You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
# pak 0.3.0
* pak functions that used to return tibbles return data frames now.
While data frames and tibbles are very similar, they are not completely
compatible. To convert the outputs of pak functions to tibbles call the
`tibble::as_tibble()` function on them. If the pillar package is loaded,
it improves the printing of the returned data frames.
Relatedly, `pak::pak_install_extra()` installs pillar now, instead of tibble.
* pak now supports `file://` repositories.
* pak now uses HTTP 1.1 to download packages on Linux, in addition to macOS.
This fixes HTTP issues with some servers (#358).
* New `?ignore-before-r` parameter to ignore optional dependencies that
need a newer R version (r-lib/pkgdepends#243).
* New `?ignore` parameter to ignore an optional dependency.
* Allow specifying downstream package parameters with the `package=?param`
syntax.
* `lockfile_install()` now works better for `any::` refs, and pak always
install the version it has planned for.
* System requirement installation is now more robust and works for
Unix shell expressions (#347).
* CRAN-like resolution is more robust now if a repository is missing
the usual metadata.
* The lock file is pretty JSON now.
* pak now handles all version requirement types properly:
'<', '<=', `==`, `>=`, `>`.
* The dependency solver now uses better heuristics and does not
(effectively) freeze if multiple repositories have multiple versions of
the same packages (e.g. RSPM and CRAN)
(r-lib/pkgdepends#277)
# pak 0.2.1
No user visible changes.
E.g.
pkg?ignore
and thenpkg
will be ignored, even if it is suggested. IDK if we should allow ignoring hard dependencies.The text was updated successfully, but these errors were encountered: