Magic `pak()` function

[gaborcsardi]

• Only works interactively.
• Automatically detects if you are developing a package / project / etc.
• Shows the status of your library / package / project.
• Offers some actions to do, depending on the status.
• Has one argument, which is a ref or list of refs, to install or update and/or add to the package.

If the current directory is not in a package

Status:

• pak version, whether there is an update (?).
• Summary of libraries (?).
• Cache summary (?).
• Summary of main (first) library.
• Packages that can be updated in the main library.

Actions:

• Update all packages.
• List all packages.

If the current directory is in a package

• pak version, whether there is an update (?).
• Dependency tree of the package, first level only, annotated.
• Whether the local package is up to date in the main library.
• Dependencies that are missing from the main library.
• Dependencies that are outdated in the main library.
• Summary of "extra" packages, that are not needed for the local package.
• Potential version conflicts of dependencies.

Actions:

• Install/update package.
• Install/update dependencies.
• Install/update dev-dependencies.

[psolymos]

I first though this magic pak function is a cli function that can be used like pip or npm. Is that on the roadmap?

[jimbrig]

I first though this magic pak function is a cli function that can be used like pip or npm. Is that on the roadmap?

@psolymos would love this feature as well.

[gaborcsardi]

@jimbrig Yeah, I agree. The difficulty is that R packages do not have a standard way to "install" executables, so we would need to work that out first.

[jimbrig]

The only thing that comes to mind for me would be to utilize the littler scripts as a form of executables (similar to how most people install R packages within Docker image builds) or an alternative via the docopt package. But good point @gaborcsardi