In my case, I've got 36 threads, and those lines cause each thread to download the full dataset, which then maxes out my available bandwidth, causing failure of threads and therefore the whole routine. (I can get around this by running fewer threads, but would prefer a proper solution to this.)
Edit: Actually, that whole call to checkrows(res.calendar.days) takes hardly any time even in single threaded, so maybe the easiest solution would be just removing the multi-core version there? Everything else then works fine.
This package is cleverly set up to download all of the release data each time it is loaded. But that means these lines can fail: https://github.com/ITSLeeds/UK2GTFS/blob/5094f1f4a447d65f2e14cea0b494c9c602192a24/R/atoc_export.R#L386-L388
In my case, I've got 36 threads, and those lines cause each thread to download the full dataset, which then maxes out my available bandwidth, causing failure of threads and therefore the whole routine. (I can get around this by running fewer threads, but would prefer a proper solution to this.)
Edit: Actually, that whole call to
checkrows(res.calendar.days)
takes hardly any time even in single threaded, so maybe the easiest solution would be just removing the multi-core version there? Everything else then works fine.