I agree that it’s very surprising this bug hasn’t been fixed, since it completely breaks one of the most important export features. (I noticed this bug shortly after the release of 2.32, but to my shame didn’t log it on Jira.) I’ve added my vote!
It looks like it’s just the user interface that’s broken, and that you can still do dependency exports via the API. Although you’re probably familiar with the API, I’ll set out full instructions below, in case others need to use this:
If you paste the following into your browser, it should download the appropriate dataset or program:
[your server URL]/api/dataSets/[UID of dataset]/metadata.xml?download=true
An example for the Play Demo site:
Within this URL, if you want to download a program instead of a dataset, just replace ‘dataSets’ with ‘programs’ ; if you want JSON instead of XML format, just replace ‘.xml’ with ‘.json’.
You can find the UID (“Id”) of a dataset or program by finding it within the list of datasets/programs in the ‘Maintenance’ app, and then selecting ‘Show details’ from the ‘Action’ menu on the right (three dots).