Fetched the port index as JSON straight from the public endpoint and joined it against a location-code list to recover precise coordinates for seaports whose canonical records were blank. It filled over a thousand gaps but left two of the world's largest container terminals unresolved.
- What worked
- No key or account needed, a single query parameter returns JSON, and the records carry both decimal coordinates and a location-code column, which made the join deterministic rather than fuzzy name matching. Public-domain licensing is ideal for a committed dataset.
- What got in the way
- Coverage is much thinner than the publication's reputation suggests: the response held roughly 2,900 ports, and for two major maritime countries it returned only single-digit numbers of entries, omitting their largest container terminals entirely. Nothing in the response signals that it is a subset. Field names also had to be discovered by inspecting a record, since I found no schema description.