi-image plots, see the API documentation for the figsize, nrows, ncols and layout_direction parameters.

• The property_filtering.GenericProperties class has been replaced with property_filtering.Properties, but remains for back compatibility.
• Property filters now support isnull and isnotnull operations. This can be very useful for properties which may or may not be present, e.g. properties.cloud_fraction.isnull | properties.cloud_fraction <= 0.2.

• The Config exceptions RuntimeError and KeyError were changed to ConfigError exceptions from descarteslabs.exceptions.
• Auth now retrieves its URL from the Config settings. If no valid configuration can be found, it reverts to the commercial service (https://iam.descarteslabs.com).

• Dependencies for the descarteslabs library have been updated, but remain constrained to continue to support Python 3.7.
• Numerous bug fixes.

• The extra requirement options have changed. There are four extra requirement options now, visualization, tables, complete, and tests. visualization pulls in extra requirements to support operating in a Jupyter notebook or environment, enabling interactive maps and graphical displays. It is not required for operating in a "headless" manner. tables pulls in extra requirements to support the Tables client. complete is the combination of visualization and tables. tests pulls in extra requirements for running the tests. As always, pip install 'descarteslabs[complete]' will install a fully enabled client.

• The Descartes Labs client now supports configuration to support operating in different environments. By default, the client will configure itself for standard usage against the GCP platform ("gcp-production"), except in the case of AWS Marketplace users, for whom the client will configure itself against the AWS platform ("aws-production"). Alternate environments can be configured by setting the DESCARTESLABS_ENV environment variable before starting python, or by using a prelude like

  from descarteslabs.config import Settings
  Settings.select_env("environment-name")
  

  before any other imports of any part of the descarteslabs client package.
• The new AWS Enterprise Accelerator release currently includes only Auth, Configuration and the Scenes client.

• The descarteslabs.client.auth package has moved to descarteslabs.auth. It is now imported into the original location at descarteslabs.client.auth to continue to work with existing code, but new code should use the new location.
• The descarteslabs.client.exceptions module has moved to descarteslabs.exceptions. It is now imported into the original location at descarteslabs.client.exceptions to continue to work with existing code, but new code should use the new location.

• Fixed an issue in scenes.DLTile.from_shape where there would be incomplete coverage of certain geometries. The function may now return more tiles than before.
• Added support for the new all_touched parameter to the different GeoContext types. Default behavior remains the same as always, but if you set all_touched=True this communicates to the raster service that you want the image(s) rastered using GDAL's CUTLINE_ALL_TOUCHED option which will change how source pixels are mapped to output pixels. This mode is only recommended when using an AOI which is smaller than the source imagery pixel resolution.
• The DLTile support has been fixed to avoid generating gaps when tiling regions that span a large distance north-to-south and straddle meridians which are boundaries between UTM zones. So methods such as DLTile.from_shape may return more tiles than previously, but properly covering the region.
• Added support for retrieving products and bands.
  • Methods added: get_product, get_band, get_derived_band, search_products, search_bands, search_derived_bands.
  • Disallows search without products parameter.
• Scaling support has been enhanced to understand processing levels for newer products. The Scene.scaling_parameters and SceneCollection.scaling_parameters methods now accept a processing_level argument, and this will be factored in to the determination of the default result data type and scaling for all rastering operations such as Scene.ndarray and SceneCollection.mosaic.
• If the user provides the rasterio package (which implies providing GDAL), then rasterio will be used to save any downloaded images as GeoTIFF, allowing for the use of compression. Otherwise, by default the tifffile support will be used to generate the GeoTIFF files but compression is not supported in this mode.
• As the Places client has been deprecated, so has any use of the place= parameter supported by several of the Scenes functions and methods.

• (Core users only) Added support for specifying the image index to use when creating a new Product.
• Added support for defining per-processing-level data_type, data_range, display_range and physical_range properties on processing level steps.

• Added support for filtering Assets by type and name fields.
  • Supported filter types blob, folder, namespace, sym_link, sts_model, and vector. Specifying multiple types will find assets matching any given type.
  • The name field supports the following wildcards:
    • * matches 0 or more of any character.
    • ? matches 1 of any character.
  • Find assets matching type of blob and having a display name of file name.json or file2name.txt but not filename.json:
    • Discover().list_assets("asset/namespace/org:some_org", filters="type=blob&name=file?name.*")
    • Discover().list_assets("asset/namespace/org:some_org", filters=AssetListFilter(type=AssetType.BLOB, name="file?name.*"))
  • Find assets of type blob or vector:
    • Discover().list_assets("asset/namespace/org:some_org", filters="type=blob,vector")
    • Discover().list_assets("asset/namespace/org:some_org", filters=AssetListFilter(type=[AssetType.BLOB, AssetType.VECTOR], name="file?name.*"))

• Metadata.products and Metadata.available_products now properly implement paging so that by default, a DotList containing every matching product accessible to the user is returned.

• If the user provides the rasterio package (which implies providing GDAL), then rasterio will be used to save any downloaded images as GeoTIFF, allowing for the use of compression. Otherwise, by default the tifffile support will be used to generate the GeoTIFF files but compression is not supported in this mode.

• Fixed an issue that caused a user's schema to be overwritten if they didn't provide a primary key on table creation.
• Now uses Discover backend filtering for list_tables() instead of filtering on the client to improve performance.
• list_tables() now supports filtering tables by name
  • Tables.list_tables(name="Test*.json")

• New Tasks images for this release bump the versions of several dependencies, please see the Tasks guide for detailed lists of dependencies.

• The new Workbench release bumps the versions of several dependencies.

• Added support for the new all_touched parameter to the different GeoContext types. See description above under Scenes.

• The Places client has been deprecated, and use thereof will generate a deprecation warning.
• The older Catalog V1 client has been deprecated, and use thereof will generate a deprecation warning. Please use the Catalog V2 client in its place.
• Documentation has been updated to include the `AWS Enterprise Accelerator" release.
• With Python 2 far in the rearview mirror, the depedencies on the six python package have been removed throughout the library, the distribution and all tasks images.

• Added support for Python 3.9.
• Removed support for Python 3.6 which is now officially End Of Life.

• Added support for organizational sharing. You can now share using the Organization type:
  • workflows.add_reader(Organization("some_org"))

• Added support for organizational sharing. You can now share using the Organization type:
  • asset.share(with_=Organization("some_org"), as_="Viewer")
• Allow user to list their organization's namespace.
  • Discover().list_asset("asset/namespace/org:some_org")
• Allow user to list their organization's users.
  • Discover().list_org_users()

• Added an alpha Tables client. The Tables module lets you organize, upload, and query tabular data and vector geometries. As an alpha release, we reserve the right to modify the Tables client API without any guarantees about backwards compatibility. See the Tables API and Tables Guide documentation for more details.

• Added the progress= parameter to the various rastering methods such as Scene.ndarray, Scene.download, SceneCollection.mosaic, SceneCollection.stack, SceneCollection.download and SceneCollection.download_mosaic. This can be used to enable or disable the display of progress bars.

• Support for Python 3.9 images has been added, and support for Python 3.6 images has been removed.
• Many of the add on packages have been upgraded to more recently released versions. In particular, tensorflow was updated from version 2.3 to version 2.7.
• GPU support was bumped up from CUDA 10 to CUDA 11.2

• Fixed a bug preventing retry-able errors (such as a 429) from being retried.

• Allow retrieving Attribute as a class attribute. It used to raise an exception.

• Fixed a bug preventing the user from writing JPEG files with smaller than 256x256 tiles.
• Allow specifying a NoData value for non-JPEG GeoTIFF files.
• Include band description metadata in created GeoTIFF files.
• Support scaling parameters as lists as well as tuples.
• Add caching of band metadata to drastically reduce the number of metadata queries when creating SceneCollections.
• DLTiles.from_shape was failing to handle shape objects implementing the __geo_interface__ API, most notably several of the Workflows GeoContext types. These now work as expected.
• Certain kinds of network issues could read to rastering operations raising an IncompleteRead exception. This is now correctly caught and retried within the client library.

• Users can now use descarteslabs.tasks.update_credentials() to update their task credentials in case they became outdated.

• We have introduced a hard limit of 120 as the number of outstanding Workflows compute jobs that a single user can have. This limit exists to minimize situations in which a user is unable to complete jobs in a timely manner by ensuring resources cannot be monopolized by any individual user. The API that backs the calls to compute will return a descarteslabs.client.grpc.exceptions.ResourceExhausted error if the caller has too many outstanding jobs. Prior to this release (1.9.0), these failures would be retried up to some small retry limit. With the latest client release however, the client will fail without retrying on an HTTP 429 (rate limit exceeded) error. For users with large (non-interactive) workloads who don’t mind waiting, we added a new num_retries parameter to the compute function; when specified, the client will handle any 429 errors and retry up to num_retries times.
• Workflows is currently optimized for interactive use cases. If you are submitting large numbers of long-running Workflows compute jobs with block=False, you should consider using Tasks and Scenes rather than the Workflows API.
• Removed ResourceExhausted exceptions from the list of exceptions we automatically catch and retry on for compute calls.

• Lots of improvements, additions, and clarifications in the API documentation.

• Workflows client no longer validates processing_level parameter values, as these have been enhanced to support new products and can only be validated server side.
• Catalog V2 bands now support the vendor_band_name field (known as name_vendor in Metadata/Catalog V1).
• Scenes support for masking in version 1.8.1 had some regressions which have been fixed. For this reason, version 1.8.1 has been pulled from PyPI.
• New task groups now default to a maximum_concurrency value of 5, rather than the previous 500. This avoids the common problem of deploying a task group with newly developed code, and having it scale up and turning small problems into big problems! You may still set values as large as 500.
• The Tasks client now provides an update_group() method which can be used to update many properties of an existing task group, including but not limited to name, image, minimum_concurrency, and maximum_concurrency.
• Improved testing across several sub-packages.
• Various documentation fixes.

** Version Deprecated ** Due to some regressions in the Scenes API, this version has been removed from PyPI.

• Added a new common.dltile library that performs geospatial transforms and tiling operations.
• Upgraded various dependencies: requests[security]>=2.25.1,<3,six>=1.15.0,blosc==1.10.2, mercantile>=1.1.3,Pillow>=8.1.1,protobuf>=3.14.0,<4,shapely>=1.7.1,<2,tqdm>=4.32.1,traitlets>=4.3.3,<6;python_version<'3.7',traitlets==5.0.5,<6;python_version>='3.7',markdown2>=2.4.0,<3,responses==0.12.1,freezegun==0.3.12,imagecodecs>=2020.5.30;python_version<'3.7',imagecodecs>=2021.5.20;python_version>='3.7',tifffile==2020.9.3;python_version<'3.7',tifffile==2021.4.8;python_version>='3.7'

• Added an alpha Discover client. Discover allows users to organize and share assets with other users. As an alpha release, we reserve the right to modify the Discover client API without any guarantees about backwards compatibility. See the Discover API documentation for more details.

• breaking Image (Scene) metadata now accepts and returns the bucket and directory fields as lists of strings, of a length equal to that of the files fields. This allows the file assets making up an image to live in different locations. When creating new images, a simple string can still be provided for these fields. It will automatically be converted to a list of (duplicated) strings as necessary. As most users will never interact with these fields, the change should not affect user code.

• derived_params field for Image (scene) metadata now supported for product-specific service-implemented "native derived bands" which may only be created for core products.

• Scenes now uses the client-side dltile library to make DLTiles. This improves performance when creating a large number of DLTile objects.
• Scenes DLTile from_shape now has a parameter to return tile keys only instead of full tile objects. Usage details can be found in the docs.
• Scenes DLTile now has new methods: iter_from_shape that takes the same arguments as from_shape but returns an iterator (from_shape docs), subtile that adds the ability to subdivide tiles (subtile docs), and rowcol_to_latlon and latlon_to_rowcol which converts pixel coordinates to spatial coordinates and vice versa (rowcol_to_latlon docs and latlon_to_rowcol docs).
• Scenes DLTile now has a new parameter tile_extent which is the total size of the tile in pixels including padding. Usage details can be found in the docs.
• breaking Removed the dependence on Raster for tiling. The raster_client parameter has been removed from the from_latlon, from_key, from_shape, and assignDLTile methods.
• Tiling using from_shape may return a different number of tiles compared to previous versions under certain conditions. These tiles are usually found in overlapping areas between UTM zones and should not affect the overall coverage.
• DLTile geospatial transformations are guaranteed to be within eight decimal points of the past implementation.
• DLTile errors now come from the dltile library and error messages should now be more informative.
• When specifying output bounds in a spatial reference system different from the underlying raster, a densified representation of the bounding box is used internally to ensure that the returned image fully covers the bounds. For certain methods (like mosaic) this may change the returned image dimensions, depending on the SRSs involved.
• breaking As with the Metadata v1 client changes, the bucket and directory fields of the Scene properties are now multi-valued lists.
• Scenes does not support writing GeoTiffs to file-like objects. Non-JPEG GeoTiffs are always uncompressed.

• dltiles_from_shape, dltiles_from_latlon, and dltile have been removed. It is strongly recommended to test any existing code which uses the Raster API when upgrading to this release.
• Fully masked arrays are now supported and are the default. Usage details can be found in the docs
• Added support to draw progress bar. Usage details can be found in the docs.
• The signature and return value of Raster.raster() have changed. The save= parameter has been removed as the resulting download is always saved to disk, to a file named by the outfile_basename= parameter. The method returns a tuple containing the name of the resulting file and the metadata for the retrieval, which is now an ordinary Python dictionary.
• As with Scenes, when specifying output bounds in a spatial reference system different from the underlying raster, a densified representation of the bounding box is used internally to ensure that the returned image fully covers the bounds. For certain methods (like mosaic) this may change the returned image dimensions, depending on the SRSs involved.

Internal release only. See 1.8.1 above.

• Upgraded various dependencies: blosc==1.10.2, cachetools>=3.1.1, grpcio>=1.35.0,<2, ipyleaflet>=0.13.3,<1, protobuf>=3.14.0,<4, pyarrow>=3.0.0, pytz>=2021.1
• Upgraded from using Travis to GitHub Actions for CI.

• Added support for the physical_range property on SpectralBand and MicrowaveBand.

• Workflows sharing. Support has been added to manage sharing of Workflow objects with other authorized users. The public option for publishing workflows has been removed now that Workflow.add_public_reader() provides the equivalent capability. See the Workflows Guide.
• Lots of improvements to API documentation and the Workflows Guide.

• Allow constructing Float instances from literal python integers.

Fixes a few buglets which slipped through. This release continues to use the workflows channel v0-18.

• Fixed a problem with the defaulting of the visual options when generating tile URLs, making it possible to toggle the checkerboard option on a layer and see the difference.
• Support axis=list(...) for Image.
• Corrected the results of doing arithmetic on two widgets (e.g. adding two IntSliders together should yieldanInt`).
• For single-band imagery VizOption will accept a single two-tuple for the scales= argument.

• Python 3.6 is now deprecated, and support will be removed in the next version.

• Added support to Bands for new processing levels and processing step specifications to support Landsat Collection 2.

• The new channel v0-18 utilizes a new and improved backend infrastructure. Any previously saved workflows and jobs from earlier channels are not accessible from the new infrastructure, so you will need to recreate and persist (e.g. publish) new versions using v0-18. Older releases and older channels can continue to access your originals if needed.
• wf.widgets lets you quickly explore data interactively. Add widgets anywhere in your code just like normal values, and the widgets will display automatically when you call .visualize.
• View shared Workflows and XYZs in GIS applications using WMTS. Get the URL with wf.wmts_url(), XYZ.wmts_url(), Workflow.wmts_url().
  • Create publicly-accessible tiles and WMTS endpoints with wf.XYZ(..., public=True). Anyone with the URL (which is a cryptographically random ID) can view the data, no login required. Set days_to_expiration to control how long the URL lasts.
  • wf.XYZ.list() to iterate through all XYZ objects you've created, and XYZ.delete to delete them.
  • Set default vizualization options (scales, colormap, bands, etc.) in .publish or wf.XYZ with wf.VizOption. These viz_options are used when displaying the published object in a GIS application, or with wf.flows.
• ImageCollection.visualize(): display ImageCollections on wf.map, and select the reduction operation (mean, median, mosaic, etc.) interactively
• Image.reduction() and ImageCollection.reduction() (like ic.reduction("median", axis="images")) to reduce an Image/ImageCollection with an operation provided by name
• wf.map.controls is accessible (you had to do wf.map.map.controls before)
• Access the parameters used in a Job with Job.arguments and Job.geoctx.

• Errors like In 'or': : operand type(s) all returned NotImplemented from __array_ufunc__ when using the bitwise-or operator | are resolved.
• Errors when using computed values in the wf.Datetime constructor (like wf.Datetime(wf.Int(2019) + 1)) are resolved.
• wf.Timedelta can be constructed from floats, and supports all binary operations that Python does (support for /, //, %, * added)
• In .rename_bands, prohibit renaming a band to a name that already exists in the Image/ImageCollection. Previously, this would succeed, but cause downstream errors.
• .bandinfo.get("bandname", {}) now works---before, providing {} would fail with a TypeError
• Indexing an Any object (like wf.Any({"foo": 1})["foo"]) behaves correctly
• wf.Datetimes constructed from strings containing timezone information are handled correctly

• .mask(new_mask) ignores masked pixels in new_mask. Previously, masked pixels in new_mask were considered True, not False. Note that this is opposite of NumPy's behavior.
• If you .publish an object that depends on wf.parameters or wf.widgets, it's automatically converted into a wf.Function.
• breaking .compute and .inspect no longer accept extra arguments that aren't required for the computation. If the object doesn't depend on any wf.parameters or wf.widgets, passing extra keyword arguments will raise an error. Similarly, not providing keyword arguments for all parameters the object depends on will raise an error.
• breaking The wf.XYZ interface has changed; construct an XYZ with wf.XYZ(...) instead of wf.XYZ.build(...).save()
• Set days_to_expiration on XYZ objects. After this many days, the object is deleted.
• Job metadata is deleted after 10 days; wf.Job.get(...) on a job ID more than 10 days old will fail. Note that Job results have always been deleted after 10 days; now the metadata expires as well.
• wf.Function has better support for named arguments. Now, f = wf.Function[{'x': wf.Int, 'y': wf.Str}, wf.Int] requires two arguments x and y, and they can be given positionally (f(1, "hi")), by name in any order(f(x=1, y="hi") or f(y="hi", x=1)), or both (f(1, y="hi")). wf.Function.from_callable will generate a Function with the same names as the Python function you decorate or pass in. Therefore, when using @wf.publish as a decorator, the published Function will automatically have the same argument names as your Python function.

• Python 3.8 is now supported in the client.
• As Python 3.5 has reached End Of Life, it is no longer supported by the descarteslabs client.

• Altered the behavior of Task function creation. Deprecation warnings will be issued when attempting to create a Task function for which support will be removed in the near future. It is strongly recommended to test any existing code which uses the Tasks client when upgrading to this release.
• New tasks public images for for use with Python 3.8 are available.

• .pick_bands supports proxy wf.Str objects; .unpack_bands supports wf.Str and wf.Tuple[wf.Str, ...].
• Better performance constructing a wf.Array from a List of numbers (like wf.Array(ic.sum(["pixels", "bands"])))
• No more error using @wf.publish as a decorator on a function without a docstring

No more irrelevant DeprecationWarnings when importing the descarteslabs package (#235). Deprecated functionality in the package will now show FutureWarnings instead.

• wf.map.geocontext doesn't raise an error about the CRS of the map
• wf.flows doesn't raise an error about versions from incompatible channels

• Example code has been cleaned up.

• Sharing of any Workflows object as a Workflow with version and access control. Browse through shared Workflows with the wf.flows browser widget.
• Upload images to the DL catalog from Workflows jobs. Usage details can be found in the docs.
• wf.np.median
• Job.cancel() to cancel running jobs.
• Transient failures in Jobs are automatically retried, resulting in fewer errors.
• Search widget on wf.map by default.

• Bitwise operations on imagery no longer fail
• wf.np.linspace no longer fails when being called correctly
• .median is slightly less prone to OOM errors

• Breaking: Workflows sharing: wf.publish() and wf.use() have new signatures, wf.retrieve() has been removed in favor of wf.Workflow.get() and wf.VersionedGraft.get_version() and the wf.Workflow object has been completely refactored. Detailed information is in the docs.
• Array.to_imagery now accepts KnownDict for bandinfo and properties.
• Numbers can now be constructed from Strs

• Output formats for .compute including GeoTIFF, JSON, PyArrow, and MessagePack. Usage details can be found in the docs.
• Destinations for Job results: download and email. Usage details can be found in the docs.
• Save .compute outputs to a file with the file= argument.
• Pixel value inspector: click in the map widget to view pixel values.
• wf.ifelse for simple conditional logic.
• NumPy functions including hypot, bitwise_and, bitwise_or, bitwise_xor, bitwise_not, invert, and ldexp
• Bitwise Array and MaskedArray operations
• size attribute on Array and MaskedArray
• astype function on Array and MaskedArray for changing the dtype
• flatten function on Array and MaskedArray for flattening into a 1D array
• MaskedArray.compressed for getting all unmasked data as a 1D array
• get function on Dict and KnownDict for providing a default value if a key does not exist
• nbands attribute on Image and ImageCollection
• proxify can handle scenes.GeoContexts
• Dict.contains, Dict.length

• Fewer failures and hanging calls when connecting to the Workflows backend (like .compute, .visualize, Job.get, etc.)
• wf.numpy.histogram works correctly with computed values for range and bins (such as range=[arr.min(), arr.max()])
• More consistent throughput when a large number of jobs are submitted
• Arrays can now be constructed from proxy Lists
• MaskedArray.filled works correctly when passed Python values
• Long-running sessions (like Jupyter kernels) refresh credentials instead of failing with auth errors after many hours of use
• wf.numpy.dot and wf.numpy.einsum no longer fail when being called correctly
• Occasional errors like ('array-89199362e9a5d598fb5c82805136834d', 0, 0) when calling wf.compute() with multiple values are resolved

• pick_bands accepts duplicate band names. Enjoy easier Sentinel-1 "vv vh vv" visualizations!
• ImageCollection.from_id is always ordered by date
• wf.numpy.percentile no longer accepts an axis argument
• breaking wf.Job construction and interface changes:
  • Use a single wf.Job(..) call instead of wf.Job.build(...).execute() to create and launch a Job
  • New Job.result_to_file method
  • Job.status is removed in favor of a single Job.stage
  • wf.TimeoutError renamed to wf.JobTimeoutError

• 191 functions from NumPy are available for Workflows Arrays, including parts of the numpy.linalg and numpy.ma submodules. See the full list on the docs.
• index_to_coords and coords_to_index methods on Image/ImageCollection/GeoContext for converting between geospatial and array coordinates
• value_at function on Image and ImageCollection for extracting single pixel values at spatial coordinates.

• Using datetimes as parameters to visualize behaves correctly.

• Fixed a bug that prevented uploading ndarrays of type uint8

• Array support for argmin, argmax, any, all
• pick_bands supports an allow_missing kwarg to drop band names that may be missing from the data without an error.
• wf.compute supports passing lists or tuples of items to compute at the same time. Passing multiple items to wf.compute, rather than calling obj.compute for each separately, is usually faster.
• Casting from Bool to Int: wf.Int(True)
• Experimental .inspect() method for small computations during interactive use.

• [breaking] Array no longer uses type parameters: now you construct an Array with wf.Array([1, 2, 3]), not wf.Array[wf.Int, 1]([1, 2, 3]). Remember, Array is an experimental API and will continue to make frequent breaking changes!
• Workflows now reuses the same gRPC client by default---so repeated or parallel calls to .compute, etc. will be faster. Calling .compute within a thread pool will also be significantly more efficient.

• wf.numpy.histogram correctly accepts a List[Float] as the range argument

1.1.2 fixes a bug which caused Workflows map layers to behave erratically when changing colormaps.

1.1.1 fixes a packaging issue that caused import descarteslabs.workflows to fail.

It also makes NumPy an explicit dependency. NumPy was already a transitive dependency, so this shouldn't cause any changes.

You should NOT install version 1.1.0; 1.1.1 should be used instead in all circumstances.

• Image.upload() now emits a deprecation warning if the image has a cs_code or projection property. The projection defined in the uploaded file is always used and applied to the resulting image in the Catalog.
• Image.upload_ndarray() now emits a deprecation warning if the image has both a cs_code and a projection property. Only one of them may be supplied, and cs_code is given preference.

• SceneCollection.download_mosaic has new default behavior for mask_alpha wherein the alpha band will be used as a mask by default if it is available for all scenes in the collection, even if it is not specified in the list of bands.

• Experimental Array API following the same syntax as NumPy arrays. It supports vectorized operations, broadcasting, and multidimensional indexing.
  • ndarray attribute of Image and ImageCollection will return a MaskedArray.
  • Over 60 NumPy ufuncs are now callable with Workflows Array.
  • Includes other useful Array functions like min(), median(), transpose(), concatenate(), stack(), histogram(), and reshape().
• ImageCollection.sortby_composite() for creating an argmin/argmax composite of an ImageCollection.
• Slicing of List, Tuple, Str, and ImageCollection.
• wf.range for generating a sequence of numbers between start and stop values.
• ImageCollectionGroupby.mosaic() for applying ImageCollection.mosaic to each group.
• wf.exp(), wf.square(), wf.log1p(), wf.arcsin(), wf.arccos(), and wf.arctan()
• Datetime.is_between() for checking if a Datetime falls within a specified date range
• FeatureCollection.contains()
• Container operations on GeometryCollection including:
  • GeometryCollection.contains()
  • GeometryCollection.sorted()
  • GeometryCollection.map()
  • GeometryCollection.filter()
  • GeometryCollection.reduce()
• List and Tuple can now be compared with other instances of their type via __lt__(), __eq__() etc.
• List.__add__() and List.__mul__() for concatenating and duplicating Lists.

• Products without alpha band and nodata value are rejected, instead of silently producing unwanted behavior.
• ImageCollection.concat_bands now throws a better error when trying to concatenate bands from another ImageCollection that is not the same length.
• Any is now promotable to all other types automatically.
• Better error when trying to iterate over Proxytypes.
• Interactive map: calls to visualize now clear layer errors.
• Interactive map: when setting scales, invalid values are highlighted in red.
• Interactive map: a scalebar is shown on the bottom-left by default.
• ImageCollection.mosaic() now in "last-on-top" order, which matches with GDAL and dl.raster. Use mosaic(reverse=True) for the same ordering as in v1.0.0.

• Better errors when specifying invalid type parameters for Proxytypes that require them.
• Field access on Feature, FeatureCollection, Geometry, and GeomeryCollection no longer fails.
• In from_id, processing level 'cubespline' no longer fails.

• There is an entirely new backend supporting asynchronous uploads of image files and ndarrays with the catalog client. There are minor changes to the ImageUpload class (a new events field has subsumed errors, and the job_id field has been removed) but the basic interface is unchanged so most code will keep functioning without any changes.
• It is now possible to cancel image uploads.
• Errors messages are now easier to read.
• Many improvements to the documentation.
• You can now create or retrieve an existing object using the get_or_create method.
• Retrieving a Band or Image by name is now possible by calling get_band or get_image on the Product instance. You can also use the Product's named_id function to get a complete id for images and bands.
• A new convenience function make_valid_name on Image and Band classes will return a sanitized name without invalid characters.
• A new property ATTRIBUTES enumerates which attributes are available for a specific catalog object.
• Trying to set an attribute that does not exist will now raise AttributeError.
• update_related_objects_permissions() should no longer fail with a JSON serialization error.
• Setting a read-only attribute will now raise an AttributeValidationError.
• Saving a new object while one with the same id already exists will now raise a ConflictError instead of BadRequestError.
• If a retrieved object has since been deleted from the catalog, saving any changes or trying to reload it will now raise a DeletedObjectError.
• Resolution fields now accept string values such as "10m" or "0.008 degrees". If the value cannot be parsed, an AttributeValidationError will be raised.
• Changes to the extra_properties attribute are now tracked correctly.

• This release no longer supports Python 2.
• This package is now distributed as a Python 3 wheel which will speed up installation.

• Handling of missing data via empty ImageCollections
  • ImageCollection.from_id returns an empty ImageCollection if no data exist for the given time/place, rather than an error
  • ImageCollection.filter returns an empty ImageCollection if the predicate is False for every Image, rather than an error
  • Image.replace_empty_with and ImageCollection.replace_empty_with for explicitly filling in missing data
  • See the Workflows guide for more information
• Docstrings and examples on every class and function!
• Assigning new metadata to Image properties & bandinfo: Image.with_properties(), Image.with_bandinfo()
• Interactive map: colorbar legends on layers with colormaps (requires matplotlib)
• Dict.from_pairs: construct a Dict from a sequence of key-value pairs
• Map displays a fullscreen button by default ([breaking] if your code adds one, you'll now get two)
• wf.concat for concatentating Image and ImageCollection objects
  • ImageCollection.concat now accepts Image objects; new Image.concat accepts Image or ImageCollection
• ImageCollection.mosaic()
• FeatureCollection.sorted(), FeatureCollection.length(), FeatureCollection.__reversed__()
• GeometryCollection.length(), GeometryCollection.__reversed__()

• wf.zip now supports ImageCollection, FeatureCollection, GeometryCollection as well as List and Str
• Get a GeoContext for the current bounds of the map in any resolution, shape, or CRS (including "utm", which automatically picks the right UTM zone for you) with wf.map.geocontext. Also now returns a Scenes GeoContext for better introspection and use with Raster.
• Better backend type-checking displays the possible arguments for most functions if called incorrectly
• arr_shape included when calling wf.GeoContext.compute()
• More readable errors when communication with the backend fails
• Interactive map: layout handles being resized, for example setting wf.map.layout.height = '1000px'
• Any is no longer callable; Any.cast encouraged
• remove_layer and clear_layers moved from wf.interactive.MapApp class to wf.interactive.Map (non-breaking change)
• [possibly breaking] band renaming in binary operators only occurs when broadcasting: red + red is just red, rather than red_add_red. red + blue is still red_add_blue. Code which depends on accessing bands by name may need to change.

• wf.where propagates masks correctly, and handles metadata correctly with multi-band inputs
• processing_level="surface" actually returns surface-reflectance-processed imagery
• ImageCollection.sorted() works properly
• Viewing global-extent WGS84 images on the Workflows map no longer causes errors
• List proxytype no longer infinitely iterable in Python
• Repeated use of axis="bands" works correctly
• ImageCollection.from_images correctly aligns the bands of the inputs
• Numeric casting (wf.Int(wf.Float(2.2))) works as expected
• More descriptive error when constructing an invalid wf.Datetime
• Computing a single Bool value derived from imagery works correctly

• Update workflows client channel
• Workflows map UI is more stable: errors and layers won't fill the screen

• Catalog client: Added an update() method that allows you to update multiple attributes at once.

• Catalog client: Images and Bands no longer reload the Product after calling save
• Catalog client: Various attributes that are lists now correctly track changes when modifying them with list methods (e.g. Product.owners.append("foo"))
• Catalog client: Error messages generated by the server have a nicer format
• Catalog client: Fix a bug that caused waiting for tasks to never complete
• The minimum numpy version has been bumped to 1.17.14 for Python version > 3.5, which addresses a bug with scenes.display

• .compute() is noticeably faster
• Most of the Python string API is now available on workflows.Str
• Interactive map: more descriptive error when not logged in to iam.descarteslabs.com
• Passing the wrong types into functions causes more descriptive and reliable errors

• RST_STREAM errors when calling .compute() have been eliminated
• Image/ImageCollection.count() is much faster
• .buffer() on vector types now works correctly
• Calling .compute() on a GeometryCollection works

• Catalog client: Added a MaskBand.is_alpha attribute to declare alpha channel behavior for a band.

• The maximum number of extra_properties allowed for Catalog objects has been increased from 10 to 50.
• Fixed bug causing SceneCollection.download to fail.

• When you call .compute() on an Image or ImageCollection, the GeoContext is included on the result object (ImageResult.geocontext, ImageCollectionResult.geocontext)

• Passing a Workflows Timedelta object (instead of a datetime.timedelta) into functions expecting it now behaves correctly
• Arguments to the reducer function for reduce are now in the correct order

• A new catalog client in descarteslabs.catalog makes searching and managing products, bands and images easier. This client encompasses functionality previously split between the descarteslabs.Metadata and descarteslabs.Catalog client, which are now deprecated. Learn how to use the new API in the Catalog guide.
• Property filtering expressions such as used in scenes.search() and FeatureCollection.filter() now support an in_() method.

• SceneCollection.download previously always returned successfully even if one or more of the downloads failed. Now if any of the downloads fail, a RuntimeError is raised, which will detail which destination files failed and why.
• Fixed a bug where geometries used with the Scenes client had coordinates with reduced precision.

• Interactive parameters: add parameters to map layers and interactively control them using widgets
• Spatial convolution with wf.conv2d
• Result containers have helpful reprs when displayed
• Datetime and Timedelta are unpacked into datetime.datetime and datetime.timedelta objects when computed.

• [breaking] Result containers moved to descarteslabs/workflows/results and renamed, appending "Result" to disambiguate (e.g. ImageResult and ImageCollectionResult)
• [breaking] .bands and .images attributes of ImageResult and ImageCollectionResult renamed .ndarray
• [breaking] When compute-ing an Image or ImageCollection, the order of bandinfo is only correct for Python >= 3.6
• Interactive maps: coordinates are displayed in lat, lon order instead of lon, lat for easier copy-pasting
• Interactive maps: each layer now has an associated output that is populated when running autoscale and deleted when the layer is removed
• Interactive maps: Image.visualize returns a Layer object, making it easier to adjust Layer.parameters or integrate with other widgets

• Composing operations onto imported Workflows no longer causes nondeterministic errors when computed
• Interactive maps: remove_layer doesn't cause an error
• No more errors when creating a wf.parameter for Datetime and other complex types
• .where no longer causes a backend error
• Calling wf.map.geocontext() when the map is not fully initialized raises an informative error
• Operations on numbers computed from raster data (like img_collection.mean(axis=None)) no longer fail when computed
• Colormap succeeds when the Image contains only 1 value

• Raster.stack max_workers is limited to 25 workers, and will raise a warning and set the value to 25 if a value more than 25 is specified.

• Interactive maps: clear_layers and remove_layer methods
• ImageCollections: reversed operator
• ImageCollections: concat and sorted methods
• ImageCollections: head, tail, and partition methods for slicing
• ImageCollections: where method for filtering by condition
• ImageCollections map_window method for applying sliding windows
• ImageCollections: Indexing into ImageCollections is supported (imgs[1])
• [breaking] Statistics functions are now applied to named axes
• DateTime, Timedelta, Geocontext, Bool, and Geometry are now computable
• ImageCollectionGroupby ProxyObject for grouping ImageCollection by properties, and applying functions over groups
• ImageCollections: groupby method
• parameter constructor

• Interactive maps: autoscaling is now done in the background
• Tiles requests can now include parameters
• median is noticeably faster
• count is no longer breaks colormaps
• map, filter, and reduce are 2x faster in the "PREPARING" stage
• Significantly better performance for functions that reference variables outside their scope, like

overall_comp = ndvi.mean(axis="images")
deltas = ndvi.map(lambda img: img - overall_comp)

• Full support for floor-division (//) between Datetimes and Timedeltas (imgs.filter(lambda img: img.properties['date'] // wf.Timedelta(days=14))

• [breaking] ImageCollection.one (in favor of indexing)

• scenes.DLTile.assign(pad=...) method added to ease creation of a tile in all ways indentical except for the padding.

• The parameter nbits has been deprecated for catalog bands.

• New interactive map, with GUI controls for multiple layers, scaling, and colormaps.
• Colormaps for single-band images.
• Map interface displays errors that occur while the backend is rendering images.
• ImageCollection compositing no longer changes band names (red does not become red_mean, for example).
• .clip() and .scale() methods for Image/ImageCollection.
• Support specifying raster resampler method.
• Support specifying raster processing level: toa (top-of-atmosphere) or surface [surface reflectance).
• No more tiles 400s for missing data; missing/masked pixels can optionally be filled with a checkerboard pattern.

• Workflows Image.concat renamed Image.concat_bands.
• Data are left in data_range values if physical_range is not set, instead of scaling to the range 0..1.
• Selecting the same band name twice (img.pick_bands("vv vv")) properly raises an error.
• Reduced DeprecationWarnings in Python 3.7.

• Alpha Workflows API client has been added. Access to the Workflows backend is restricted; contact support for more information.
• Workflows support for Python 3 added in channel v0-5.

• Scenes API now supports band scaling and output type specification for rastering methods.
• Methods in the Metadata, Raster, and Vector service clients that accepted GeoJSON geometries now also accept Shapely geometries.

• Add support for user cython modules in tasks.

• Tasks webhook methods no longer require a group_id if a webhook id is provided.
• catalog_id property on images is no longer supported by the API
• Fix scenes.display handling of single band masked arrays with scalar masks
• Fix problems with incomplete UploadTask instances returned by vectors.FeatureCollection.list_uploads

• Metadata, Catalog, and Scenes now support a new storage_state property for managing image metadata and filtering search results. storage_state="available" is the default for new images and indicates that the raster data for that scene is available on the Descartes Labs Platform. storage_state="remote" indicates that the raster data has not yet been processed and made available to client users.
• The following additional colormaps are now supported for bands – 'cool', 'coolwarm', 'hot', 'bwr', 'gist_earth', 'terrain'. Find more details about the colormaps here.
• Scene.ndarray, SceneCollection.stack, and SceneCollection.mosaic now support passing a string as the mask_alpha argument to allow users to specify an alternate band name to use for masking.
• Scenes now supports a new save_image function that allows a user to save a visualization given a filename and extension.
• Tasks now allows you to unambiguously get a function by group id using get_function_by_id.
• All Client APIs now accept a retries argument to override the default retry configuration. The default remains the same as the prior behavior, which is to attempt 3 retries on errors which can be retried.

• Bands of different but compatible types can now be rastered together in Scene.ndarray() and Scene.download() as well as across multiple scenes in SceneCollection.mosaic(), SceneCollection.stack() and SceneCollection.download(). The result will have the most general data type.
• Vector client functions that accept a geometry argument now support passing Shapely shapes in addition to GeoJSON.

• Removed deprecated method Metadata.sources()
• FeatureCollection.filter(geometry) will now raise an InvalidQueryException if you try to overwrite an existing geometry in the filter chain. You can only set the geometry once.

• Many old and obsolete examples were removed from the package.
• Scene.ndarray, SceneCollection.stack, and SceneCollection.mosaic now will automatically mask alpha if the alpha band is available in the relevant scene(s), and will set mask_alpha to False if the alpha band does not exist.
• FeatureCollection.add, FeatureCollection.upload, Vector.create_feature, Vector.create_features, and Vector.upload_features all accept a fix_geometry string argument that determines how to handle certain problem geometries including those which do not follow counter-clockwise winding order (which is required by the GeoJSON spec but not many popular tools). Allowed values are reject (reject invalid geometries with an error), fix (correct invalid geometries if possible and use this corrected value when creating the feature), and accept (the default) which will correct the geometry for internal use but retain the original geometry in the results.
• Vector.get_upload_results and Vector.get_upload_result now accept a pending parameter to include pending uploads in the results. Such pending results will have status: PENDING and, in lieu of a task id, the id attribute will contain the upload id as returned by Vector.upload_features
• UploadTask.status no longer blocks until the upload task is completed, but rather returns the current status of the upload job, which may be PENDING, RUNNING, SUCCESS, or FAILURE.
• The FutureTask.ready and UploadTask.ready property has been added to test whether the task has completed. A return value of True means that if get_result(wait=True) were to be called, it would return without blocking.
• You can now export features to a storage data blob. To export from the vector client, use Vector.export_product_from_query() with a storage key and an optional query. This returns the task id of the export task. You can ask for status using Vector.get_export_results() for all export tasks or Vector.get_export_result() for a specific task by task id.
• FeatureCollection has been extended with this functionality with a FeatureCollection.export() method that takes a storage key. This operates on the filter chain that FeatureCollection represents, or the full product if there is no filter chain. It returns an ExportTask which behaves similar to the FutureTask.
• Catalog.upload_image() and Catalog.upload_ndarray() now will return an upload_id that can be used to query the status of that upload using Catalog.upload_result(). Note that the upload id is the image id and if you use identical image ids Catalog.upload_result() will only show the result of the most recent upload.

• Several typical kinds of non-conforming GeoJSON which previously caused errors can now be accepted or fixed by the FeatureCollection and Vector methods for adding or uploading new vector geometries.

• Fixed issues with Catalog.upload_ndarray() under Windows
• Added header to client requests to better debug retries

• Improved error messages for Catalog client upload methods

• Tasks methods create_function, create_or_get_function, and new_group now have image as a required parameter
• The name parameter is renamed to product_id in Vector.create_product, and FeatureCollection.create and FeatureCollection.copy. The 'name' parameter is renamed to new_product_id in Vector.create_product_from_query. Using name will continue to work, but will be removed completely in future versions.
• The name parameter is no longer required, and is ignored for Vector.replace_product, Vector.update_product, FeatureCollection.update and FeatureCollection.replace. This parameter will be removed completely in future versions.

• Metadata.paged_search has been added and essentially supports the original behavior of Metadata.search prior to release 0.16.0. This method should generally be avoided in favor of Metadata.features (or Metadata.search).

• Fixed typo in UploadTask.status which caused exception when handling certain failure conditions
• FeatureCollection.upload parameter max_errors was not being passed to Vector client.
• Ensure cloudpickle==0.4.0 is version used when creating Tasks.
• Eliminate redundant queries from FeatureCollection.list.

• FeatureCollection.upload and Vector.upload_features now accept an optional max_errors parameter to control how many errors are acceptable before declaring an upload a failure.
• UploadTask (as returned by FeatureCollection.upload and Vector.list_uploads) now has added attributes to better identify what was processed and what errors occurred.
• Storage now has added methods set_file and get_file to allow for better uploading and downloading, respectively, of large files.
• Storage class now has an exists() method that checks whether an object exists in storage at the location of a given key and returns a boolean.
• Scenes.search allows limit=None
• FeatureCollection.delete_features added to support deleting Features that match a filter
• FeatureCollection.delete_features and FeatureCollection.wait_for_copy now use AsyncJob to poll for asynchronous job completion.
• Vector.delete_features_from_query and Vector.get_delete_features_status added to support new FeatureCollection and AsyncJob methods.

• Fixed tasks bugs when including modules with relative paths in sys.path

• Tasks now support passing modules, data and requirements along with the function code, allowing for a more complex and customized execution environment.
• Vector search query results now report their total number of results by means of the standard len() function.

• Metadata.search no longer has a 10,000-item limit, and the number of items returned will be closer to limit. This method no longer accepts the continuation_token parameter.

• Raster client can now handle arbitrarily large numbers of tiles generated from a shape using the new iter_dltiles_from_shape() method which allows you to iterate over large numbers of tiles in a time- and memory-efficient manner. Similarly the existing dltiles_from_shape() method can now handle arbitrarily large numbers of tiles although it can be very slow.
• Vector client upload_features() can now upload contents of a stream (e.g. io.IOBase derivative such as io.StringIO) as well as the contents of a named file.
• Vector FeatureCollection add() method can now handle an arbitrary number of Features. Use of the upload_features() method is still encouraged for large collections.
• Vector client now supports creating a new product from the results of a query against an existing product with the create_product_from_query() method. This support is also accessible via the new FeatureCollection.copy() method.
• XYZTile GeoContext class, helpful for rendering to web maps that use XYZ-style tiles in a spherical Mercator CRS.

• Tasks client FutureTask now instantiates a client if none provided (the default).
• Catalog client methods now properly handle add_namespace parameter.
• Vector Feature now includes valid geojson type 'Feature'.
• Tasks client now raises new GroupTerminalException if a task group stops accepting tasks.
• General documentation fixes.

• Scenes and raster clients have a processing_level parameter that can be used to turn on surface reflectance processing for products that support it

• scenes.GeoContext: better defaults and bounds_crs parameter
  • bounds are no longer limited to WGS84, but can be expressed in any bounds_crs
  • New Scene.default_ctx uses a Scene's geotrans to more accurately determine a GeoContext that will result in no warping of the original data, better handling sinusoidal and other non-rectilinear coordinate reference systems.
  • Important: the default GeoContexts will now return differently-sized rasters than before! They will now be more accurate to the original, unwarped data, but if you were relying on the old defaults, you should now explicitly set the bounds to geometry.bounds, bounds_crs to "EPSG:4326", and align_pixels to True.
• Scene.coverage and SceneCollection.filter_coverage accept any geometry-like object, not just a GeoContext.

• FutureTask inheritance changed from dict to object.

• Can now specify a GPU parameter for tasks.
• Vectors.upload allows you to upload a JSON newline delimited file.
• Vectors.list_uploads allows you to list all uploads for a vector product.
• UploadTask contains the information about an upload and is returned by both methods.

• Vector.list_products and Vector.search_features get query_limit and page_size parameters.

• Vector.upload_features handles new response format.

• Vector client support for retrieving status information about upload jobs. Added methods Vector.get_upload_results and Vector.get_upload_result.

• Shapely is now a full requirement of this package. Note: Windows users should visit https://docs.descarteslabs.com/installation.html#windows-users for installation guidance.
• Reduced the number of retries for some failure types.
• Resolved intermittent SceneCollection.stack bug that manifested as AttributeError: 'NoneType' object has no attribute 'coords' due to Shapely thread-unsafety.
• Tracking system environment to improve installation and support of different systems.

• The vector service is now part of the public package. See descarteslabs.vectors and descarteslabs.client.services.vector.

• Fixed SSL problems when copying clients to forked processes or sharing them among threads
• Removed extra keyword arguments from places client
• Added deprecation warnings for parameters that have been renamed in the Metadata client
• Scenes now exposes more parameters from raster and metadata
• Scenes descarteslabs.scenes.search will take a python datetime object in addition to a string
• Scenes will now allow Feature and FeatureCollection in addition to GeoJSON geometry types
• Fixed Scenes issue preventing access to products with multi-byte data but single-byte alpha bands

• Scene.download, SceneCollection.download, and SceneCollection.download_mosaic methods
• Colormaps supported in descarteslabs.scenes.display
• Task namespaces are automatically created with the first task group

• Moved metadata property filtering to common
• Deprecated create_or_get_function in tasks
• Renamed some examples

• Namespaced auth environment variables: DESCARTESLABS_CLIENT_SECRET and DESCARTESLABS_CLIENT_ID. CLIENT_SECRET and CLIENT_ID will continue to work.
• Tasks runtime check for Python version.

• Documentation updates
• Example updates

• Scenes package
• More examples

• Deprecated add_namespace argument in catalog client (defaults to False now, formerly True)

• Added org to token scope
• Removed deprecated key usage

• Tasks service

• Patched bug in catalog service for py3

• Catalog service
• Storage service

• Switched to start_datetime argument pattern instead of start_date
• Fixed minor regression with descarteslabs.ext clients
• Deprecated token param for Service class

• Raster stack method

• Removed deprecated searching by const_id
• Removed deprecated raster band methods
• Deprecated sat_id parameter for metadata searches
• Changed documentation from readthedocs to https://docs.descarteslabs.com

• Dot notation access to dictionaries returned by services

• Reorganization into a client submodule

• Fix regression for NotFoundError

• Reverted descarteslabs.services.base to descarteslabs.services.service

• Reorganization of services
• Places updated to v2 backend, provides units interface to statistics, which carries some backwards incompatibility.

• Places updated to v2 backend, provides units interface to statistics, which carries some backwards incompatibility.

• Blosc Support for raster array compression transport
• Scrolling support for large metadata searches

• Offset keyword argument in metadata.search has been deprecated. Please use the metadata.features for iterating over large search results

• Complex filtering expressions for image attributes

• Raise explicitly on 409 response
• Keep retrying token refresh until token fully expired
• Fixed race condition when creating .descarteslabs directory

• Added ext namespace
• Metadata multi-get

• Fix OpenSSL install on OSX

• Automatic retry on 504
• Internal API refactoring / improvements for Auth

• Add raster bands methods to metadata service.
• Deprecate raster band methods.
• Add require_bands param to derived bands search method.

• Test suite replaces original token when finished running script tests.

• Support for derived bands endpoints.
• Direct access to const_id to product translation.

• descarteslabs scripts on windows OS.

• Fix auth login

• Add metadata.bands and metadata.products search/get capabilities.
• Add bands/products descriptions
• Additional Placetypes

• Better error messages with timeouts
• Update to latest version of requests

• Major refactor of metadata.search
  • Introduction of "Products" through Metadata.products()
  • metadata entries id now concatenate the product id and the old metadata keys. The original metadata keys are available through entry['key'].
  • Additional sorting available.

• Search & Raster using DLTile Feature GeoJSON or key. Uses output bounds, resolution, and srs to ease searching and rasterizing imagery over tiles.

• Better Error messaging

• DLTile notebook
• save and outfile_basename in Raster.raster()

• Fix metadata.features

• Strict "requests" versions needed due to upstream instability.

• Fix python 3 command line compatibility

• API Change descarteslabs, raster, metadata have all been merged into 'descarteslabs'. 'descarteslabs login' is now 'descarteslabs auth login', 'raster'' is now 'descarteslabs raster', etc.

• A Changelog
• Testing around command-line scripts

• Searching with cloud_fraction = 0
• dltile API documentation

• Fix login bug
• Installation of "requests[security]" for python < 2.7.9

• Doctests

• Python 3 login bug

• Search by Fractions

• Initial release of client library