- tabledap() requests can now be downloaded as a parquet file, making for a much smaller download
- units have been added to tabledap() output
- griddap() bug fixed when a coordinate has a very large value, such as for some projected data.
- browse() now returns the URL if base::interactive is FALSE, as the documentation states
- 'tabledap()' responses now have the datatype given in the file .dds
- fixes problem with time bounds check and "last"
- Provides exta checks on time bounds.
- Fixes some typos and a mistake showing 'global_search()' in vignette
Ensure '[' and ']' properly encoded in URL
Changed default cacheing behavior and 'cache_setup()'
- griddap dataframe now uses the same coordinate names returned in 'rerddap::info()'
- all grids can now be "melted" into a dataframe, not just lat-lon grids
- fixed some bugs accessing some datasets not on lat-lon grid
- vignette now included in package.
- Added global search function
- fixed bug when dataset has a decreasing coordinate that is not latitude or longitude
- fixed a bug in dealing with trailing slashes in URLs
- fix a broken test
- vignettes only on package documentation site now (#87)
server()
(to fetch known ERDDAP server URLs) now uses the list maintained byirishmarineinstitute/awesome-erddap
on GitHub (#86)- better error handling for
griddap()
: if no dimension arguments passed, we error saying so (and no http requests made); in addition, if a dataset is passed togriddap()
, to which the output ofinfo()
was also passed, then we can check if the dataset has griddap data or not, and fail saying so if not (#91) griddap()
andtabledap()
: ifinfo()
output passed to these two funcitons, we will now use the url within that info output, and use a message telling the user we are doing so; now you don't have to set the url if you pass info output (#92)
- fix a
convert_units
test that was failing because remote service had changed the response
- fix to internal fxn
err_handle()
for handling http errors - ERDDAP servers changed to some weird JSON-ish type format (#85)
- change all
tibble::as_data_frame
/tibble::data_frame
totibble::as_tibble
(#79) info()
gains new element in its output list,base_url
, the base url for the ERDDAP server under consideration (#80)- improved docs for
griddap()
with respect to what's returned from the function (#81) - fix some test fixtures to use preserve exact bytes so that cran checks on debian clang devel don't fail (#83)
- add .github files: contributing, issue template, pull request template
- fix for lat/lon parsing within
griddap()
to account for cases when min and max are reversed from the order they should be in (#78) - fix to
griddap()
to parse additioanl dimensions returned; previously we were only returning time, lat, and lon, plus one more (#82) thanks @afredstonhermann
- added new
Caching
section to package level manual file (?rerddap
) about caching (#52) - use markdown docs in package (#75)
- replace
httr
withcrul
(#54) - cache most tests with HTTP requests using
vcr
(#76) - add test for
read
parameter ingriddap()
(#47) - use default url via
eurl()
; used as default in main functions; set default url with env vars, see?eurl
(#41) - improve handling and reporting back to user of ERDDAP server errors (#70) (#73)
- change to
griddap()
: when nc format gridded datasets have latitude and longitude we "melt" them into a data.frame for easy downstream consumption. When nc format gridded datasets do not have latitude and longitude components, we do not read in the data, throw a warning saying so. You can readin the nc file yourself with the file path (#74) - for for
griddap()
to support cases in wihch lat/lon runs north to south and south to north (#68)
memory()
usage ingriddap()
wasn't working. fixed now (#77)
- Now using
hoardr
to manage caching paths and such (#60). Also now asking users where they want to cache files, either in arappdirs
user cache dir or a temp directory. Now on tests and examples we use temp dirs. - Related to above, new functions
cache_info()
to get cache path and number of cached files, andcache_setup()
to set cache path. - Related to above,
cache_details()
,cache_list()
, andcache_delete()
lose theircache_path
parameter - now cache path is set package wide and we use the same cache path, so no need to set in the fxn call.
- Fixes to a number of
griddap()
andtabledap()
examples to use datasets that still exist (previous examples used datasets that are no gone)
- New vignette added that goes in to much more depth than the original vignette (#51) thx to @rmendels
info()
function gains new attributeurl
with the base url for the ERDDAP server used (#42)- Replaced usage of internal compact data.frame code to
use
tibble
package (#45)
- Added another ERDDAP server to
servers()
function (#49) - Changed base URLs for default ERDDAP server from
http
tohttps
(#50) - Added note to docs for
griddap()
andtabledap()
for how to best deal with 500 server errors (#48) - Replaced all
dplyr::rbind_all
uses withdplyr::bind_rows
(#46)
- Removed use of
ncdf
package, which has been taken off CRAN. Usingncdf4
now for all NetCDF file manipulation. (#35) - Failing better now with custom error catching (#31)
- Added many internal checks for parameter inputs, warning or stopping as necessary - ERDDAP servers silently drop with no informative messages (#32)
- Using now
file.info()$size
instead offile.size()
to be backwards compatible with R versions < 3.2
- Cache functions accept the outputs of
griddap()
andtabledap()
so that the user can easily see cache details or delete the file from the cache without having to manually get the file name. (#30)
- All package dependencies now use
importFrom
so we only import functions we need instead of their global namespaces.
- Fixed bug in parsing data from netcdf files, affected the
griddap()
function (#28)
- Added a suite of functions to manage local cached files (#17)
- Added new ERDDAP server to list of servers in the
servers()
function (#21)
- Fixed a few cases across a number of functions in which an empty list
passed to
query
parmaeter inhttr::GET()
caused an error (#23) - Fixed retrieval of path to file written to disk by
httr::write_disk()
(#24) last
is a value accepted by ERDDAP servers, but internal functions weren't checking correctly, fixed now. (#25)as.info()
wasn't passing on theurl
parameter to theinfo()
function. fixed now. (#26)
- released to CRAN