[Tagging] V3 to V4 Mapillary image id migration

[Water-Map]

Hi Kai,

Yes, it does seem that reaching consensus and wiki alignment with this user
will be very difficult.

Another argument for only having the pKey (id) in the mapillary tag is that
the v4 notation referenced in your email:
*https://www.mapillary.com/app/?pKey=
<https://www.mapillary.com/app/?pKey=> *seems to link to the mapillary app
which shows the image and not the image itself. I believe that one now
needs to use the v4 api to obtain the thumbnail image url. (maybe I am
wrong).

https://blog.mapillary.com/update/2021/06/23/getting-started-with-the-new-mapillary-api-v4.html

Best regards,

Stuart


On Wed, 3 Nov 2021 at 22:00, Kai Michael Poppe - OSM <osm at poppe.dev> wrote:

> *sigh*
>
> If anyone wants to follow the changeset discussion, having myself insulted
> for a hobby that I have enjoyed at some point in time, feel free.
>
> Kai
> On 03.11.2021 19:57, Kai Michael Poppe - OSM wrote:
>
> Hi,
>
> Well "the wiki is wrong" is quite a bold statement when every language
> page says exactly the same with the English version having done so for 6
> years.
>
> Fortunately, I have a very extensive list of the current values (*g*) and
> can confirm: There are 202,203 "valid" old values in the database of which
> "only" 37,615 values use the http-notation (18,6%). I would dispute that
> that's the correct way to tag it.
>
> Also: The wiki specifically states "The reason for adding the key and not
> the full URI is that the URI might change (and already did in the past),
> but the key will stay the same. For the same reason the direct link to the
> image (e.g. *https://images.mapillary.com/.../thumb-2048.jpg
> <https://images.mapillary.com/.../thumb-2048.jpg>*) should not be
> tagged." ... We're QUITE LITERALLY seeing that the v3 notation of the
> web-app was https://www.mapillary.com/map/im/... and the v4 notation is
> https://www.mapillary.com/app/?pKey=...
>
> I'm not in favour of telling a contributor off that has 3k+ edits, but I
> really don't want to leave this uncommented.
>
> Kai
> On 03.11.2021 19:37, Peter Neale via Tagging wrote:
>
> @Kai and everyone,
>
> I have hit a problem.  I have corrected a "mapillary=*" tag, where another
> user had put the whole mapillary URL into the key field.  I removed it and
> left only the v4 key (16 digit numeric key), with a comment that they were
> in error by putting the whole URL into the key field.  This was the second
> instance of this with the same user.
>
> They have now responded to my comment, saying that "the wiki is wrong".
>
> Please see https://www.openstreetmap.org/changeset/113303225 and the
> changeset comments.
>
> Who is correct?  How do we resolve this difference of opinion?
>
>
>
> @Kai.  No problem; I have actually quite enjoyed doing it.  (Only 19 more
> to go....)
>
> Regards,
> Peter
> (aka PeterPan99)
>
>
> On Wednesday, 3 November 2021, 18:20:36 GMT, Kai Michael Poppe - OSM
> <osm at poppe.dev> <osm at poppe.dev> wrote:
>
>
> @CJ,
>
> Thanks for clearing that up for me, I'll guess for the "cleanup" of the
> no-longer-existing-old-valid-values I'll have to put two Challenges, one
> with Nodes/Ways and one with any one member of a Relation and a different
> description - as MapRoulette only has GeoJSON or an overpass query which
> might not be useful in this case (have to check).
>
> @Peter,
>
> Thanks for clearing the whole Challenge!
>
> Kai
>
> On 03.11.2021 16:29, Cj Malone wrote:
> > On Wed, 2021-11-03 at 14:01 +0100, osm at poppe.dev wrote:
> >> Q1 - yes. If there's a way/node in the Challenge that has no
> >> mapillary=* tagging then it is most likely part of a relation where
> >> that's the case. When setting up the challenge, I was 99% sure I
> >> added the Relations as their own objects and put them into the
> >> GeoJSON that's the basis for the Challenge. It might very well be
> >> that MR just ignores relations (as it might not know how to display
> >> them properly) and add the objects that are part of it (as they are
> >> in the GeoJSON as well) or I was just too stupid to deliver the
> >> correct data.
> > GeoJSON doesn't support relations. It's kinda a bad format for complex
> > data structures, like anything that comes from OSM.
> >
> > Another one is, if a way has a node in it with a tag on it, it's
> > actually stored as a point in the way, and separately a node on it's
> > own (with the tags) then the program that reads the data has to
> > correlate the points with nodes by finding anything that shares a
> > location. This then means if you have 2 nodes with the same location
> > they can erroneously get merged into one point with some software (like
> > JOSM).
> >
> > I used to love geojson before running into weird issues like this. Now
> > I try to keep things in .osm as much as possible.
> >
> > Cj
> >
> >
>
> _______________________________________________
> Tagging mailing list
> Tagging at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
>
> _______________________________________________
> Tagging mailing listTagging at openstreetmap.orghttps://lists.openstreetmap.org/listinfo/tagging
>
>
> _______________________________________________
> Tagging mailing listTagging at openstreetmap.orghttps://lists.openstreetmap.org/listinfo/tagging
>
> _______________________________________________
> Tagging mailing list
> Tagging at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20211104/54fd7e48/attachment.htm>