Removed LazyData, unused.
Extra info now provided by RNetCDF was causing non-nested data frame problems, fixed by PR from @mjwoods in #44 and #45.
Changed default branch name to “main”, so please check if you have a fork and submit PRs.
Fix bug where Scalar variables were treated as an axis. Picked up in stars PR #399.
Simplified error handling when file not able to be opened. https://github.com/ropensci/tidync/issues/98
Fixed bug in nc_atts() https://github.com/hypertidy/ncmeta/issues/36.
Extra checks and fixes to align with stars, and a future release of RNetCDF.
Condition on version of tidyr for new nest() syntax
> 0.8.3.
The output of nc_atts() is now more consistent, with
the same structure given for only global attributes, or a mix of
variable attributes and global attributes. If there are no attributes at
all the output has zero rows, but now has the correct column
types.
nc_grids() now normal form, with nested variables so
we can easily link grids to variables.
New functions nc_coord_var to find coordinate
variables (if any), gm_to_prj to determine PROJ string in
use, and nc_grid_mapping_atts to determine grid mapping
parameters; #19, #14, #12, #9.
Fix for grid organization providing variables out of native order.
Added ‘variable’ argument to ‘nc_atts’, per #8.
system.file() where the file does not exist)*updates from CRAN feedback
ncmeta now provides support for dimensions that do not have explicit coordinates: dimension and variable tables now have information about “dimvals”, in the form of “dim_coord” and “coord_dim” i.e. if a variable is 1D and its name corresponds to a dimension name, then it is a rectlinear coordinate vector of that dimension. (The coordinates can be of type “char”, and that must be dealt with down stream as it now is in tidync).
new function nc_grids, the spaces available to variables
new function nc_axes, nc_axis for the instances of dimensions
nc_vars now returns only variables
First working version.