Changes in version 0.4.2.9000                      

  - starts_with() was renamed to begins_with() to prevent name conflicts
    with tidyverse.
  - Plain keys are now assumed to be unprefixed. This eliminates the
    warning message when using unprefixed plain keys. The "plain:"
    prefix is still recognized by chrkeys_to_rawkeys() however
    rawkeys_to_chrkeys() doesn't use it anymore. Percent encoding of
    plain keys now encodes : in order to prevent collisions.
  - New make_actor_keys() function to create actor keys from UniqueIDs.
  - New generate_uuid() function to create random UUIDs.
  - default_db() now defaults to the first opened db instead of the most
    recently opened one. This not only is easier to remember but also
    has simpler bookkeeping. The default db can still be set manually,
    and if the current default db is closed, the next db opened will be
    remembered as the default.

                 Changes in version 0.4.2 (2025-11-09)                  

  - Vanilla data was updated on 2025-11-09.
  - Legacy checksum support was updated for compatibility with the most
    recent version of digest. (@eddelbuettel #14)

                 Changes in version 0.4.1 (2025-08-24)                  

  - Undefined behavior identified by the M1-San system was fixed.

                 Changes in version 0.4.0 (2025-08-23)                  

Breaking Changes

  - This release includes a new API for reading and writing from the
    world database. The new API is designed to better support pipes and
    use a common database connection across functions. Most functions
    associated with reading and writing values to a world database were
    impacted. Consider the process of getting a set of values from a
    common prefix.
    
    db <- bedrockdb(dbpath)
    # Previous API
    dat <- get_data(db, starts_with = prefix)
    # New API
    #   - automatically uses db
    #   - prefixes created with helper function
    dat <- get_data(starts_with(prefix))
    
    # Previous API
    put_data(db, dat)
    # New API
    put_data(dat)
    
    # Previous API
    put_values(db, names(dat), dat)
    # New API
    put_data(dat, names(dat))

  - As part of the new API, the get_values(), set_values(), and related
    _values() functions were removed.

  - The API for working NBT and rNBT values has been reimagined.
    
      - NBT lists are now treated as vectors where ever possible.
      - NBT longs are now stored as strings to protect against data
        loss.
      - NBT integers are now stored as doubles to protect against data
        loss.

  - New get_blocks_* and put_blocks_* functions have replaced
    get_chunk_blocks_* and put_chunk_blocks_*. The new functions work
    per chunk and always return blocks as a 16 x 384 x 16 array,
    regardless of the dimension.

New Features

  - New default_db() function to get/set a default db connection.
  - New print() etc. commands for printing NBT using a tree structure.
  - New with_db() and local_db() for temporarily changing the default db
    connection.

Miscellaneous Fixes and Features

  - New pkgdown-based website for rbedrock.
  - The requirements for a key to be interpreted as a chunk key has been
    relaxed. Now any raw key of length 9, 10, 13, or 14 will be
    considered a chunk key if it contains at least one non-printable
    character. Collisions between plain and chunk keys can still occur,
    but they are not likely in this scheme.
  - New biome_name() function for converting from a numeric id to biome
    name.
  - New read_rnbt_once() function to parse NBT data that is mixed in
    with non NBT data. It returns a single NBT value along with how may
    bytes the value was.
  - New blocks_nbt() and blocks_str() functions to convert block data
    between NBT and string representations.
  - spawning_area() no longer requires a chunk to be withing 96 blocks
    of a player.
  - Vendored LevelDB library is now built with exceptions enabled to
    avoid libcxx including abort() calls if exceptions are not enabled.
  - The following dependencies were removed stringr, dplyr, purrr,
    magrittr, reader, vctrs, tidyr, and fs.

                 Changes in version 0.3.3 (2025-03-25)                  

  - Fix notes identified by CRAN. (Unused calls to stderr and abort.)
  - Update included leveldb library.

                 Changes in version 0.3.2 (2023-12-01)                  

  - Fixed errors in usage of Rf_error().
  - Use R_len_t in more locations.

                 Changes in version 0.3.1 (2023-10-06)                  

  - Remove hardcoded C++17 requirement when calling CMake.

                 Changes in version 0.3.0 (2023-03-20)                  

  - Added support for reading and writing 1.18.30+ actor digests.
  - Chunk data functions have been expanded.
  - [Breaking Change] Biome functions now return CNC biomes data (CnC
    biome functions have been renamed.) Accessing legacy biome data can
    be done with legacy biome functions.
  - [Breaking Change] Chunk data functions and files have been renamed
    to match internal Mojang names.
  - [Breaking Change] Chunk data functions no longer default to reading
    all keys from db if no keys are present.
  - [Breaking Change] Chunk tag key descriptions have been changed to
    better match internal Mojang descriptions.
  - Maximum chunk key tag has been increased to 96.
  - [Breaking Change] Change key format to require prefixes in front of
    every key. This allows rbedrock to support multiple key formats.
  - Add nbt_raw_string to support reading and writing nbt_string data
    that contains embedded nulls.
  - Change default parameters to bedrockdb to match Minecraft. This
    reduces the size of worlds modified by rbedrock.
  - Generate random world seeds using 64-bits.
  - [Fix] Generate default package options on loading rbedrock. This
    fixes users not being able to find the world directory on Window.
  - Create 1.19+ worlds by default when using create_world()
  - Add functions for reading and writing actors data for a chunk.
  - Add utility functions for working with unique ids.

                 Changes in version 0.2.0 (2022-03-10)                  

  - Support subchunk version 9 introduced for Caves and Cliffs.
  - Support Caves and Cliffs' 3DMaps.
  - Change tag format to @x:z:d:t:s. This change was made because
    subtags can now be negative.
  - Add locate_blocks() function.
  - Introduce intermediate rnbt (raw nbt) format.
  - Introduce new NBT API based on S3 classes.
  - Remove layer and simplify arguments to get_subchunk_layers_value()
    and friends.
  - Improved ChunkVersion API
  - Remove -Werror flag used by leveldb on some platforms
  - Added 1.18 example world
  - Add support for 1.18's empty palettes.
  - Rename chunk tag to match internal bedrock names.
  - Add starts_with option to get_values(), and db$mget_prefix() and
    bedrock_leveldb_mget_prefix() function.
  - Add get_subchunk_blocks_from_chunk() function to efficiently lead
    subchunks from a single chunk.
  - Add utility functions for calculating spawning area and simulation
    area.

                 Changes in version 0.1.1 (2021-09-11)                  

  - Fixed errors identified by UBSAN and Valgrind
  - Improved the detection of cmake binary.
  - RBedrock can now be compiled on Solaris using GCC.
  - RBedrock can now be compiled by CRAN's MacOS builder.
  - Updated README.md

                 Changes in version 0.1.0 (2021-07-14)                  

  - Initial Release