Wie kann der Goetheweg bei der St. Mariakirche denn `bicycle:backward=no` haben und `foot:backward=designated`, also in dieser Richtung nicht erlaubt für Fahrräder, aber besonders geeignet für Fußgänger? Und die andere Richtung hat nur `foot:forward=yes`. Ist das hier wirklich wichtig um die Realität abzubilden oder tut es auch ein einfaches `oneway=yes` (=Fahrräder dürfen nur in eine Richtung fahren und Fußgänger in beide)?

D.h. aber mit der Kreuzung Pariser/Brüsseler/Stockholmer Straße wo der via node zuerst war hat das nichts zu tun. Gemeint war das Linksabbiegen von der Pariser auf die Dubliner Straße? Ich hab das repariert.

Hier ist eine turn relation kaputt gegangen. Dort fehlt jetzt der 'from' member: relation/16615
leider gibt iD in solchen Fällen keine Warnung o.ä. aus: https://github.com/openstreetmap/iD/issues/9346
Mal abgesehen davon müsste die via node dieser Restriktion wohl eher node/269774196 sein und statt 'only_right_turn' würde 'no_left_turn' hier besser passen.

I posted to the tagging mailing list

How can we determine whether there is unity in the community? Judging from the number of cases of combined restriction+restriction:xyz usage I'd say there is unity already, even before I changed anything. And who is responsible for the wiki page?

Yes, that would be ideal. I only changed cases where judging from the history I was pretty sure that they were simply mistakes (note that for example iD has limited support and even had a bug until recently for restriction:xyz, https://github.com/openstreetmap/iD/issues/9337). And in other cases like the present one I only left a comment or contacted the author directly, like here: changeset/127760119

One thing I tried to figure out was if there is an *actual* use-case for using both schemes and I don't think so.

Ah yes thanks I just added myself as author in the wiki discussion.
No, the wiki does not mention this (or I missed it), but IMO it should. The wiki doesn't seem very complete overall btw.

Ok, sure. Do you mean small in terms of the number of changed elements or small in terms of total area covered by all changes in the changeset (obviously quite large here).

Yes, 6854367 is missing a to-member (already before my changeset).

One more: I'm pretty sure there are traffic signs that state a turn is restricted for hgv, but none that state that a turn is restricted for everything except a long list of vehicles. So 'except=bicycle' seems to make sense, but not 'except=motorcar;psv;moped;bicycle'.

What makes the second scheme particularly unclear is that it doesn't seem to be defined what the 'complete' except list is. If you start excluding stuff when your intention is that only hgv remains you'd need to know what are all the possible values of the except tag (I even asked this here: osm.wiki/Talk:Relation:restriction#Possible_values_for_the_except_key), but again using (only) restriction:hgv, this problem doesn't even occur.

I see two (possible) disadvantages:

1) there is room for ambiguity, like what if the value of the restriction tag is 'no_left_turn' and the one for restriction:xyz is 'no_right_turn'? Yes this example is contrived, but if only one scheme is used this problem doesn't even come up.

2) using two schemes on a single restriction is just more complicated than using one, so it is harder to understand (for both humans and computer programs)

Now, I may ask the inverse question: What is the advantage of using both schemes at once?

I see BRouter does not understand restriction:xyz: http://brouter.de/brouter/suspect_scanning_readme.txt, GraphHopper understands it only since 2019. Anyway we shouldn't tag for (a specific) router, so that wouldn't be a good reason to use both schemes, especially because it doesn't really help as there are very many restrictions that only use restriction:hgv anyway.

I'm not sure, does
restriction:hgv=no_left_turn
mean the same as
restriction=no_left_turn
except=bicycle;moped;motorcar;psv
?
That depends on which transportation modes there 'are'. Is all minus bicycle;moped;motorcar;psv the same as hgv? And what would be the advantage of the second scheme, i.e. why use the except list when restriction:hgv already expresses what we need?
Also this kind of tagging seems very rare. This overpass query: `relation["restriction"]["restriction:hgv"];` only yields 14 (!) matches. IMO using restriction + restriction:xyz should be avoided, or at least I am trying to find out if there is any reason to use it.

Why did you add a restriction tag where there already was a restriction:hgv tag? If the turn is only restricted for hgv, restriction:hgv is sufficient, if the turn is restricted for all vehicles use only the restriction tag, but not both?

This turn restriction seemed fine before this changeset: `restriction:hgv=no_left_turn` means 'no left turns for hgv'. Why did you add an extra 'restriction' with a list of 'except' transportation modes?

It looks like the restriction relation/166650 is missing to- and via-members after your change. Can you fix or delete it?

Irgendwas ist hier glaube ich mit der relation/103243 kaputt gegangen. Sie hat jetzt nur noch einen member way mit to-Rolle. Es fehlen die from- und via-member.

Yes, that does not look very accurate, indeed. Honestly, I don't know why it wasn't more precise, but thanks for fixing it in the meantime.

Der Grund dass ich frage ist übrigens, dass wir das highway=ford tag in GraphHopper nicht mehr berücksichtigen werden, s. hier: https://github.com/graphhopper/graphhopper/pull/2538#issuecomment-1067272330

Hi @fkv,

mittlerweile gibt es noch genau 36 `highway=ford` tags. Davon sind 34 hier in der Gegend. Das tag ist (mittlerweile?) auch auf der Wikiseite deprecated: ford=*

Wenn `ford=yes` (noch) nicht in allen Karten gerendert wird, wäre es nicht richtiger das bei den Kartenrenderern zu ändern? Und wegen dem Problem dass eine Furt doppelt gerendert wird, wäre das nicht auch behoben wenn es einfach nur noch `ford=yes` gäbe?