rixpress 0.12.1 (2026-02-04)
Improvements
-
Python Version Detection:
rixpressnow correctly detects the Python version specified indefault.nix(e.g.,python311Packages,python312Packages) and uses the corresponding interpreter (e.g.,${defaultPkgs.python311}/bin/python) instead of defaulting topython3. This prevents ABI version mismatch errors when using compiled Python packages likenumpy.
rixpress 0.12.0 (2026-01-31)
New Features
Chronicler integration: Added support for detecting
{chronicler}packageNothingvalues in pipeline outputs. When usingchronicler’srecord()decorated functions, errors and warnings are captured asNothingvalues instead of failing the build. This can mask pipeline failures.-
rxp_check_chronicles(): New function to scan all pipeline outputs for chronicle objects and report their status:- ✓ Success:
Justvalue, no warnings or errors - ⚠ Warning:
Justvalue, but warnings were captured - ✗ Nothing: Failed computation, errors captured
- ✓ Success:
Enhanced
rxp_read()andrxp_load(): Now automatically warn when reading chronicle objects that containNothingvalues, and inform about chronicles with captured warnings.
rixpress 0.11.2 (2026-01-28)
CRAN release: 2026-01-28
Changes
- Updated
rxp_visnetwork()styling to use a dual-encoding approach (similar torxp_ggdag()) whencolor_by = "pipeline". Now, node interiors are colored by derivation type (e.g., R, Python) while node borders are colored by pipeline group. - Spellchecked
vignettes/sub-pipelines.Rmdto British English. # rixpress 0.11.1 (2026-01-25)
rixpress 0.11.0 (2025-01-24)
New Features
Sub-pipelines with
rxp_pipeline(): New function to organize derivations into named logical groups (e.g., “ETL”, “Model”, “Report”). Pipeline groups are visually distinguished in DAG visualizations with custom colors.Pipeline-based DAG coloring:
rxp_visnetwork()andrxp_ggdag()now support acolor_byparameter. When set to “pipeline” (the default when pipelines are defined), nodes are colored by their pipeline group rather than derivation type.Master Script workflow: Enables organizing large projects across multiple R scripts, then combining them in a master script using
rxp_pipeline().
Internal Changes
-
rxp_populate()now flattensrxp_pipelineobjects while preserving metadata. -
dag.jsonschema extended withpipeline_groupandpipeline_colorfields. -
get_nodes_edges()extracts pipeline metadata for visualization.
rixpress 0.10.2 (2025-01-24)
Fixed bug: relative paths for nix_env (e.g., "../../default.nix") now work correctly. Previously, path characters were converted to underscores, causing invalid Nix variable names like ______defaultBuildInputs. Now only the filename is used, producing valid identifiers like defaultBuildInputs.
rixpress 0.10.1 (2025-10-07)
CRAN release: 2025-10-14
Fixed bug: functions inside of arbitrary folder were not being imported into the build sandbox correctly.
rixpress 0.2.0 (2025-05-12)
Submission for RopenSci review.
-
rxp_rmd(): build RMD documents. -
rxp_list_logs(): list the logs of the builds, and possible to read artifacts of previous builds withrxp_read()(or load withrxp_load()as well. - DAG of pipeline can be visualised with visNetwork or with ggdag.
rixpress 0.1.0 (2025-04-14)
First release (only on GitHub).
New features
- Possibility to define pipelines with R, Python or Quarto outputs. Data transfer between R and Python is made using reticulate.
- Basic plotting of DAG of pipeline.
- Demos available at: https://github.com/b-rodrigues/rixpress_demos
