NEWS.md

brickster 0.2.15

• The workspace Connections pane identifies managed and foreign Iceberg tables as ICEBERG, distinguishes Delta tables with Iceberg reads as DELTA (UniForm: Iceberg), and shows Iceberg metadata instead of Delta protocol versions. Tables without a source format, including metric views, no longer fail when opening metadata (#311, @zacdav).
• The RStudio connection pane now shows the storage location, last commit time, and minimum reader and writer versions of Delta tables (#304, @m-muecke).
• db_jobs_repair_run() now sends job_parameters correctly (#307, @m-muecke).
• access_control_req_group() now sends the group as group_name, so jobs created with group permissions no longer fail with "Principal name not defined" (#305, @m-muecke).
• dbListFields() no longer returns extra rows for partitioned and clustered tables (#306, @m-muecke).
• Fixed db_sql_query(), dbGetQuery(), and dbFetch() silently returning only the first chunk of INLINE results. Additional chunks are fetched in order up to the requested row limit (#303, @zacdav).
• db_list_all_pages() collects all records from token-paginated listings, omitting pagination metadata automatically. Pass a listing function and its arguments, for example db_list_all_pages(db_cluster_list, page_size = 100) (#301, @zacdav-db).
• Cluster wrappers now use Clusters API 2.1. Breaking change: db_cluster_list() and db_cluster_events() return full single-page responses, preserving pagination tokens; extract $clusters or $events for records. Both accept page_size and page_token. The Connections pane follows all cluster pages (#301, @zacdav-db).
• db_cluster_events() deprecates offset/limit in favor of page_token/page_size ahead of Databricks' November 30, 2026 migration deadline. Legacy arguments default to NULL; explicitly supplied values still work with a warning but cannot be mixed with token pagination. Obtain tokens from preceding responses rather than converting numeric offsets. Millisecond timestamps are now preserved without integer overflow (#301, @zacdav-db).
• cumprod() now uses built-in Databricks SQL functions instead of requiring sparklyr's session-registered aggregate. Cumulative products use a row frame and propagate NULL values to subsequent rows in each group (#295, @zacdav-db).
• Breaking change: db_jobs_list() and db_jobs_runs_list() now return the full single-page API response, including pagination tokens and metadata. Extract $jobs or $runs to access the records. offset now defaults to NULL to enable token pagination; explicit numeric offsets are still forwarded. Use page_token from the preceding response to fetch another page. Job and run get wrappers also accept page_token for additional array elements. Use named arguments with these four wrappers because their argument order has changed (#282, @zacdav).
• Fixed the workspace connection pane's Catalog availability check so accessible Unity Catalog resources appear (#283, @zacdav).
• The RStudio connection pane now follows all Unity Catalog listing pages for catalogs, schemas, tables, volumes, models, functions, and model versions, including empty pages with continuation tokens. Volume details are fetched by name so volumes beyond the first listing page can be inspected. db_uc_volumes_get() now sends include_browse as a query parameter, and the pane handles optional metadata omitted from browse-only responses (#283, @zacdav).
• Volume-based dbWriteTable() now creates the schema requested by field.types, including CHAR/VARCHAR length limits. Volume writes and appends use INSERT to convert staged values to the target column types (#287, @zacdav).
• db_volume_list() now accepts page_size and page_token. Directory downloads and recursive deletion follow every listing page, preventing files beyond the first page from being omitted. Listing failures now stop recursive deletion instead of being treated as empty directories (#281, @zacdav).
• db_volume_upload_dir() now creates each parent directory only once, avoiding repeated API calls for files in the same directory (#269, @zacdav-db).
• dbWriteTable() and dbAppendTable() now write POSIXct values as epoch microseconds without shifting their time zone, using the same microsecond conversion as volume-based writes (#298, @m-muecke).
• Standard-path table writes now detect binary columns once per column, avoiding quadratic serialization time for lists of raw vectors (#268, @zacdav-db).
• Volume file requests now encode reserved characters and literal percent sequences in paths correctly, preventing # and ? in filenames from changing the request target (#267, @zacdav-db).
• db_volume_dir_delete() now respects perform_request = FALSE when recursive = TRUE, returning a directory-delete request without listing or deleting contents (#266, @zacdav-db).
• db_cluster_create() and db_cluster_edit() no longer ignore log_conf (#299, @m-muecke).
• db_vs_indexes_upsert_data() no longer rounds numbers to four decimal places (#308, @m-muecke).
• Volume-backed dbWriteTable() and dbAppendTable() avoid an extra staging-directory creation request before uploading files (#309, @m-muecke).
• Databricks CLI authentication now limits token retrieval to 60 seconds instead of waiting up to 60,000 seconds (#296, @m-muecke).

brickster 0.2.14

• Fixed OAuth U2M and M2M authentication across multiple Databricks workspaces in one R session by isolating OAuth clients and cached tokens per workspace (#256, @zacdav-db)
• OAuth tokens are now refreshed 40 seconds before expiry, avoiding Databricks API rejections during the final 30 seconds of a token's lifetime; this requires {httr2} 1.3.0 or later (#258, @zacdav-db)
• Added databricks-cli authentication support for profiles created by databricks auth login; {brickster} obtains tokens through databricks auth token, caches only the short-lived access token in memory, and leaves durable credentials and refresh to the CLI (#258, @zacdav-db)
• Authentication mode overrides now come from auth_type in the selected .databrickscfg profile; replace DATABRICKS_AUTH_TYPE with the profile field. Environment-only authentication continues to infer the mode from available credentials (#258, @zacdav-db)
• Fixed dbWriteTable() and dbAppendTable() standard-path writes for binary columns, which now use Databricks BINARY types and X'...' literals when no staging volume is configured (#246, @zacdav-db)
• dbConnect() now preserves Databricks API error details when its validation query fails (#247, @zacdav)
• db_perform_request() and db_perform_response() now preserve useful error messages when Databricks returns non-JSON error bodies such as empty, plain-text, or HTML responses (#247, @zacdav)
• Fixed git_source() erroring when type was left at its default (#225, @m-muecke)
• Fixed Unity Catalog volume file requests so db_volume_* paths containing spaces are encoded correctly (#233, @zacdav-db)
• db_cluster_events() now forwards the event_types argument to the API, which was previously ignored (#236, @m-muecke)
• db_sql_warehouse_create() and db_sql_warehouse_edit() now forward the tags argument to the API, which was previously ignored (#240, @m-muecke)
• Unity Catalog list helpers now preserve next_page_token metadata in list responses for catalogs, schemas, tables, and volumes; their list arguments are sent as documented query parameters (#244, @zacdav-db)
• db_uc_volumes_list() now forwards the max_results, include_browse, and page_token arguments to the API, which were previously ignored (#228, @m-muecke)
• db_vs_indexes_query() now sends the score_threshold argument to the API, which was previously ignored (#235, @m-muecke)
• Added show_progress to dbConnect() for the DBI backend; dbGetQuery(), dbFetch(), dbWriteTable(), and dbplyr collect() now use the connection default while preserving per-call show_progress overrides (#223, @zacdav-db)

brickster 0.2.13

• Enabled db_request() retries for transient low-level HTTP request failures, improving resilience to intermittent curl/HTTP2 framing errors (#215)
• Added a DBI connection-level disposition setting so dbSendQuery() and default dbGetQuery() calls can use INLINE results when direct cloud-storage downloads are blocked (#205)
• Added Azure AD service principal OAuth M2M support (ARM_CLIENT_ID, ARM_CLIENT_SECRET, ARM_TENANT_ID) with optional DATABRICKS_AUTH_TYPE override (oauth-m2m, azure-client-secret, oauth-u2m); default auth resolution now prefers Azure M2M over U2M when ARM credentials are present (#185)
• Marked DBFS REST wrappers (db_dbfs_*) as deprecated and moved them to internal-only, guiding users towards using volumes (db_volume_*)
• Added db_volume_download_dir() for parallel directory downloads from Unity Catalog volumes to local directories
• Renamed preserve_structure to recursive in db_volume_upload_dir() and db_volume_download_dir() for consistent directory traversal semantics
• Added a new vignette for working with volumes
• Added compact, colorized {cli} S3 print methods for db_cluster_get()/db_cluster_list(), db_sql_warehouse_get()/db_sql_warehouse_list(), and db_jobs_get()/db_jobs_list() that preserve nested list structures while improving at-a-glance summaries
• Standardized DBI write progress argument naming: dbWriteTable() and internal db_write_table_volume() now use show_progress (#201)

brickster 0.2.12

• Fixed dbplyr::compute interaction for DBI backend (#184)
• Added OAuth M2M support for workspace-level service principal authentication (#173, #103)
• DBI connections accept http_path as an alternative to warehouse_id and extract the warehouse ID from /sql/1.0/warehouses/<id>.
• Added RStudio/Positron Connections Pane support for the DBI backend.
• Added DBI helpers for dbCreateTable(), dbReadTable(), and dbRemoveTable() with Id/AsIs support, plus offline tests for DBI table helpers.
• Optimized SQL result fetching for single-chunk external links by using the inline link from the initial response when available.
• Increment version of testthat required (>= 3.3.0)

brickster 0.2.11

• Added Lakebase workspace database helpers (db_lakebase_*) including credential generation, instance listing, instance lookup by name or UID, and catalog retrieval (#113)
• Moving all filesystem related calls to {fs} (#140)
• The DBI backend now always respects a staging volume when specified, even for small data (#143)
• schemaEvolutionMode is now always none when writing to tables with DBI backend and staging volumes (#147)

brickster 0.2.10

• Increment version of httr2 required (>= 1.1.1)
• DBI connections expose max_active_connections and fetch_timeout to control result download concurrency and timeouts
• DBI/dbplyr write table methods now make two transactions (create empty table --> insert into) to ensure type correctness
• Allow optional schedules in db_jobs_reset() and propagate parameters in reset/update requests.
• DBI/dbplyr inline writes now preserve single quotes in character columns via explicit escaping (#130)

brickster 0.2.9

• Added DBI + dbplyr backend support: DatabricksSQL() driver for standard DBI operations
• Increase support for job level parameters
• Added db_jobs_repair_run

brickster 0.2.8

• Added SQL Queries API coverage
• Updated Jobs to 2.2
• Added additional tasks for jobs: for_each_task, condition_task, sql_query_task, sql_file_task, run_job_task
• Removing the Python SQL connector as db_sql_query supersedes it.
• Added db_sql_query to simplify execution of SQL
• Adjusted db_repl to handle mulit-line expressions (R only)
• Removed RStudio Addins to send lines/selection/files to console
• Moved arrow to Suggests

brickster 0.2.7

• Exporting UC table functions (db_uc_table*) (#72)
• Adding support for direct_download option in db_workspace_export()
• Exporting UC Catalog/Schema get/list functions (#72)
• Adding support for UC Volume management (#72)
• Fixing command execution context cancel (#86, #87)
• Adding stricter version requirements for {httr2} (#81, #63)

brickster 0.2.6

• Fixing db_volume_delete() function (#73, @vladimirobucina)
• Adjustments to ensure {httr2} changes don't break things (#75, @hadley)

brickster 0.2.5

• Adding db_repl() a remote REPL to a Databricks cluster (#53)
• Removing defunct RStudio add-in for browsing Databricks compute
• Changes to DESCRIPTION file in preperation for CRAN (#64)
• Removal of notebook_use_posit_repo() and notebook_enable_htmlwidgets() as they are incompatible with CRAN (#64)
• Removing kntir engine due to many render edge cases not being solvable
• Adding shortcuts for REPL under addins
• Added db_context_command_run_and_wait
• Adjusted tests to use withr (#68)

brickster 0.2.4

• open_workspace() and the rstudio connection pane have been heavily revised to enhance browsing unity catalog and also remove DBFS and WSFS browsing (#52)

brickster 0.2.3

• Adding NEWS.md
• Renamed set_lib_path to add_lib_path and added the after parameter
• Adding OAuth U2M support (workspace level), considered the default when DATABRICKS_TOKEN isn't specified (e.g db_token() returns NULL)
• Updating authentication vignette to include information on OAuth
• Updating README.md to include quick start and clearer information
• Adding vector search index functions