### Context We maintain a downstream fork of QField tailored for field vegetation surveys in Poland (WorkField), so first of all: thank you for the new in-app NTRIP client — it removes a whole layer of third-party mock-location tooling for RTK workflows and has been working nicely in our tests. ### Current behaviour Looking at src/core/positioning/ntripclient.cpp and positioningsource.cpp , when the caster connection drops or errors out, NtripClient emits errorOccurred() / streamDisconnected() and PositioningSource records the state — but no reconnection is attempted. The client is only (re)started when the receiver connection state changes, so after a transient network hiccup the user has to toggle NTRIP manually. ### Why it matters in the field On cellular connections in rural areas, short drops (handover, coverage gaps) are routine. Each one currently silently degrades the fix from RTK back to autonomous until the surveyor notices and reconnects by hand — easy to miss while collecting data. ### Suggested behaviour Automatic retry with growing backoff (e.g. 1 s doubling up to ~30 s), gated by the isPermanent flag that errorOccurred(message, isPermanent) already carries — permanent errors (bad credentials, unknown mountpoint) would stop retrying and surface to the user, transient ones would retry quietly. We would be happy to prepare a PR along these lines if the approach sounds right to you.