Skip to content

Latest commit

 

History

History
113 lines (92 loc) · 3.28 KB

CHANGELOG.md

File metadata and controls

113 lines (92 loc) · 3.28 KB
Raw

Tesselate Release Notes

0.7

  • Added function to obtain pixel value from a coordinate.
  • Imporoved logging on traininglayer ingestion.

0.6

Breaking changes:

  • Users and group listing endpoints have been removed.
  • Syntax for updating permissions has changed. User user and group keywords to pass user and group IDs for permission changes.
  • Removed deprecated wmts layer endpoint. The wmts service is now automatically configured using the permissions of a user.

0.5.2

  • Updated the export function to use the changed api output (extent not geom).

0.5.1

  • Added date column to trainingsample ingestor.

0.5

Breaking change: Updated api ur to new address.

Other changes:

  • Allowed ingesting continuous training layers.
  • Fixed pagination on paginated results.
  • Added WMTS creation endpoint.

0.4

Breaking change:

  • Switched all pk arguments to id. The api expects the id keyword, so to stay consistent, the all pk arguments were replaced.

0.3.4

  • Improved warning for paginated queries.

0.3.3

  • Allowed to specify zoom level for aggregation.

0.3.2

  • Slightly improved api error messaging.

0.3.1

  • Fixed asynchronous mode for aggregation utility.

0.3

Breaking change: the data model for training samples has changed. Training samples are no longer assigned to a classifier directly. Instead, they are assigned to a training layer and then the layer as a whole is assigned to the classifier. This improves training data management. It also prevents having "free floating" training samples that are not associated to anything.

Other changes:

  • Fixed nodata bug on RGB exports.
  • Added option to force object deletion without user confirmation.

0.2.2

  • Fixed zoomlevel bug on export, zoomlevel specification was not working.
  • Fixed export of areas that contain empty tiles.

0.2.1

Fixed bug in compositebuild trigger.

0.2.0

First full stack implementation. Tesselate can now be used to create composites, upload training data, train an algorithm, predict a layer and export the result. I.e. the full workflow from zero to predicted layer is now covered!

In detail, the following changes:

  • Added predicted layer endpoint.
  • Added trigger functions for long running tasks: build composite, train classifier, predict layer.
  • Created shapefile ingestor to upload trainig data.

0.1.2

  • Added possibility to manage permissions on objects.
  • Added possibility to request user or group permission lists.
  • Added unittests for auth and main dispatcher.
  • Changed import method for tesselate functionality. Now use from tesselate import Tesselate and then instantiate ts = Tesselate(). The rest works the same way.
  • Added experimental functionality for classify endpoints.
  • Added create and update functionality to rest endpoints.

0.1.1

  • Fixed requests with custom grouping parameter.
  • Fixed synchronous valuecount calculation.
  • Changed base_path to file_path on export function. Setting the full filepath is more flexible than a base path with automatic naming.

0.1

  • First version
  • Endpoints for composites, formula, area, export and valuecounts.
  • Automatic "loose" filter interpretation as kwargs on endpoints.
  • Authentication either by setting token or logging in using credentials.