rstudio
DuckDB tables throw error when passed to `create_agent()`
Prework
- [x] Read and agree to the code of conduct and contributing guidelines.
- [x] If there is already a relevant issue, whether open or closed, comment on the existing thread instead of posting a new issue.
Description
DuckDB-backed tables cannot be passed to create_agent() - this emits an ! argument is of length zero error. This manifests on MacOS, Windows and Ubuntu. Bizarrely, it only happens when running code line by line. When executing a script, it does not happen.
The same issue does not manifest with SQLite. It does also show up when the DuckDB table is a view representing a registered Arrow dataset. It does not manifest when the tbl() call is put directly inside the create_informant() call (see reprex below).
The cause is in pointblank:::get_tbl_dbi_src_info() which does not capture the DB signature because it begins with newlines. It is not clear why in some situations (line by line execution) the newlines are there and not in others.
More broadly, could create_informant() rely on the class of the object to detect its type? e.g.
class(tbl_db)
> [1] "tbl_duckdb_connection" "tbl_dbi" "tbl_sql" "tbl_lazy"
> [5] "tbl"
Reproducible example
- [x] Post a minimal reproducible example so the maintainer can troubleshoot the problems you identify. A reproducible example is:
- [x] Runnable: post enough R code and data so any onlooker can create the error on their own computer.
- [x] Minimal: reduce runtime wherever possible and remove complicated details that are irrelevant to the issue at hand.
- [x] Readable: format your code according to the tidyverse style guide.
library(pointblank)
library(duckdb)
library(dplyr)
con <- dbConnect(duckdb::duckdb(), "test.duckdb")
tbl_db <- copy_to(con, airquality,
"airquality_cp", temporary = TRUE, overwrite = TRUE)
tbl_db2 <- tbl(con, "airquality_cp")
create_agent(tbl(con, "airquality_cp")) # runs in any case
create_agent(tbl_db) # error if run line by line
create_agent(tbl_db2) # error if run line by line
tbl_db$src$con
pointblank:::get_tbl_dbi_src_details(tbl_db) # returns empty string "" if run line by line
dbDisconnect(con)
Expected result
I would expect this to work no matter the mode of execution.
Session info
> sessionInfo()
R version 4.4.1 (2024-06-14)
Platform: aarch64-apple-darwin20
Running under: macOS 15.5
Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
locale:
[1] en_US.UTF-8/UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
time zone: Europe/Prague
tzcode source: internal
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] dplyr_1.1.4 duckdb_1.3.0 DBI_1.2.3 pointblank_0.12.2.9000 reprex_2.1.1
[6] devtools_2.4.5 usethis_3.1.0
loaded via a namespace (and not attached):
[1] gt_1.0.0 sass_0.4.10 rappdirs_0.3.3 generics_0.1.4 xml2_1.3.8
[6] stringi_1.8.7 digest_0.6.37 magrittr_2.0.3 pkgload_1.4.0.9000 fastmap_1.2.0
[11] pkgbuild_1.4.8 sessioninfo_1.2.2 urlchecker_1.0.1 promises_1.3.2 purrr_1.1.0
[16] coro_1.1.0 httr2_1.1.2 cli_3.6.5.9000 shiny_1.10.0 rlang_1.1.6
[21] dbplyr_2.5.0 litedown_0.7 commonmark_2.0.0 ellipsis_0.3.2 remotes_2.5.0
[26] withr_3.0.2 cachem_1.1.0 tools_4.4.1 memoise_2.0.1 nanonext_1.6.0
[31] httpuv_1.6.15 vctrs_0.6.5 R6_2.6.1 mime_0.13 lifecycle_1.0.4
[36] stringr_1.5.1 mcptools_0.0.0.9000 fs_1.6.6 htmlwidgets_1.6.4 miniUI_0.1.1.1
[41] pkgconfig_2.0.3 pillar_1.11.0 later_1.4.1 glue_1.8.0 profvis_0.3.8
[46] Rcpp_1.1.0 xfun_0.52 tibble_3.3.0 tidyselect_1.2.1 xtable_1.8-4
[51] htmltools_0.5.8.1 ellmer_0.2.1 compiler_4.4.1 blastula_0.3.6 S7_0.2.0
[56] markdown_2.0
the solution is to generalize is_tbl_mssql() to DuckDB and other backends and detect by class
I implemented a bug fix here: https://github.com/rstudio/pointblank/pull/650
