[Tagging] Implied oneway tag for highway=*_link, wiki edits

Tobias Knerr osm at tobias-knerr.de
Thu Nov 26 19:28:07 GMT 2009

There were some changes to the tag documentation pages for
highway=*_link several weeks ago:


All of these tags were previously documented to imply oneway=yes. Now
only the page for highway=motorway_link still contains that implication.

Generally, I don't think it's acceptable to change the definitions of
tags in this way. Maybe you believe that these implications didn't
actually make much sense or weren't implemented consistently. But they
were documented in the wiki for quite some time, so it's inevitable that
at least some mappers and applications used them. A change like this
should be preceded by at least some discussion, and we should try to
reliably inform application developers. There also needs to be a way of
of how to deal with existing data based on the old documentation.

I also wonder how we should deal with this specific situation. Can we
still assume any default oneway information for *_link values (whether
it's yes or no) or is oneway=yes/no a required second tag for *_links?

Should an application developer decide to assume oneway=no (which might
lead to longer-than-optimal routing results) or oneway=yes (which might
send you the wrong way up a trunk link)?

I'm not sure myself - for example, I'm still wondering how I should deal
with this for my GraphView plugin -, but I'd like to know other people's

Tobias Knerr

More information about the Tagging mailing list