Skip to content

Latest commit

 

History

History
74 lines (56 loc) · 6.11 KB

Changes.md

File metadata and controls

74 lines (56 loc) · 6.11 KB
Raw

0.59.0.0

  • Breaking Change : Changed plug type of extraDeadlineSettings and extraEnvironmentVariables to AtomicCompoundDataPlug. This allows the values to be set by registering userDefault metadata and prevents adding non-sensical data such as shaders to these plugs. It will break existing expressions connected to these plugs. The broken expression nodes will still exist and can be reconnected by replacing the __disconnected = IECore.CompoundObjectData( YourCompoundData ) variable assignment with parent[YourNodeName]["dispatcher"]["deadline"]["extraDeadlineSettings"] = IECore.CompoundData( YourCompoundData ) or parent[YourNodeName]["dispatcher"]["deadline"]["extraEnvironmentVariables"] = IECore.CompoundData( YourCompoundData ).

0.58.0.0

  • Add menu entry /Dispatch/Deadline Dispatch for compatibility with Gaffer 1.4.

  • Breaking change : Changed the naming of the temporary files created at submission time to send settings to Deadline. Files are now named by the hash of the task node.

  • Fixed error Context has no variable named "frame" when dispatching with DeadlineDispatch.

  • Breaking change : Changed the API for GafferDeadlineJob.submitJob(). It now takes a single directory where the job and plugin submission files will be saved.

  • Breaking change : Temporary job submission files are given random names by Python's tempfile module instead of attempting to use the hash of the dispatch node.

  • API : Added GafferDeadlineJob.environmentVariables() method.

  • Fixed bug that prevented context variables from being substituted in the deadlineSettings and environmentVariables plugs.

  • Fixed bug that prevented context variables from being substituted into GafferDeadline plugs. (#79)

  • Added extraDeadlineSettings and extraEnvironmentVariables plugs. These can be set by an expression to add arbitrary numbers of Deadline settings and environment variables. Entries in these plugs will take precedence over identically named settings / variables in the deadlineSettings and environmentVariables plugs.

  • Breaking change : Changed job names to be ${dispatcher.jobName}.${taskNodeName}. (#80)

  • Added batchName plug to allow easy customization of the batch name. Previously it would be set to the dispatcher's jobName plug value unless overridden in deadlineSettings. The default value is the same as dispatchers' jobName default value, so unless you change the batchName plug value, batches will be named the same as previously.

  • Updated supported Gaffer versions to 1.3.16.7 and 1.4.11.0.

0.57.3.0

  • Fixed bug causing an error when dispatching when passing pathlib.Path values to GafferDeadlineJob.setAuxFiles().
  • Added support for GafferScene.RenderPassWedge. This fixes the error TypeError: unsupported operand type(s) for -: 'NoneType' and 'NoneType' raised when attempting to dispatch a RenderPassWedge.

0.57.2.0

  • Reduced the severity of the check on no-op node batch size from an error to a warning.
  • Removed support for Gaffer versions 1.2.1.0 and 1.2.6.0
  • Added support for Gaffer versions 1.3.7.0 and 1.2.10.5
  • Added support for Arnold progress updates in the Deadline plugin.
  • Update Deadline Plugin to use Python 3.
  • Update Deadline Plugin to be compatible with Python 3.10.

0.57.1.0

  • Added GafferDeadline.DeadlineTask node. Instead of submitting a Gaffer plugin to Deadline, this node will submit any plugin, set by the plugin plug. Plugin parameters can be added to the parameters plug and will be submitted with the job.

0.57.0.0

  • Added controls for Gaffer's -threads command line parameter.
    • The GafferDeadline dispatcher plug threads can be used to control the thread count on a per job basis. If threads is set to 0, it will be ignored when executing the Deadline job.
    • GafferDeadline will also recognize a Deadline Worker's CpuAffinity property, if set, and use the number of enabled CPUs as the thread count.
    • If the threads plug is non-zero and a Worker has its CpuAffinity property enabled, Gaffer will use the lesser of the two values as its thread count.
  • Added logLevel :
    • Added plug to the dispatcher settings to control the IECORE_LOG_LEVEL environment variable for submitted jobs.
    • API : Added GafferDeadlineJob.setLogLevel() and GafferDeadline.getLogLevel() methods.
  • Added output support :
    • Added outputs plug to GafferDeadline settings. The values of this plug will use all string substitutions except frame substitutions. This allows Deadline to substitute frame numbers itself.
    • API :
      • Added GafferDeadlineJob.addOutput() to add an output with an optional context for substitutions.
      • Added GafferDeadlineJob.getOutputs() to return the current job outputs.
      • Added GafferDeadlineJob.clearOutputs() to remove all job outputs.

0.56.1.0

  • Added Alfred style progress (commonly used by Houdini renderers) updates in the Deadline plugin.
  • Added support for Gaffer 0.61.1.2, 1.2.0.0a1 and 1.2.0.0a2.
  • Added the ability to use pathlib.Path objects in auxiliary files. This can be done, for example, in a prespool signal handler that modifies the jobs auxiliary files via setAuxFiles().
  • DeadlineDispatcherTest : Removed support for Python 2.

0.56.0.1

Bugs Squished

  • Fixed error "This application failed to start because no Qt platform plugin could be initialized." when running dispatch app on a headless system.

0.56.0.0

New Features

  • Deadline Plugin : Added this Changes.md file to track changes. For previous updates, see the release history at https://github.com/hypothetical-inc/GafferDeadline/releases
  • Added support for Gaffer 0.60.7.0 and 0.60.8.0.
  • Task Logging : Fixed logging of Gaffer errors. Previously the Gaffer Deadline plugin would terminate at the first line containing "ERROR : ", but Gaffer prints the more interesting error information on subsequent lines.

Breaking Changes

  • A task will now only error if Gaffer exits with a non-zero exit code. To accomplish this in a PythonCommand, you can raise RuntimeError("Error Message") or assert. Outputting "ERROR : ", for example from IECore.msg(IECore.Msg.Level.Error...), will no longer fail the task.