                     Changes in version 0.3.0.9000                      

                 Changes in version 0.3.0 (2026-10-01)                  

Accepted on CRAN (2026-10-01).

Breaking changes

  - bdpe_fetch_chunks() now requests chunk_size = 50000 records per
    chunk by default (was 500000), matching 'apifetch'. Very large
    chunks were prone to timeouts and high memory use; pass chunk_size
    = 500000 to restore the old behaviour.
  - bdpe_fetch_data() now drops the API's Mensagem status column, as
    bdpe_fetch_chunks() already did, so both functions return the same
    columns.

Other changes

  - Now requires 'apifetch' >= 0.2.0, which brings its bug fixes to
    every bdpe_* function: bdpe_fetch_chunks() accepts chunk_size = Inf
    and never returns more than total_limit rows; the query argument
    drops NULL/NA values, repeats multi-valued parameters and works with
    an endpoint that already has a query string; dataset names with
    accents map to the same plain-ASCII token name on every platform
    (macOS used to produce names such as BigDataPE_Sa'ude); and
    HTTP 401/403, empty and non-JSON responses give clearer messages.
  - bdpe_store_token() gains overwrite = FALSE, to replace an expired
    token. Previously the "already defined" warning suggested overwrite
    = TRUE, which bdpe_store_token() did not accept.
  - Added a test suite (with mocked HTTP) for the Big Data PE service
    conventions.
  - Added authors' ORCID iDs and a BugReports field.

                 Changes in version 0.2.0 (2026-07-15)                  

Accepted on CRAN (2026-07-15).

Changes

  - BigDataPE is now a thin wrapper around the new generic 'apifetch'
    package: every bdpe_* function delegates to its apifetch::af_*
    counterpart with the Big Data PE service conventions (token sent
    verbatim in the Authorization header, limit/offset as HTTP headers,
    the "Mensagem" status column dropped when combining chunks). The
    public API is unchanged, and the environment-variable token names
    (BigDataPE_<dataset>) are identical, so existing code and stored
    tokens keep working.
  - Imports is now just apifetch; the direct dependencies on cli, dplyr,
    tibble, and httr2 are inherited transitively through apifetch.

                 Changes in version 0.1.0 (2026-04-07)                  

New Features

  - Migrated all user-facing messages, warnings, and errors from base R
    (message(), stop(), stopifnot()) to the cli package
    (cli::cli_abort(), cli::cli_alert_success(),
    cli::cli_alert_warning(), cli::cli_alert_info()). This provides
    richer, more informative, and consistently formatted console output.
  - Added progress feedback in bdpe_fetch_chunks() when verbosity > 0,
    showing the number of records fetched per chunk and the running
    total.
  - Added a completion summary message in bdpe_fetch_chunks() reporting
    the total number of records retrieved.
  - Improved input validation in bdpe_fetch_data() and
    bdpe_fetch_chunks() with descriptive cli error messages replacing
    stopifnot() calls.
  - Numeric parameters (limit, offset, total_limit, chunk_size,
    verbosity) are now automatically coerced to integer when possible
    (e.g., limit = 50 now works without requiring limit = 50L). An
    informative error is raised only when conversion is not possible
    (e.g., non-numeric or fractional values).
  - Improved HTTP error handling in bdpe_fetch_data(): connection
    failures and HTTP errors (e.g., 503 Service Unavailable) now produce
    clear, user-friendly messages with guidance on network requirements
    and next steps, instead of raw httr2 errors.

Changes

  - Added cli as a package dependency in Imports.
  - Added Marcos Wasiliew and Carlos Amorim as package authors.
  - Added vignette "Exploring Dengue Data with BigDataPE" demonstrating
    the full package workflow with the public dengue dataset.

                 Changes in version 0.0.96 (2025-04-02)                 

Bug Fixes

  - Fixed an issue in bdpe_fetch_chunks() where the offset parameter was
    being formatted as a scientific double (e.g., 1e+05), which caused
    the API to fail. The value is now properly formatted as an integer
    string, ensuring compatibility with the API which expects an exact
    integer format.
