<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=utf-8">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="postmsg">
<p>Hi,<br>
</p>
<p>we're currently trying to improve the time estimates for ferry
routes in GraphHopper. See <a
href="https://github.com/graphhopper/graphhopper/pull/519"
rel="nofollow" target="_blank"><a class="moz-txt-link-freetext" href="https://github.com/graphhopper/graphhopper/pull/519">https://github.com/graphhopper/graphhopper/pull/519</a></a></p>
<p>But there are several unrealistic times detected for a world
wide import: <a
href="https://gist.github.com/karussell/da39dcbea2b8b9cc9e06"
rel="nofollow" target="_blank"><a class="moz-txt-link-freetext" href="https://gist.github.com/karussell/da39dcbea2b8b9cc9e06">https://gist.github.com/karussell/da39dcbea2b8b9cc9e06</a></a></p>
<p>Now my question is: is the duration just the time the ship
takes from A to B or is an average waiting time included? We
guess the waiting time is not included. Can we update the wiki,
saying that the waiting time is not included?</p>
<p>Also we need to say there, that if the ferry is tagged as a
relation, then the duration tag has to be set as part of the
relation, not of the way. Because if two ferry relations share
the same way the durations 'collides'.</p>
<p>Additionally tagging ferries as way is simpler but we think
tagging them as relations is better. Especially for extremely
long routes the OSM ways should be avoided, but that is probably
just a different opinion. See the GraphHopper issue because of
this <img src="cid:part3.03060309.07000601@gmx.de" alt="wink"
height="15" width="15"><br>
<a href="https://github.com/graphhopper/graphhopper/issues/435"
rel="nofollow" target="_blank">https://github.com/graphhopper/graphhopper/issues/435</a><br>
</p>
</div>
Regards,<br>
Peter<br>
</body>
</html>