Known issues

These are the problems we know about in DataFlow 1.0.1, the version on the download page. Each says what happens, who it affects and what to do until it's fixed. Three are already fixed for the next version, and the changelog lists what each version changes.

Found something that isn't here? Write to hello@letdataflow.com with what you did, what happened, and your DataFlow version and system (Settings → Updates shows the version). From the next version, Settings → About → Send feedback opens that email with the version and system filled in.

Installing

macOS asks you to confirm DataFlow the first time

What happens. DataFlow for Mac isn't notarized by Apple yet. The first time you open it, macOS says it can't check it for malicious software and doesn't open it.

Who it affects. Everyone installing DataFlow on a Mac, once. Windows and Linux aren't affected.

What to do. Open it once with Open Anyway in System Settings → Privacy & Security; the download page has the steps. After that it opens like any other app.

Status. Developer ID signing is coming, and this step goes away with it.

Connecting to your server

SurveyCTO: interviews held for review arrive only once approved

What happens. On a form that holds submissions for review, DataFlow 1.0.1 downloads only approved submissions. Interviews waiting for review aren't checked until someone approves them, so problems in them are found later than they could be.

Who it affects. SurveyCTO forms that use the review workflow. Forms that don't hold submissions aren't affected.

What to do. Review held submissions promptly, so they reach DataFlow soon after they arrive, or hold only some submissions for review, so the rest arrive straight away.

Status. Fixed for the next version: DataFlow asks for submissions waiting for review too. Rejected ones stay out.

Edits made on the server don't come through

What happens. DataFlow downloads each interview once. If it's changed on the server after that, by a correction in SurveyCTO or an edit in ODK Central, DataFlow keeps the version it first downloaded and goes on checking that one.

Who it affects. Teams that correct interviews on the survey server while DataFlow is syncing them.

What to do. Make corrections in DataFlow instead. Each one is logged with who made it, when and why, can be undone, and is replayed by the do-file in every download, while the raw data stays as it arrived.

Status. A change for a later version; it isn't scheduled yet.

KoboToolbox: no direct connection

What happens. DataFlow connects to SurveyLoopr, SurveyCTO and ODK Central, but not to KoboToolbox, so Kobo interviews don't sync on their own.

Who it affects. Teams collecting with KoboToolbox.

What to do. Download Kobo's XLS export and import it, and import each new export the same way: DataFlow matches interviews across imports by _uuid. In Kobo's export settings, keep Include groups in headers off, so headers stay as question names. With it on, a header such as household/hhsize isn't matched to the form's hhsize. Rosters on their own sheets aren't read yet (see Excel below). The KoboToolbox guide has the rest of the settings.

Importing files

Excel imports read the first sheet only

What happens. DataFlow 1.0.1 reads the first sheet of an Excel file and leaves the rest. Repeat groups exported on sheets of their own, such as household rosters in KoboToolbox's XLS export or SurveyCTO's long format, aren't imported or checked.

Who it affects. Anyone importing an Excel file whose rosters sit on their own sheets. The main sheet is checked as usual.

What to do. Until the next version, check the roster sheets with your own scripts.

Status. Fixed for the next version: rosters on their own sheets are read and put back with their interview.

CSV imports expect commas

What happens. DataFlow 1.0.1 reads CSV files with commas between the columns. A file separated by semicolons or tabs isn't read correctly.

Who it affects. Anyone importing a CSV file separated by semicolons or tabs, as some tools and regional settings produce.

What to do. Convert the file to comma-separated before importing it, or import the tool's Excel export instead.

Status. Fixed for the next version: DataFlow uses whichever of comma, semicolon or tab the header row uses.

Scheduled runs

On Linux, schedules run only while DataFlow is open

What happens. On macOS and Windows, the system starts DataFlow for each scheduled sync while it's closed (launchd on macOS, Task Scheduler on Windows). On Linux nothing starts it, so a schedule runs only while DataFlow is open.

Who it affects. Linux users who turn on Sync on a schedule.

What to do. Leave DataFlow open for as long as the schedule should run.