detect highway=motorway, primary, secondary, tertiary (etc) with foot=yes as bogus

[mkoniecz]

foot=designated also is certainly (I think) bogus

​osmwww:browse/way/126376304 was affected till manual detection and fix

suggested list of highway types included in this check: motorway, motorway_link trunk, trunk_link, primary, primary_link, secondary, secondary_link, tertiary, tertiary_link

Repository Root: http://josm.openstreetmap.de/svn
Build-Date: 2013-11-27 17:47:56
Last Changed Author: Don-vip
Revision: 6418
Repository UUID: 0c6e7542-c601-0410-84e7-c038aed88b3b
URL: http://josm.openstreetmap.de/svn/trunk
Last Changed Date: 2013-11-26 17:24:06 +0100 (Tue, 26 Nov 2013)
Last Changed Rev: 6418

Identification: JOSM/1.5 (6418 en_GB) Windows 7 32-Bit
Memory Usage: 87 MB / 247 MB (14 MB allocated, but free)
Java version: 1.7.0_45, Oracle Corporation, Java HotSpot(TM) Client VM
Dataset consistency test: No problems found

Plugin: OpeningHoursEditor (29854)
Plugin: notes (v0.6)

[simon04]

This would result in many false-positives in Austria: Here, many primary/secondary/tertiary streets have sidewalks. It would be okay for motorway.

[mkoniecz]

So it is a third method for marking sidewalks after separate footways and sidewalk key? Great.

[Don-vip]

in this case foot=* should indeed be replaced by sidewalk=*

[simon04]

Hm, I wasn't so sure about the tagging scheme for sidewalks.

To summarize, neither foot=yes nor foot=no makes sense (since foot=no is the default), and foot=yes can/should be replaced by sidewalk?

[mkoniecz]

foot=yes can be replaced by sidewalk field, or by a separate footway, or removed - in some cases this value is bogus.

[simon04]

In 6549/josm:

fix #9379 - Add validator warning: "foot used with highway=motorway|trunk|primary|secondary|tertiary, use sidewalk or separate footway instead"

[aceman]

This warning is triggered also for foot=no (+highway=motorway). That is bogus as in may countries foot=yes is the default for most road classes (specifically when there is no sidewalk). So foot=no is needed for some roads where pedestrians are specifically forbidden (via road sign). See ​http://wiki.openstreetmap.org/wiki/OSM_tags_for_routing/Access-Restrictions .
Please tweak the patch. If you want the warning for foot=yes, then fine, but for foot=no it the suggestion text is bogus. I don't see where you check for foot=yes in the patch.

[mkoniecz]

I am surprised by ​http://wiki.openstreetmap.org/wiki/OSM_tags_for_routing/Access-Restrictions#Default

on highway=trunk foot defaults to yes? It is ridiculous.

[aceman]

You should probably make a separate table for your country if the defaults are not right there. In our table (Slovakia) we have foot=no on trunk. But in some country-specific tables you can see people have specifically set foot=yes, so it does not appear that ridiculous.

[Don-vip]

Replying to Bulwersator:

on highway=trunk foot defaults to yes? It is ridiculous.

+1, as horse and bicycle...

[simon04]

What about classifying this validation result as "information" instead of "warning"?

[aceman]

I do not fully understand what is actually wrong with foot=yes on a road (you can officially walk on the road surface itself even when there are no sidewalks) and what this ticket wants to achieve. Anyway, the description of the ticket mentions foot=yes, not foot=no so if the check suddenly does something also with foot=no (and it does) then the feature should be properly described and documented here. Without that, we can't decide if the check is working properly and what class it whould be.

[simon04]

The code is readable like a book, so please refer to comment+diff of r6549 and Help/Validator/MapCSSTagChecker before asking for even more documentation.

foot=yes might indicate a problem since you might not be allowed to walk on major roads, and foot=no might be superfluous since that might be considered superfluous.

[mkoniecz]

"you can officially walk on the road surface itself even when there are no sidewalks"

Note "motorway|trunk|primary|secondary|tertiary" check. Can you give example of road in this class where you can walk on carriageway?

[aceman]

Yes, in Slovakia (and I think also all countries that specified foot=yes on ​http://wiki.openstreetmap.org/wiki/OSM_tags_for_routing/Access-Restrictions) you can walk on primary|secondary|tertiary and people regularly do so (e.g. walk between villages).

The law says literally: A pedestrian must use the sidewalk. Where there is no sidewalk he walks on the left border of the road [my explanation: the 0-3m buffer of road surface that is outside the lanes where motor vehicles run (carriage ways)]. Where there is no border of the road, he walks on the leftmost possible part of the carriage way [my explanation: inside lanes where motor vehicles run].

There is even NO exception for motorway/trunk in this paragraph, but it MAY be somewhere else, so I would not argue about that. It also feels natural to me that motorway/trunk is forbidden.

So in my understanding, I would understand a JOSM warning in these cases:
-sidewalk=yes + foot=yes (as the pedestrian should use the sidewalk, not the road surface)
-foot=yes (with no tag about sidewalk) - isn't only this combination the use case this ticket is about?

But these cases seem fine and should produce no warning:
-sidewalk=no + foot=yes
-sidewalk=yes + foot=no
-foot=no (even though it may be superfluous with the defaults, it is no logical problem)

(In all my texts here I assume the definition of sidewalk from ​http://wiki.openstreetmap.org/wiki/Sidewalk , i.e. a special footway along the road separated by kerb. The "buffer" edge of a road I talk about above is on the same vertical level as the lanes (carriage ways) and is physically part of the surface (asphalt) and is only separated by a solid white line.)

[aceman]

This is the "buffer" part of road I meant: ​http://en.wikipedia.org/wiki/Shoulder_%28road%29 .

[aceman]

All of this only works if we assume all the access tags apply to the carriageway+shoulder road only and not to the sidewalk. But in that case any tagged sidewalk will have implicit foot=yes even when that is not appropriate (e.g. it could be foot=private if the main road has access=private). The wiki also admits this.
I think there are open issues with using sidewalk tag. So it appears better to me to still map the sidewalk as separate footway alongside it (of course, then I have to have enough crossing points to allow for effective routing).

[simon04]

So to summarize, the warning should only appear if foot=yes and sidewalk is missing?

[aceman]

Yes, that is what I would expect. (Also foot=designated and maybe foot=permissive). The sidewalk=yes + foot=yes may be country dependent so better not pollute the warnings for now.

Let's wait for the ticket reporter's opinion.

[simon04]

In 6632/josm:

fix #9379 - warn foot on major roads only if foot=yes and sidewalk is missing

[alv]

The warning message is suggesting to draw errors on the map. The countryside has lots of roads up to trunk status that do not have sidewalks, but walking is not banned, and people do walk there. I wouldn't tag foot=yes, but drawing non-existant sidewalks as separate paths is bonkers, as is adding nonexisting sidewalk. At least the warning message must point out that foot=yes + sidewalk=no (or is it none) would be correct.

[Don-vip]

What about removing this check ? It's not a major issue and I'm tired of this non-ending discussion on special cases. People may define their own country-specific checks in centralized validator rules service, now.

[simon04]

Replying to Don-vip:

What about removing this check ?

+1

It's not a major issue and I'm tired of this non-ending discussion on special cases.

Me too. :-)

[anonymous]

Replying to alv:

The warning message is suggesting to draw errors on the map. The countryside has lots of roads up to trunk status that do not have sidewalks, but walking is not banned, and people do walk there. I wouldn't tag foot=yes, but drawing non-existant sidewalks as separate paths is bonkers, as is adding nonexisting sidewalk. At least the warning message must point out that foot=yes + sidewalk=no (or is it none) would be correct.

I do not understand this problem. When do you get this warning from JOSM? It should only trigger when there is foot=yes and no mention of sidewalk. You say you would not put foot=yes on a road you describe. In that case the warning should not show. So what is the problem? Can you please explain it more?

If other people in your country have put foot=yes on such a road and now you get the warning, adding sidewalk=none is not "bonkers", but adding more real information. Yes, if it is not easy to guess from the text that you need to add sidewalk=, then it can be improved. But I don't think the test itself is bogus in the current refined implementation.

[aceman]

Sorry, I got logged off. The comment 27 is mine.

[alv]

Replying to anonymous:

So what is the problem? Can you please explain it more?
adding sidewalk=none is not "bonkers"

Adding isn't. Saying (to anyone happening to edit a way where somebody has at some point added foot=yes) that "warning: road with foot" is implying "Foot on (major) road looks wrong, don't use foot on road. Add or draw sidewalk instead." People don't read the wiki, they too easily generalize random warning messages and apply that elsewhere. So I'd be content with dropping to information level, and rephrasing, like (with drafting I think the last one is short but specific enough?) :

• sidewalk unknown but foot allowed on major road, tag or draw separate footway where exists
• unknown if major road has sidewalk, specify with sidewalk or draw a separate footway
• foot belongs to the sidewalk (separate footway) or add sidewalk=none
• sidewalk=none missing, or foot belongs to separate footway

I couldn't find sidewalk in the presets, so users couldn't know of the values no or none, or the others.

[simon04]

Vincent is right: 29 comments and a never ending discussion for a relatively unimportant validator test. I don't think that longish tagging discussions should be performed in JOSM's bugtracker since for JOSM development it's cumbersome to follow the discussion and very easy to overlook a minor detail hidden somewhere in the discussion. I feel that there more important changes/improvements to be done for JOSM than this issue for instance.

We will most likely remove this test. Feel free to write and experiment with custom tests via Rules.

[aceman]

Replying to alv:

Adding isn't. Saying (to anyone happening to edit a way where somebody has at some point added foot=yes) that "warning: road with foot" is implying "Foot on (major) road looks wrong, don't use foot on road. Add or draw sidewalk instead." People don't read the wiki, they too easily generalize random warning messages and apply that elsewhere. So I'd be content with dropping to information level, and rephrasing, like (with drafting I think the last one is short but specific enough?) :

Thanks for the exaplanation. Yes, I'd be OK with this solution too. And the original reporter would still have the test he wanted.

[mkoniecz]

I agree. I thought that it is simple and obvious (and created this ticket) but it degenerated into a complete and complex mess.

Now I only need to find how I can add my own validator rules. Unfortunately ​http://josm.openstreetmap.de/wiki/Help/Validator/MapCSSTagChecker is not providing an answer, googling also failed.

[simon04]

Replying to Bulwersator:

Now I only need to find how I can add my own validator rules. Unfortunately ​http://josm.openstreetmap.de/wiki/Help/Validator/MapCSSTagChecker is not providing an answer, googling also failed.

Create and test a validation test: Create a .validator.mapcss file, add some tests (according to Help/Validator/MapCSSTagChecker) and add it via the JOSM preferences.

To make a test public: Add a subpage to Rules, see the source code of Rules/Brazilian-Specific for an example.

Btw: extending the documentation is very welcome. :-))

[simon04]

The removal ticket #9686 (for correct "milestoning") …