Retrospective development diary · Week beginning 2026-06-29. Written in September 2026 from the project history; this is not an original publication date or a release announcement.
At the beginning of July we started turning import into a workflow rather than a file picker. Choosing an export from another application should not immediately fill a vault with records that have been guessed incorrectly. The first useful step was to show what had been read and let the user decide what belonged where.
The first-pass CSV work introduced parsing and review before adding data. Column mapping made it possible to connect an unfamiliar export’s headings to the licence fields we understood. Individual rows could be selected instead of treating the entire file as an all-or-nothing operation.
This changed the emphasis of importing. Success was no longer just whether a file opened; it was whether the person importing could understand the proposed result. That became especially important for exports with gaps, unusual column names or several different kinds of information mixed together.
The work committed on 4 and 5 July was still a first pass. It established a place to review and correct data, while leaving plenty of real-world export formats to handle later. That review step would become one of the most useful foundations when the project expanded beyond software licences.