[OSM-dev] Merkaartor creates invalid highway tags like highway=unclassified; track

Andy Townsend ajt1047 at gmail.com
Mon Mar 14 22:57:24 UTC 2016


On 14/03/2016 08:58, Gerd Petermann wrote:
>
> Hi all,
>
> I hoped that this problem was solved whe iD was corrected, but
>
> some less often used editors still allow to create these erroneous tags.
>
> Since iD was corrected the only changesets with these errors are created
>
> by Merkaartor  and Potlatch 2. Fortunately these editors are rather 
> rarely used.
>
> I hope the programmers of these editors are also working on a fix ?
>
>

I think perhaps that we need to to a step back here.  You seem to be 
equating "valid" data with "correct" data, when in reality if something 
is "invalid" it's usually a very good indication that further survey is 
required, and if it's been made "valid" by someone picking a key/value 
combination that "the original user might have meant", there's no 
guarantee it's correct.

For example, on https://www.openstreetmap.org/node/1057585745/history 
you've changed the obviously invalid "footway=sigtnpost" to 
"information=guidepost; tourism=information".

What you've changed it to is "valid", but not necessarily correct; it 
might instead be an "information=route_marker", or it might be a 
bridleway marker suggesting that one or both of 
https://www.openstreetmap.org/way/5037727 and 
https://www.openstreetmap.org/way/167425506 should actually be 
"designation=public_bridleway" and have "foot=yes", "horse=yes" and 
perhaps "bicycle=yes" tags on them.

Now that you've "fixed" the data there there's no longer something 
that's "obviously invalid" that people can see and say "I wonder what 
that is supposed to be" before going out and surveying it. Also, it's 
perhaps worth mentioning in passing that telling me via 
https://www.openstreetmap.org/changeset/35996416 to "Please contact the 
original mapper" has perhaps lost some politeness in the German to 
English translation (as I'm sure it would the other way if I'd tried to 
write it in German); it suggests that your time is far more valuable 
than other mappers'.

It's a similar issue with semicolon-separated values.  If someone has 
created (say) "sidewalk=both;right" then it's pretty easy to see what's 
happened - someone has merged two ways together and not known to check 
all the tags. However, imagine if instead of "sidewalk=both" or 
"sidewalk=right" one of the ways hadn't got a sidewalk tag on it at 
all.  The result would still be incorrect, but there's no way to easily 
detect it.  The only way to prevent this problem occurring is to educate 
mappers about tags that they might not be aware of, and the best way to 
do that is to actually talk to them, not to suggest $technical_fix to 
whatever editing software they happen to be using.  Editors can help, 
sure (for example, the way that Potlatch 2 highlights 
semicolon-separated values is one way of dealing with it) but ultimately 
a human needs to answer the question "why did that just happen" in order 
to prevent new and not-so-new mappers making the same mistake again.

Best Regards,

Andy (SomeoneElse)

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20160314/6731ab20/attachment.html>


More information about the dev mailing list