Inconsistent file paths returned from multi_download when resumed.

[h-a-graham]

Hello,

Firstly, thanks for this package it is hugely appreciated!

I recently noticed that when using curl::multi_download the destfile column of the returned data.frame object differs depending on whether the file is resumed on completed on first go. When a file is resumed, the destfile is absoloute but when it completes in one it is relative. See here for a reprex:

suppressPackageStartupMessages({
  library(curl)
  library(R.utils)
})

# Example: some large files
urls1 <- sprintf(
  "https://d37ci6vzurychx.cloudfront.net/trip-data/yellow_tripdata_2021-%02d.parquet", 1:2
)

# cancel download after 5 seconds
res1 <- try(withTimeout(multi_download(urls1, resume = TRUE), timeout = 5)) # You can interrupt (ESC) and resume
# now escape the download and resume
res1 <- multi_download(urls1, resume = TRUE)

urls2 <- sprintf(
  "https://d37ci6vzurychx.cloudfront.net/trip-data/yellow_tripdata_2021-%02d.parquet", 3:4
)
# Do not escape this.
res2 <- multi_download(urls2, resume = FALSE) # You can interrupt (ESC) and resume

# full filepaths returned when resumed
print(res1$destfile)
#> [1] "/tmp/RtmpbPZ3Zs/reprex-146af8714e091d-awake-topi/yellow_tripdata_2021-01.parquet"
#> [2] "/tmp/RtmpbPZ3Zs/reprex-146af8714e091d-awake-topi/yellow_tripdata_2021-02.parquet"
# relative filepaths returned when completed in one go.
print(res2$destfile)
#> [1] "yellow_tripdata_2021-03.parquet" "yellow_tripdata_2021-04.parquet"

^{Created on 2024-09-12 with reprex v2.1.1}

Is the solution perhaps as simple as setting destfiles to normalizePath(destfiles, mustWork = FALSE ) in:
https://github.com/jeroen/curl/blob/master/R/multi_download.R#L186

Cheers,
Hugh