Skip to content

Latest commit

 

History

History
92 lines (64 loc) · 6.7 KB

NEWS.md

File metadata and controls

92 lines (64 loc) · 6.7 KB

1.1

Add check for valid Socrata resource URL. Add check for supported download file format. Add support for Socrata short dates.

1.2

Use comma-separated file format for Socrata downloads.

1.3

  • Added support for human-readable URL. Users can now copy and paste URLs of Socrata-hosted datasets, which will be transformed into a valid SoDA API web query.

  • Added additional RUnit tests to validate new functionality.

1.4

Add json file format for Socrata downloads. Switch to RJSONIO from rjson.

1.5 Several changes:

  • Swapped to jsonlite from RJSONIO
  • Added handling for long and short dates
  • Added unit test for reading private datasets

1.5.1

Deprecated httr::guess_media() and implemented mime::guess_type()

1.6.0 Several changes:

  • New function, ls.socrata to list all datasets on a Socrata portal.
  • New optional argument, app_token, which lets users supply an API token while using read.socrata() to minimize throttling.
  • Repairs a bug where read.socrata failed when reading in a date with a column, but there are null values in that column.
  • Minor changes to the DESCRIPTION documentation to point users to GitHub for issues and provides new contact information.

1.6.1 Bug fixes:

  • Resolved potential name collision issue
  • Cleaned-up documentation with contributor instructions #23 and #28)
  • Moved test coverage in RUnit to testthat and implemented code coverage monitoring (#41)
  • Clean-up DESCRIPTION (#40)
  • Add continuous integration for Windows (#39)
  • Migrate Travis-CI to "proper" R YAML (#46)

1.7.0 Several changes

New features:

  • Users can upload data with write.socrata() to upload data to Socrata data portals (using "upsert" and "replace" methods).
  • Download private datasets by using Socrata credentials with email and password fields in read.socrata().

Bug fixes:

  • Updated unit testing on ls.socrata() to check for @type field is available.
  • Converts a Socrata money field into a proper numeric field, instead of a factor.
  • Updated build method for Travis to test using the current CRAN packages, not beta packages from GitHub.

1.7.1 Bug fixes:

  • Users provided an option to handle string-like fields as characters (chr) or factors. Default is to handle string-like fields as character vectors (#27)
  • Fixes bug where dates are incorrectly read when first date is a blank (#68)
  • Dates are now handled using POSIXct instead of POSIXlt (#8)
  • Added additional unit testing (#28)
  • Artifacts from Appveyor CI can now be directly submitted to CRAN (#77)
  • Fixed issue where JSON may occasionally come back with a final NULL that is not "[]" (in this example it was "[]\n"). This caused getDataFrame to get stuckin an infinite loop while waiting for "[]". Thank you @kevinsmgov for documenting this bug in issue (#96)
  • Resolved an error users would see when downloading the JSON file on a SoDA call, Error in rbind(deparse.level, ...) : numbers of columns of arguments do not match. (#19)
  • The LIMIT argument was incorrectly interpreted by RSocrata, which sometime causes slower performance (#83, #14)
  • The mandatory @context, @id, conformsTo, and describedBy fields were not shown when using ls.socrata(). These elements are now included as attributes in the resulting data frame. (#72)
  • Fixed a bug which a user-defined token was ignored when the user passed along a $where argument in a SoDA query. (#105)
  • Fixes a bug where the POSIX date conversion would not occur when using the JSON SoDA url (#85)
  • Queries now follow SoDA guidelines and are explicitly sorted. This could have caused a silent error where some row may not have downloaded. (#15)

1.7.2 Several changes:

Performance improvements:

  • By default, 1000 rows at a time were being retreived for each API call while paging through data. Increased to retreive 50,000 rows for each API call. (#129)

Bug fixes:

  • Fixed a bug which caused an error if a select=count() statement was present in a URL. (#120)
  • Fixed a bug when data types are not found because there are no X-SODA2-* headers available in the API response. Users will now get a warning and data will be returned as character. (#118)
  • Fixed a bug which did not recognize URLs listed from ls.socrata() as valid URLs to be used in read.socrata(). Users will now be able to use URLs from the former function in the latter function. (#124)
  • Removed unit tests for older releases of R. (#136)

Deprecation:

RSocrata's core development team has stated a formal policy to only support the most recent release of R. Until now, RSocrata was tested against the penultimate release of R; however, testing will be limited to the current version of R and the current development release. The project's contributing guidelines have been updated to reflect that accepted changes to RSocrata must pass tests on the current and penulimate versions of R.

While RSocrata is only tested on the current and penultimate version, the core development team expects it will work on older versions most of the time. See #132 for more information.

1.7.9-4

CRAN bug fixes

The updates from 1.7.8 to 1.7.9 are all realated to minor bug fixes to pass CRAN tests, which were caused by changes by the data portal vendor.

  • Tests have been updated to reflect changes in the order in which columns are returned
  • Tests have been skipped to reflect the vendor's deletion of user accounts that are consistent with their security policy.