Dataverse mangles "original format" filename extensions

[donsizemore]

Only slightly related to #2720: Dataverse still renames certain file format extensions on upload/ingest. In our particular case, a .tab file is returned in original format as .tsv — verified on 4.14 and 4.15.

This breaks automated reproducibility when filenames are hard-coded in user-submitted scripts.

[oscardssmith]

As the culprit, this was intentional, since from everything I could find suggested that tsv was the much more common name for tab separated data. I understand the frustration though.

[djbrooke]

Thanks @donsizemore for the report. We'll change this.

[djbrooke]

We should add a solution that makes the reproducibility folks happy (this issue) and the preservation folks happy (#2720). We'll bring both into this sprint.

[djbrooke]

Closing in favor of newly-created #6006.