Change log¶
Version 2.0.4 May 6, 2023
Update of Python versions in documentation and source.
Version 2.0.3 Jun 13, 2022
Use strings for Python version in .github/workflow files. (Changed in documentation and Roberto’s own CI files.)
Version 2.0.2 Oct 18, 2021
Upgrade pip before using it.
Version 2.0.1 Oct 18, 2021
Fix nuke-testenv for conda.
Minor documentation and output improvements.
Use f-strings in source code.
Fix few new types of pylint issues.
Version 2.0.0 Mar 20, 2021
Support for two types of test environments: conda and venv. (It is also possible not to activate any environment.) The default is venv, which is the fastest of the two. venv does not support all features of Roberto, but can handle most.
Support for two types of package managers: conda and pip. (Obviously, conda can only be used in a conda environment.)
Improved output readability.
All CI tools have corresponding classes in roberto/tools.py.
Improved activate script handling.
Many small fixes.
Switch from Travis to Github Actions.
Version 1.11.0 Aug 27, 2020
Derive development classifier from git tag.
Version 1.10.0 Aug 16, 2020
Improved support for deployment of pre-releases.
Version 1.9.0 Aug 15, 2020
Configure channels with prepend instead of append.
Version 1.8.0 May 14, 2020
Fix some URLs (Miniconda)
Install pytest-regressions extensions by default
Test requirements in Conda environment.
Version 1.7.0 May 16, 2019
Fix typo (CXXFLAGS).
Skip dependency installation for absent conda recipes without crashing.
Use
dist_name
instead ofconda_name
in the configuration.
Version 1.6.1 April 14, 2019
Small fixes, mostly for cpp projects.
Version 1.6.0 April 12, 2019
Add formatter dependencies for cardboardlint to default config.
Add rst-lint dependency for cardboardlint to default config.
Add ROBERTO_ABS=1 option to disable relative linting.
Version 1.5.0 April 09, 2019
Call cardboardlinter with parallel option.
Improved processing of coverage resutls for C++ code.
Minor fixes.
Version 1.4.1 March 30, 2019
Fix typo in tasks.py
Version 1.4.0 March 30, 2019
Only conda requirements in default config. Required packages were added to conda-forge.
Start using stable versions of cardboardlint.
Add more requirements that were missing from default config.
Document API with sphinxcontrib-autodoc.
Generate files for doc from conf.py. Separate script no longer needed.
Various website improvements.
Fix issue with building for python != 3.7.
Version 1.3.0 March 26, 2019
Option to add project-specific extra requirements, which may be convenient when using certain Sphinx extensions. This was also possible in the previous version by rewriting the corresponding build-docs tool, but this was a rather cumbersome approach.
Documentation improvements, also for the CI setup on Travis.
conda* packages are installed and updated in the base environment. This makes sure the most up-to-date versions are used. Upgrading them in a sub-environment may have no effect.
The stderr is hidden for commands that are allowed to fail.
A few minor fixups.
Version 1.2.1 March 22, 2019
Fix the detection of a pull request, such that it also works on Travis-CI.
Version 1.2.0 March 21, 2019
Make pytest run in parallel.
Parallel in-place build with CMake.
Default task has become “quality”, i.e. the in-place subset of the tests.
More configuration options for in-place builds (paths and flags).
Fix SDK blues on OSX.
Various small fixes.
Versoin 1.1.1 March 18, 2019
Fix bugs in SDK download and usage on OSX.
Re-activate conda after every install.
Use yaml.safe_load
Roberto no longer crashes outside a git repo.
Stop using LD_LIBRARY_PATH and use RPATH for dynamic in-place linking instead.
Report SIP status on OSX, which could be useful info in case of troubles.
Fix mistake in generation of PATH variables.
Version 1.1.0 March 17, 2019
Download MacOSX SDK when needed.
Fix PyPi upload (no sha256).
Small documentation fixes.
Version 1.0.1 March 17, 2019
Small documentation improvements.
Fix missing requirement for static cardboard linting
Version 1.0.0 March 17, 2019
This is the first official release, all previous ones being just testing artifacts.