v0.2.3 (October, 2023)

• fix for default Project creation in JupyterLab
• better handling of non-reloadable packages like numpy
• interface for removing unused Projects respects non-interactive mode
• codespell workflow for spell checking files
• lots of updates to the paper!

v.0.2.2 (August, 2023)

This release includes four major updates:

1. We have added a "project" infrastructure for isolating davos-related changes from the user's runtime environment. When Davos is imported, a new lightweight "virtual environment-like infrastructure" is created. These environments are different from (and somewhat lighter-weight than) standard virtual environments in three ways: (a) they don't contain their own python or pip executables, (b) they don't need to be activated or deactivated in order to use them, and (c) they extend rather than replace the current runtime environment. Any smuggled packages (and their dependencies) that are not available in the runtime environment are automatically installed to the notebook's environment instead of altering the user's system. This means that davos-enhanced notebooks will no longer interfere with each other, or with the user's system. By default each davos-enhanced notebook gets its own associated environment. However, this may be customized by setting the davos.project attribute to a user-defined string, pathlib.Path, or davos.Project. If the project is shared across several notebooks, Davos can share package installations across those notebooks, rather than creating new copies of each package for each individual notebook. Setting davos.project to None will revert Davos to its pre-v0.2.0 behavior by installing any "smuggled" packages to the current runtime environment. Project can be inspected and managed from within the notebook — e.g., deleting them/“clean up after yourself” when you’re done with a project. Here's a screencast as an example (credit: @paxtonfitzpatrick):

2. To enable easier customization of Davos's behavior, we now allow many of the configurable attributes to be tunable from the main davos module variable rather than solely via davos.config. For example, to suppress verbose output when new packages are installed, either of the following will now work:

# pre v. 0.2.2 version (legacy support)
import davos
davos.config.suppress_stdout = True

# new syntax option as of v 0.2.2
import davos
davos.suppress_stdout = True

3. We've included new syntax for requiring a specific Python version or pip version, using davos.require_python(...) and davos.require_pip(...), respectively. This is useful when specified dependencies only run on particular versions of Python, etc.

4. We have added (unofficial and incomplete) support for using davos in IPython shells. More extensive and robust support for IPython shells will be implemented in a future release.

This release also includes many bug fixes, enhancements, and other optimizations. A complete changelog is available here.

v0.1.1 (November, 2022)

New companion paper

• check out our new paper describing how davos can be used to simplify and enhance sharing reproducible research code!

Other fixes & improvements

• smuggle statements that install packages into alternate Python environments (by changing davos.config.pip_executable) now successfully import the package in addition to installing it
• enabling davos's non-interactive mode now ensures the installer program doesn't request user input (in addition to davos itself)
• updates to environment-specific configuration following colab's transition to new IPython version
• fixes to CI tests to accommodate recent selenium changes
• new intro sections for README (+assorted typo fixes)

Full Changelog: v0.1.0...v0.1.1

v0.1.0 (January, 2022)

Initial release! See the README for documentation.