Changeset 10984 in ntrip


Ignore:
Timestamp:
Jul 31, 2026, 1:15:44 PM (8 hours ago)
Author:
stuerze
Message:

First read after startRequest/reconnect now gets 5 × _timeOut (100 s) instead of the strict 20 s, giving Windows' slow WPAD/proxy‑autodetect + TLS handshake room to finish.

File:
1 edited

Legend:

Unmodified
Added
Removed
  • trunk/BNC/src/bncnetqueryv2.cpp

    r10960 r10984  
    223223    timer.setSingleShot(true);
    224224    connect(&timer, SIGNAL(timeout()), _eventLoop, SLOT(quit()));
    225     timer.start(_timeOut);
     225    // The very first read of a request can take much longer than a
     226    // steady-state read: it includes connection setup, TLS handshake and
     227    // (on Windows) potentially slow system/WPAD proxy auto-detection. Only
     228    // apply the strict outage-detection timeout once the stream is live.
     229    timer.start(_firstData ? 5 * _timeOut : _timeOut);
    226230    _eventLoop->exec();
    227231    timedOut = !timer.isActive();
     
    253257  else {
    254258    outData.append(_reply->readAll());
     259    _firstData = false;
    255260  }
    256261}
Note: See TracChangeset for help on using the changeset viewer.