CSV Import and Export Just Got More Reliable

Three CSV bugs, each one only showing up with a specific kind of file, are fixed today. If you ran into any of these, here’s what changed.

Importing a CSV saved on Windows

CSV files exported from Excel or edited on Windows end each line with an extra character your import might not have expected. That could leave a stray mark stuck to the last column of every row — enough to make a column you’d typed as an email address fail validation on every single row, or, for an untyped column, get written into your collection along with the rest of the value.

That’s fixed. A CSV exported from Windows now imports exactly like one saved anywhere else — the last column comes through clean, row by row.

An export that failed to upload for no visible reason

Exporting an instance occasionally failed at the very last step — after the export had already finished building — with no clear connection to anything in your data. That was an internal upload issue on our end, not a problem with your export. It’s fixed, and exports now upload reliably every time.

A wide CSV reporting itself as “empty”

The import wizard previews your file before you commit to it, so you can check your columns landed on the right fields. If your CSV had an unusually wide header row — a lot of columns — the preview could show your entirely valid, non-empty file as CSV file is empty, even though nothing was wrong with it.

That false error is gone. A wide header now previews correctly, the same as any other file.

The bottom line

All three fixes shipped with tests written against the exact file that triggered each bug, so if you hit any of these before, you shouldn’t again.