« Changelog History


LynxKite v1.2.0 Release Notes

    • Graph visualizations are now cut to only show on the corresponding side. The previous behavior was especially disturbing when combined with map view.
    • The user registry file now has to be specified in .kiterc as KITE_USERS_FILE. Earlier versions used the user registry from kite_XXX/conf/users.txt. If you run LynxKite with authentication, please copy this file to a preferable location and add KITE_USERS_FILE to your .kiterc when upgrading.
    • Improvements in the configuration for number of cores per executor:
      • One common .kiterc option (NUM_CORES_PER_EXECUTOR) regardless of deployment mode
      • As a consequence, YARN_CORES_PER_EXECUTOR option has been removed. Please update .kiterc on YARN deployments by renaming YARN_CORES_PER_EXECUTOR to NUM_CORES_PER_EXECUTOR.
      • Use correcty the value of the above option internally to optimize calculations knowing the amount of cores available
      • In case of EC2, correctly set up this new kiterc option based on the cluster configuration
    • Fixed a bug that caused "hard" filtering to segmentations to fail.
    • Spaces in ACL lists are now ignored.
    • Error reports now go to [email protected] instead of [email protected].
    • Previously hidden utility operations (like "Delete vertex attribute") are now revealed on the interface. This is so that they can be accessed when editing the history.
    • Import and simple operations should now work even if there is insufficient memory to hold the entire graph.
    • New icons: triangle, pentagon, star, sim, radio.
    • File sizes are logged for import and upload. (This will be useful for troubleshooting.)
    • Operation search button. (This is the same as pressing /.)
    • Delimiters in load CSV operations now support standard java escapes. (Most importantly, you can import from a tab delimited file using \t as the value of the delimiter parameter.)
    • Project history (the list of operations that have been applied) can now be viewed and also edited. This is an experimental feature. Operations from earlier versions cannot be viewed or edited.
    • Fixed an issue with undo. The fix introduces an undo incompatibility. Operations from earlier versions cannot be undone after the upgrade.
    • Added a "Filter by attributes" operation. This is identical to filtering the usual way.
    • Regex filters can now be applied to string attributes. The syntax is regex(R) for example for finding strings that contain R. Or regex(^Ab) for strings that start with Ab. Or regex((.)\1) for strings with double letters. Etc.
    • Non-neighboring vertices are partially faded out when hovering over a vertex.
    • Fixed a leak in visualization that would gradually degrade performance in the sampled view.
    • In addition to scrolling, double-clicking and right double-clicking can be used now to zoom in and out. With Shift they adjust the thickness.
    • Upgraded to Apache Spark 1.3.0. Users need to download this and update SPARK_HOME in their .kiterc.
    • Fixed logout behavior on the splash page (/).