Closed perttolj closed 2 years ago
The route that CDISC has been looking has long been to get away from export files in general. Dataset.XML was a failed experiment because of inflated file sizes specifically but Dataset.JSON is a successor to that and an initiative that I'm hopeful for, because it's much less verbose than XML.
From an architecture standpoint, version 8 would be a simple switch with haven under the hood and some minor changes to the checks performed around the datasets, them flipping a parameter on the haven::write_xpt()
call.
Agreed with @mstackhouse. Right now V5 is hardcoded: https://github.com/atorus-research/xportr/blob/e5987358e4573276dd09d8652a22c614109e69bb/R/write.R#L55 If version 8 ever becomes supported it would be a quick update to that line. In the meantime, its not clear what they are going with in terms of data so I think best course of action is to wait until that is clear
In the presentation given here https://www.diaglobal.org/en/conference-listing/meetings/2022/04/dia-biostatistics-industry-and-regulator-forum/agenda/07/session-6-yes-you-can-use-r-in-regulatory-submissions?ref=Session6YesYouCanUseRinRegulatorySubmissions a person from FDA stated the following:
This gives hope that V8 would be in the future accepted instead of the very restrictive V5.
The presentation gives some further details as well:
It would be great to confirm this (possibly at a repeat of this presentation https://hopin.com/events/the-r-adoption-series-using-r-in-regulatory-review) and if confirmed, would be very helpful if xportr could optionally produce V8 xpt datasets as well.