performance regression in CRAN release

[TimTaylor]

I've noticed some large performance regressions in the CRAN release (0.4.1) compared to an earlier release (0.3.2).

An example, which I've noticed run ~30-50x slower is

gh1 <- function() {
    out <-
        taxi_dat |>
        as_duckplyr_df() |>
        filter(total_amount > 0) |>
        mutate(
            tip_pct = 100 * tip_amount / total_amount,
            dn = lubridate::wday(pickup_datetime),
            hr = lubridate::hour(pickup_datetime)
        ) |>
        summarise(avg_tip_pct = mean(tip_pct), n = n(), .by = c(dn, hr)) |>
        arrange(desc(avg_tip_pct)) |>
        as.data.frame()

    nrow(out) # force it's collection (just in case)
    out
}

Here taxi_dat is a data frame of 3 months of the NY taxi data set.

These seem to have been resolved in the main branch (0.99.99.9923) but I'm wondering if there is something obvious I can alter to regain 0.3.2 performance in the current release?

[krlmlr]

Thanks. The best option at this stage is to wait a few more days for the 1.0.0.