Release Notes

1.4.0

Change Log

  • Resolved an issue that resulted in Mosaic not downsampling the current expression.
  • Improved logging from compotational server component of Mosaic to assist in diagnosing issues.
  • Completed integration with Blaze 0.11 and Bokeh 0.12.

1.3.0

Change Log

  • Users can now load Excel spreadsheets containing tabular data into Mosaic. Simply add a spreadsheet like any other file, then begin to create expressions or plots to explore the data. Users can also join their Excel spreadsheet data with other datasets by using the join operation.
  • Improved the downsampling functionality and controls to further enable interactive exploration of large datasets.
  • Enhanced the “Add Dataset” dialog to enable custom arguments for every computational backend. This gives users full control over the configuration parameters for the various backends. Additionally, the dataset is automatically reloaded when a configuration parameter is changed.
  • Added an “Edit Metadata” dialog to allow the user to easily edit the metadata and custom arguments associated with a dataset.
  • Resolved an issue that corrupted the column alignment after transform operations.
  • Addressed an issue that caused incorrect rendering of the ancestry tree.

1.2.3

Change Log

  • Changed the Blaze dependency information.

1.2.2

Change Log

  • Addressed an issue where the user could enter an incorrect expression and proceed without correcting it.
  • Addressed an issue with correctly evaluating derived datasets.

1.2.1

Change Log

  • The user can edit the expression’s breadcrumb steps before commiting to the catalog. The user can also insert and remove prior steps. If a error occurs in any step as a result of the change, the user is presented with each error and given an opportunity to correct the problem.
  • The user can revert the expression in the input box.
  • Given the potential for lengthy computations on large datasets, statistics are now computed on demand within the Stats pane.
  • Addressed an issue with computations intermittently not completing on Windows.
  • Resolved an issue where Blaze incorrectly evaluated coerce operations.

1.1.0

Change Log

  • Significant enhancements to the downsampling functionality. Users can now specify different downsample methods and parameters. Additionally, the application now retains the downsampling configuration between sessions.
  • To help provide provenance, datasets are now depreecated rather than deleted. This ensures that the ancestry of an expression is preserved. The catalog now provides a toggle to show or hide deprecated datasets, with an option to restore previously deprecated datasets.
  • Users can now force datasets into memory on the computation server to ensure the fastest possible analytics.

Known Limitations

  • Reverting an in-progress “applied” expression may not update the data table as expected when working with multiple tabs. Any changes to the expression, however, will be correctly applied.

1.0.1

Change Log

  • Lux datasets can be configured through the Mosaic interface. YAML configuration files are no longer necessary.
  • HDF5 files are now supported on Windows.
  • Expressions can now reference both ground truth datasets and derived expressions.
  • Completed changes to allow a dataset to have the same name as a column.
  • The coerce operation now operates as expected.
  • Improved downsampling controls are easier to use and allow the user to specify no downsampling for smaller datasets.
  • Resovled an issue that caused statistics from SQL datasets to not display correctly.
  • Fixed a problem that resulted in Firefox not loading the Mosiac page at startup.
  • To maintain provenance, dataset URIs are no longer editable.

1.0.0

Known Limitations

  • HDF5 on Windows is not currently supported due to issues with the underlying libraries.
  • To use a Lux dataset (flat file repository) you must configure a startup-up YMAL file. Please refer to the Lux tutorial.
  • An expression in Mosaic can only reference other ground truth datasets in the catalog, that is, an expression cannot reference another derived expression.
  • All datasets are automatically downsampled. Future releases will expost greater user control and flexibility of this feature.