Fetch results for previously submitted tasks

Retrieves results for all tasks in data that are still in the PENDING state. Newly received results overwrite the corresponding rows' .task_state and .task_result values, and are unnested into regular columns.

Usage

da_fetch(data, wait = 10, poll = FALSE, interval = 5, timeout = Inf)

Arguments

data

A data frame previously produced by da_submit(). Must contain a .task_id column.

wait

Integer. Seconds to wait server-side per request for the task to complete before returning. Defaults to 10.

poll

Logical. If FALSE (default), make a single pass over pending tasks. If TRUE, repeat until no task is pending, timeout is reached, or the user interrupts.

interval

Numeric. Seconds to pause between polling rounds. Only used when poll = TRUE. Defaults to 5.

timeout

Numeric. Maximum total polling time in seconds. Defaults to Inf (poll until done or interrupted).

Value

The input data frame with updated .task_state values and unnested result columns. If polling was interrupted or timed out, the results received so far are returned.

Details

With poll = TRUE, the function keeps fetching in rounds until no task is pending, the timeout is reached, or the user interrupts. After each round it prints a summary of task states and counts down to the next round.

Interrupting

Press Esc (RStudio, Positron) or Ctrl+C (terminal) at any time to stop polling. The interrupt is caught, so the results collected so far are unnested and returned normally, e.g. res <- da_fetch(df, poll = TRUE) still assigns res. Call da_fetch() again on the result to resume.

Examples

if (FALSE) { # \dontrun{
res <- df |>
  da_submit(text, rules) |>
  da_fetch(poll = TRUE)
} # }