Merge branch 'main' into add_date_functions

Author: krlmlr

.github/copilot-instructions.md

@@ -1,3 +1,22 @@
-# GitHub Copilot Instructions for dm Package
+# GitHub Copilot Instructions for duckdb Package
 
 Read and follow the development guidelines outlined in [CLAUDE.md](../CLAUDE.md).
+
+## Working Effectively
+
+### Bootstrap, Build, and Test the Repository
+
+- `sudo apt-get install -y r-base r-base-dev build-essential` -- installs R and development tools
+- `mkdir -p ~/R/library && echo '.libPaths("~/R/library")' > ~/.Rprofile` -- sets up local R library
+- `export MAKEFLAGS="-j$(nproc)"` -- enables parallel compilation
+- `UserNM=true R CMD INSTALL . --no-byte-compile` -- builds and installs the package. NEVER CANCEL: takes 10-15 minutes on first build. Set timeout to 30+ minutes.
+
+### Run Tests
+
+- `R -q -e "testthat::test_local()"` -- runs all tests. Takes about 45 seconds. NEVER CANCEL: set timeout to 5+ minutes.
+- `R -q -e "testthat::test_local(filter = '^name$')"` -- runs specific test file by name
+
+### Format Checking (Optional - Has Known Issues)
+
+- `pip3 install cmake-format && export PATH="$HOME/.local/bin:$PATH"` -- installs formatter
+- `make format-check` -- checks code formatting

CLAUDE.md

@@ -2,14 +2,65 @@
 
 R package that contains a vendored copy of the DuckDB C++ library and glue code for R, including a DBI and a relational interface.
 
-## Package layout
+## Working Effectively
 
-- `R/`: R scripts and package code
-- `src/*.cpp`: C++ glue
-- `src/duckdb/`: C++ source code for DuckDB, can be reviewed but only modified in rare cases
-- `tests/`: Unit tests for the package
+### Bootstrap, Build, and Test the Repository
 
-## Compilation
+- `sudo apt-get install -y r-base r-base-dev build-essential` -- installs R and development tools
+- `mkdir -p ~/R/library && echo '.libPaths("~/R/library")' > ~/.Rprofile` -- sets up local R library
+- `export MAKEFLAGS="-j$(nproc)"` -- enables parallel compilation
+- `UserNM=true R CMD INSTALL . --no-byte-compile` -- builds and installs the package. NEVER CANCEL: takes 10-15 minutes on first build. Set timeout to 30+ minutes.
+
+### Run Tests
+
+- `R -q -e "testthat::test_local()"` -- runs all tests. Takes about 45 seconds. NEVER CANCEL: set timeout to 5+ minutes.
+- `R -q -e "testthat::test_local(filter = '^name$')"` -- runs specific test file by name
+
+### Manual Validation
+
+- ALWAYS test basic DuckDB functionality after making changes by running a complete scenario
+- Test connection, table creation, data insertion, and querying:
+
+    ```r
+    library(duckdb)
+    con <- dbConnect(duckdb())
+    dbExecute(con, "CREATE TABLE test (id INTEGER, name VARCHAR)")
+    dbExecute(con, "INSERT INTO test VALUES (1, 'Alice'), (2, 'Bob')")
+    result <- dbGetQuery(con, "SELECT * FROM test ORDER BY id")
+    print(result)
+    dbDisconnect(con, shutdown=TRUE)
+    ```
+
+### Format Checking (Optional - Has Known Issues)
+
+- `pip3 install cmake-format && export PATH="$HOME/.local/bin:$PATH"` -- installs formatter
+- `make format-check` -- checks code formatting (currently shows formatting differences)
+
+## Validation
+
+- ALWAYS run through a complete end-to-end scenario after making changes to ensure DuckDB R package functionality works correctly.
+- The package can be built and tested successfully, though some formatting issues exist in the current codebase.
+- One test failure is expected in clock function tests (unrelated to core functionality).
+- Format checking will show differences but should not block development.
+
+## Repository Structure and Key Locations
+
+- `R/`: R source code (17 files) - DBI interface, connection handling, result processing
+- `src/*.cpp`: C++ glue code (~30 files) - R to DuckDB interface
+- `src/duckdb/`: Vendored DuckDB C++ source code (~1700 C++ files, ~1400 headers) - DO NOT modify except in rare cases
+- `tests/testthat/`: Unit tests (~40 test files) - comprehensive test coverage
+- `scripts/`: Build and maintenance scripts - vendor.sh, format.py, setup-makeflags.R
+- `configure`/`configure.win`: Build configuration scripts
+- `DESCRIPTION`: R package metadata and dependencies
+- `README.md`: Main documentation with build instructions
+- `CLAUDE.md`: Operational instructions for AI
+- `.github/workflows/`: CI/CD workflows for testing on multiple platforms
+
+## Common Tasks
+
+The following are validated commands and their typical execution times:
+
+### Building
 
 ```bash
 # From the duckdb-r directory
@@ -42,3 +93,25 @@ R
 - Test files located in `tests/testthat/`
 - Use `testthat::test_local(filter = "name")` for running specific test files
 - Always add tests when fixing bugs to prevent regression
+
+## Code Style Guidelines
+
+- All files must end with an end-of-line (EOL) character
+- Ensure proper code formatting and consistent indentation
+- Follow R package development best practices
+
+## Dependencies
+
+System requirements already satisfied in typical development environment:
+- R >= 4.1.0
+- build-essential (gcc, g++, make)
+- Standard R packages: DBI, testthat, methods, utils
+- Optional: cmake-format for code formatting
+
+## Package Information
+
+- **Package Type**: R package providing DBI interface to DuckDB
+- **Core Functionality**: In-process SQL OLAP database for R
+- **Installation Time**: Up to 60 minutes from source (mentioned in README)
+- **Build Architecture**: C++ database engine with R bindings
+- **Key Features**: DBI compliance, Arrow integration, analytical query performance

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: duckdb
 Title: DBI Package for the DuckDB Database Management System
-Version: 1.3.2.9020
+Version: 1.3.2.9021
 Authors@R: c(
     person("Hannes", "Mühleisen", , "[email protected]", role = "aut",
            comment = c(ORCID = "0000-0001-8552-0029")),

NEWS.md

@@ -1,5 +1,34 @@
 <!-- NEWS.md is maintained by https://fledge.cynkra.com, contributors should not edit this file -->
 
+# duckdb 1.3.2.9021
+
+## Bug fixes
+
+- Fix timezone conversion for invalid timestamps with `tz_out_convert = "force"`.
+
+- Substitute invalid UTF-8 characters in error messages to avoid a failure when reporting the error.
+
+- Fix index calculation for retrieval of arrays (#1473).
+
+## Features
+
+- `dbGetInfo()` gets the version from a hard-coded value and not from a DuckDB query.
+
+- Package uses two cores by default for compilation.
+
+## Chore
+
+- Add AI instructions.
+
+- Build-ignore.
+
+- Add Claude instructions.
+
+## Testing
+
+- Add `local_con()` test fixture for cleaner DuckDB connection management.
+
+
 # duckdb 1.3.2.9020
 
 ## Continuous integration

R/Result.R

@@ -119,5 +119,5 @@ tz_force_one <- function(x, timezone) {
   # convert to character in ISO format, stripping the timezone
   ct <- format(x, format = "%Y-%m-%d %H:%M:%OS", usetz = FALSE)
   # recreate the POSIXct with specified timezone
-  as.POSIXct(ct, tz = timezone)
+  as.POSIXct(ct, format = "%Y-%m-%d %H:%M:%OS", tz = timezone)
 }

R/dbGetInfo__duckdb_connection.R

@@ -3,7 +3,8 @@
 #' @param dbObj An object inheriting from class [duckdb_connection-class].
 #' @usage NULL
 dbGetInfo__duckdb_connection <- function(dbObj, ...) {
-  version <- dbGetQuery(dbObj, "select library_version from pragma_version()")[[1]][[1]]
+  # Use hard-coded version instead of querying database to avoid establishing connection
+  version <- get_duckdb_version()
 
   list(
     dbname = dbObj@driver@dbdir,

R/rethrow.R

@@ -6,5 +6,13 @@ rethrow_error_from_rapi <- function(e, call) {
   # the caller.
 
   msg <- conditionMessage(e)
+  tryCatch(
+    # Called for side effect, rlang::abort() eventually calls nchar()
+    nchar(msg),
+    error = function(e) {
+      msg <<- iconv(msg, from = "UTF-8", to = "UTF-8", sub = "?")
+    }
+  )
+
   rlang::abort(msg, call = call)
 }

R/test-fixtures.R

@@ -0,0 +1,25 @@
+#' Local DuckDB connection fixture for testing
+#'
+#' Creates a DuckDB connection that will be automatically disconnected
+#' when the calling function exits. This is a test fixture that follows
+#' testthat guidelines for resource management.
+#'
+#' @param ... Additional arguments passed to [duckdb()]
+#'
+#' @return A DuckDB connection object
+#' @noRd
+#'
+#' @examples
+#' \dontrun{
+#' # In a test
+#' test_that("my test", {
+#'   con <- local_con()
+#'   # Use the connection...
+#'   # It will be automatically disconnected when the test exits
+#' })
+#' }
+local_con <- function(..., drv = duckdb()) {
+  con <- dbConnect(drv, ...)
+  withr::defer_parent(dbDisconnect(con, shutdown = TRUE))
+  con
+}

R/version.R

@@ -0,0 +1,9 @@
+# Generated by rconfigure.py, do not edit by hand
+# DuckDB version information
+
+duckdb_version <- "1.3.3-dev231"
+
+# Function to get DuckDB version without establishing a connection
+get_duckdb_version <- function() {
+  duckdb_version
+}

configure

@@ -2,6 +2,19 @@
 
 set -ex
 
+# Setup parallel compilation if MAKEFLAGS not already set
+cd "$(dirname "$0")"
+if [ -z "${MAKEFLAGS}" ]; then
+  makeflags_value=$(Rscript scripts/setup-makeflags.R 2>/dev/null || echo "")
+  if [ -n "${makeflags_value}" ]; then
+    export MAKEFLAGS="${makeflags_value}"
+    echo "Setting MAKEFLAGS=${makeflags_value} for parallel compilation"
+    echo "Set NOT_CRAN or MAKEFLAGS to override"
+  fi
+else
+  echo "MAKEFLAGS already set to: ${MAKEFLAGS}"
+fi
+
 cd "$(dirname "$0")/src"
 
 # Not used on Linux, for completeness.