Skip to content

Node clusters, Sankey diagrams, and clustermq_staged parallelism
Compare
Choose a tag to compare
@wlandau wlandau released this 19 Jul 11:25
v5.3.0
0f01361
  • Allow multiple output files per command.
  • Add Sankey diagram visuals: sankey_drake_graph() and render_sankey_drake_graph().
  • Add static_drake_graph() and render_static_drake_graph() for ggplot2/ggraph static graph visualizations.
  • Add group and clusters arguments to vis_drake_graph(), static_drake_graph(), and drake_graph_info() to optionally condense nodes into clusters.
  • Implement a trace argument to evaluate_plan() to optionally add indicator columns to show which targets got expanded/evaluated with which wildcard values.
  • Rename the always_rename argument to rename in evaluate_plan().
  • Add a rename argument to expand_plan().
  • Implement make(parallelism = "clustermq_staged"), a clustermq-based staged parallelism backend (see #452).
  • Implement make(parallelism = "future_lapply_staged"), a future-based staged parallelism backend (see #450).
  • Depend on codetools rather than CodeDepends for finding global variables.
  • Detect loadd() and readd() dependencies in knitr reports referenced with knitr_in() inside imported functions. Previously, this feature was only available in explicit knitr_in() calls in commands.
  • Skip more tests on CRAN. White-list tests instead of blacklisting them in order to try to keep check time under the official 10-minute cap.
  • Disallow wildcard names to grep-match other wildcard names or any replacement values. This will prevent careless mistakes and confusion when generating drake_plan()s.
  • Prevent persistent workers from hanging when a target fails.
  • Move the example template files to https://github.com/ropensci/drake/tree/master/inst/hpc_template_files.
  • Deprecate drake_batchtools_tmpl_file() in favor of drake_hpc_template_file() and drake_hpc_template_files().
  • Add a garbage_collection argument to make(). If TRUE, gc() is called after every new build of a target.
  • Remove redundant calls to sanitize_plan() in make().
  • Change tracked() to accept only a drake_config() object as an argument. Yes, it is technically a breaking change, but it is only a small break, and it is the correct API choice.
  • Move visualization and hpc package dependencies to "Suggests:" rather than "Imports:" in the DESCRIPTION file.
  • Allow processing of codeless knitr reports without warnings.
Assets 2
Loading