Kirill Müller comments

Results 2418 comments of


                                            Kirill Müller

Use `as.numeric(seq_len(n))` in favor of `seq(1, n)`

After reviewing, I think `as.numeric(seq_len(n))` is best.

Slow with `dm::dm_filter()`

Much simpler: It's not the duck tibbles inside the dm, it's dm using duckplyr. ``` r library(dm) Sys.setenv(DUCKPLYR_META_SKIP = TRUE) if (!file.exists("dm_local.qs")) { dm ℹ Restoring dplyr methods. dm_local |>...

Slow with `dm::dm_filter()`

Half of that is fallback collection, another quarter perhaps is `cli::cli_abort()` in `check_df_for_rel()` .

Slow with `dm::dm_filter()`

But also the version where dm uses dplyr and the tables are duck tibbles is slow in itself: ``` r library(dm) #> #> Attaching package: 'dm' #> The following object...

Slow with `dm::dm_filter()`

I'm getting wildly different timings for the second example, though. Probably okay?

`dm_filter()` does too much work

@claude: Give it a try.

duckdb fails to compile on RHEL7.9 on R 4.1 session.

Thanks. This is a fairly old OS. Can you please try duckdb 1.2.2 or older? https://cran.r-project.org/src/contrib/Archive/duckdb/

duckdb fails to compile on RHEL7.9 on R 4.1 session.

Thanks. Can you compile the upstream C++ library on your OS: https://github.com/duckdb/duckdb ?

R: graph.data.frame converts factors to character

Old implementation by @thomasp85: #193.

duckdb 1.4.0 Error: Assertion '__n < this->size()' failed.

Thanks. A DuckDB update is scheduled for October 6. I'll resume vendoring soon so that you can check with an updated version from GitHub or r-universe. If you're very brave,...