<div dir="ltr">for background info: a couple of links about Navads and what they do for Shell (and other brands) <a href="http://www.prweb.com/releases/navads/shell/prweb13779126.htm">http://www.prweb.com/releases/navads/shell/prweb13779126.htm</a> and <a href="https://navads.eu/businesslistings/">https://navads.eu/businesslistings/</a><div><br></div><div>It would be interesting for OSM[UK] to be an official recipient of up-to-date, clean data if indeed it is proved to be clean enough for import.</div></div><br><div class="gmail_quote"><div dir="ltr">On Fri, 12 May 2017 at 08:59 Robert Whittaker (OSM lists) <<a href="mailto:robert.whittaker%2Bosm@gmail.com">robert.whittaker+osm@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On 11 May 2017 at 23:25, Frederik Ramm <<a href="mailto:frederik@remote.org" target="_blank">frederik@remote.org</a>> wrote:<br>
> Link to discussion so far on imports@:<br>
> <a href="https://lists.openstreetmap.org/pipermail/imports/2017-May/004956.html" rel="noreferrer" target="_blank">https://lists.openstreetmap.org/pipermail/imports/2017-May/004956.html</a><br>
><br>
>> My concern would be from where to they get their geocoding. Most<br>
>> businesses, and particularly chain businesses, tend to use postcode<br>
>> centroids, which are not accurate enough, probably get them from Google.<br>
><br>
> I voiced the same general concern, but a random sample I checked of the<br>
> (actually rather few) stations that are proposed to be newly added<br>
> seemed to be impeccably placed.<br>
<br>
In which case, there is a different concern: have they done their<br>
geo-coding from an acceptable source for use in OSM? If they've e.g.<br>
used Address Base (or a similar product) or got coordinates from a<br>
non-OpenData OS map, then there could be problems. I think we need<br>
more information on the data sources here.<br>
<br>
Some other comments:<br>
<br>
* If a ref/id is to be used, it should probably be Shell's branch<br>
reference number, not that of the third-party data provider. (These do<br>
exist, and at least in some cases are verifiable on the ground, as<br>
I've found at least one on a pump at a Shell garage up the road from<br>
me.)<br>
<br>
* There's an addressing edge-case error on a station near me, which is<br>
located on the Five Ways Roundabout near Mildenhall:<br>
<a href="http://www.openstreetmap.org/way/478902268" rel="noreferrer" target="_blank">http://www.openstreetmap.org/way/478902268</a> . We currently have<br>
(incorrectly) "addr:place=5 Ways Roundabout", but the script is<br>
proposing adding "addr:street=Ways Roundabout" and<br>
"addr:housenumber=5".<br>
<br>
* The script shouldn't just add source=Navads to objects it's only<br>
modifying, as that would imply the whole object was sourced from<br>
there. If existing tags and position are retained, then this needs to<br>
be acknowledged somehow. If there's an existing source tag, then<br>
Navads could just be added to the list (I haven't checked to see if<br>
this is the case). If not, then there's more of a challenge. The<br>
script current just adds source=Navads in this case. I think the<br>
importers need to propose a better solution for this.<br>
<br>
* As others have said, there needs to be more information about what<br>
happens if there are multiple amenity=fuel objects within 50m, and<br>
also what happens if any existing tags conflict with what the script<br>
would like to add.<br>
<br>
* The proposed website tag appears to point to <a href="http://www.shell.co.uk" rel="noreferrer" target="_blank">http://www.shell.co.uk</a><br>
for all the branches. Would it be better pointing to a specific URL<br>
for that branch (assuming this exists)?<br>
<br>
* The opening_hours from the import script for<br>
<a href="http://www.openstreetmap.org/way/248030653" rel="noreferrer" target="_blank">http://www.openstreetmap.org/way/248030653</a> don't match those displayed<br>
on Shell's own website for the same station. One as open till 11pm on<br>
Saturday, the other only 10pm. So is the data accurate / up-to-date?<br>
<br>
Robert.<br>
<br>
--<br>
Robert Whittaker<br>
<br>
_______________________________________________<br>
Talk-GB mailing list<br>
<a href="mailto:Talk-GB@openstreetmap.org" target="_blank">Talk-GB@openstreetmap.org</a><br>
<a href="https://lists.openstreetmap.org/listinfo/talk-gb" rel="noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/talk-gb</a><br>
</blockquote></div>