Changelog • csdb

Bug Fixes

PostgreSQL methods (create_table, keep_rows_where, drop_table) now quote role_create_table via DBI::dbQuoteIdentifier() when emitting SET ROLE. Previously the role name was interpolated raw, which broke on identifiers containing hyphens, mixed case, or reserved words (e.g. SET ROLE token-user -> syntax error), and was a SQL-injection vector if the value came from an env var.

New Features

Added authentication hook system (csdb_set_auth_hook(), csdb_get_auth_hook()) to allow automatic credential refresh (e.g., Kerberos tickets) when connection fails

Bug Fixes

Improved database connection error messages to include the original driver error details for easier debugging

Bug Fixes

Fixed namespace loading error by importing methods::initialize generic
Improved error handling in .onLoad() function for S7 operations
Package now passes CRAN namespace loading requirements

Bug Fixes

Fixed DBConnection_v9 print method to display “SSL mode” instead of “trusted connection” for PostgreSQL connections

Development

Added S7 package to imports in preparation for S3 to S7 method conversion to improve CRAN compliance
S7 transition planned to reduce exported API surface while maintaining functionality through R6 classes

Updated package for CRAN submission with comprehensive improvements
Added comprehensive documentation with examples for all exported functions
Fixed critical CRAN compliance issues including system tool availability checks
Added proper R6 class documentation with detailed usage examples
Improved all validator function documentation with clear examples
Added missing dependencies and fixed import declarations
Updated .Rbuildignore to exclude system files and build artifacts
Added CLAUDE.md for future development guidance
Fixed vignette title and improved documentation quality
All functions now pass R CMD check with only acceptable NOTEs

Including role_create_table in drop_table for PostgreSQL.

role_create_table is now included for dbconnection_v9/dbtable_v9, so that the role can be changed when creating tables in PostgreSQL.

csdb now supports PostgreSQL databases as well as MS SQL Server.

Including use_count as an argument in nrow in DBTable_v9, which is slower but more accurate.

Including confirm_insert_via_nrow in DBTable_v9. Checks nrow() before insert and after insert. If nrow() has not increased sufficiently, then attempt an upsert.

Including validator_field_types_csfmt_rts_data_v2 and validator_field_contents_csfmt_rts_data_v2.

Including georegion in validator_field_contents_csfmt_rts_data_v1.

get_table_names_and_info is now ordered according to table_name.

get_table_names_and_nrow is now changed to get_table_names_and_info and also includes size_total_gb, size_data_gb, size_index_gb.
info is now included as a method for DBTable_v9

confirm_indexes is now added to DBTable_v9, which confirms that the names and number of indexes in the database are the same as in the R code. It does not confirm the contents of the indexes!
nrow is now added to DBTable_v9, which is an application of the new get_table_names_and_nrow function.
get_table_names_and_nrow added as an exported function, that will get all the table names and the nrows from a dbconnection.

create_table now automatically adds the indexes.

Removing info messages from drop_rows_where.

connect() in DBConnection_v9 is smarter, more robust with error checking and making fewer useless calls to the db. Tries to connect twice now before throwing an error.
autoconnection is now more robust in DBConnection_v9.

Package is created.