owner
in the request payload is no longer supported. Project
is now required and must be included.There are two noteworthy API changes for users performing data processing tasks in Crunch:
1. There is no longer a “Personal Project” to put datasets, which previously was the default place for creating datasets. This affects both creating/uploading new datasets, and making forks of existing ones.
newDataset(df, project = "/My Project/")
, or a project newDataset(df, project = projects()[["My Project"]]))
. There is also an option to add a path to the environment variable R_CRUNCH_DEFAULT_PROJECT
(or option crunch.default.project
) to use when no project is specified.loadDataset("ds name")
) will no longer be useful, because it only finds datasets in your personal project or shared via “direct share”, which is also being deprecated. Instead, you must load datasets by ID/URL or with the path from a project.2. Until further notice — The ability to convert a “derived” variable into a “materialized” one is soft deprecated (e.g., via is.derived(ds$var) <- FALSE
). Rcrunch is preparing for this by changing these behaviors now:
R_CRUNCH_DEFAULT_DERIVED
) or the option crunch.default.derived
to FALSE so that newly created variables are materialized from the start. It’s also better documented in functions like VarDef
and deriveArray
that allow you to set the derive flag.is.derived()
: Will warn once per session that this is deprecated. You can install now via remotes::install_github("Crunch-io/rcrunch")
or it should make its way to CRAN in in the coming 2–6 weeks, depending on other items for the longer CRAN release cycle.drop
and rename
, preventing common workflows such as variable deletion-and-re-creation. These restrictions have been relaxed, but users should exercise caution when making backward-incompatible changes that could render other variables or user artifacts invalid.