Esri has been adding more options for handling dates. In geodatabases, they now offer new data types for “Date Only” “Time Only” and “Datetime with timezeone offset” for data in geodatabases.
On feature services in ArcGIS Online and ArcGIS Enterprise, publishers can now define the timezone of the data underlying the service, and the timezone for display to clients. This allows service publishers to define how they prefer dates from the service should be displayed, and how dates in data being written to the service should be translated for storage in the underlying dataset. However, from my rough testing, the Rest services are still sending dates and UNIX values, so the timezone definition on the service is just there so clients (like ArcGIS Pro, ArcGIS Enterprise, etc) know what to do with the unix values on the client side before displaying or writing back.
Plus from what I saw at the 2025 user conference, they are adding more datetime configuration options to Pro in future releases.
It would be helpful if the new feature service reader could tap into these settings and control how data from date attributes gets pulled into the workspace initially. It’s just one less thing to have to translate as data comes into the workspace when the creator of the service already defined how they would prefer users to interact with the dates in that service.
On the reader, I could see this as a parameter that has options like
DateTime Output Format
Unix: Values from Date columns are brought in as the UNIX values from Esri.
FME UTC: Values from date columns are brought in with FME Datetime format and timezones not translated
FME with Timezone from Service: Values from date columns are brought in with FME Datetime format but with the timezone added.
On the writer, I could see a parameter to control how FME date attributes can be translated on writing to the service so they honor the “Time zone of the data” setting.
I personally just started looking at migrating from the old ArcGIS Portal reader to the new ArcGIS Feature Service reader. With the old reader, date info was automatically translated into FME datetime format. With the new reader, date columns initially load in Unix format. That means I’ve now got to do a datetime conversion on data from any reader if I want to work with them on the FME workbench. But maybe I’m missing something on how best to work with this kind of data in the workbench. I haven’t found anything yet about this change in any of the documentation or blog articles. I only found out about it after submitting a ticket. So if anyone has more detail on whether the change is intentional, or how do deal with it now that it’s in the new version, please let me know.