[mkgmap-dev] Turn Restrictions using three ways - design guide for OSM mappers?

Tue Oct 8 07:59:59 BST 2019

Sorry, forgot to add the link for [1]: http://gis.19327.n8.nabble.com/Turn-restrictions-with-role-via-ways-tp5799637.html

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Gerd Petermann <gpetermann_muenchen at hotmail.com>
Gesendet: Dienstag, 8. Oktober 2019 08:49
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] Turn Restrictions using three ways - design guide for OSM mappers?

Hi blc,

maybe the misunderstanding is on my side.  I've implemented the support for via-ways in 2014 [1]. For such a no_*  restriction there is no way to express the same situation with via-nodes.  The Garmin IMG format only knows restrictions which forbid to follow a certain list of nodes in a given order. So, I tend to interpret OSM restrictions like that and for an OSM "no" restriction this works fine.

I searched the tagging list and found
http://gis.19327.n8.nabble.com/mandatory-restriction-with-via-way-as-members-tp5924100.html
Following this thread the relation 2155309 means: If you are on "from" way 223720277 you must use the "via" way followed by the "to"  way.

I fear this is currently not implemented in mkgmap, it more or less translates the relation into a no_u_turn restriction with way 161220789 as "to" way and thus
allows a right turn from way 172716357 via node 1732097310 to way 161220701.
IIGTR the code in mkgmap should be changed to add more restrictions in the IMG data...

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von blc <blc at mail.vanade.com>
Gesendet: Dienstag, 8. Oktober 2019 06:05
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] Turn Restrictions using three ways - design guide for OSM mappers?

Still having a bit of misunderstanding, I think.

Maybe take a look at relation 2155309 which also triggers the warning.
https://www.openstreetmap.org/relation/2155309

Here there appears to be no other relations describing this turn other
than the one more that prohibits left turns on the right turn ramp which
is only tangentally related.  My take is that the mapper decided to write
it this way for some reason, perhaps there's a sign that says such, I don't
know.  While this too can be written as two TRs:
- one to prohibit right turns
- one to prohibit U-turns.

iD even seems to understand this way-as-via does both with just one
restriction relation, you can click and mouse over the voodoo doll
sidebar and it seems to understand the original mapper's intent.  In fact
I think you can make iD can create way-as-via only relations.

So what should a mapper have done here from the get go?  Should they
have omitted it completely?  What about a QA fixer?  Or should these
warnings simply be ignored -- there are hundreds of them in the USA.

Thanks, and sorry for any misunderstandings.

On Sat, 5 Oct 2019, Gerd Petermann wrote:

> Date: Sat, 5 Oct 2019 14:31:29 +0000
> From: Gerd Petermann <gpetermann_muenchen at hotmail.com>
> Reply-To: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
> To: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
> Subject: Re: [mkgmap-dev] Turn Restrictions using three ways - design guide
>     for OSM mappers?
>
> Hi blc,
>
> I meant the relation 2256354 is obsolete, wrong, or meaningless, whatever you want to call it. I can't think of any situation in which a only-* restriction with a via way makes sense. Can you give one? What would be restricted by such a restriction?
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von blc <blc at mail.vanade.com>
> Gesendet: Samstag, 5. Oktober 2019 16:12
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] Turn Restrictions using three ways - design guide for OSM mappers?
>
> Gerd,
>
> Ahhh... Now I understand what you mean.  I think you initially mean
> "redundant" versus "obsolete" - I was confused by "obsolete" thinking that
> the restriction should use a different method due to an outdated
> methodology.
>
> For the example, turn restriction 3843893 was the one that made the
> initial turn restriction 2256354 redundant.  If 3843893 had not been
> there, what would the suggestion be?
>
> I'd suspect that relation 3843894 is still needed regardless as the
> startpoint is different.
>
> Thanks
>
> On Sat, 5 Oct 2019, Gerd Petermann wrote:
>
>> Date: Sat, 5 Oct 2019 13:31:09 +0000
>> From: Gerd Petermann <gpetermann_muenchen at hotmail.com>
>> Reply-To: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
>> To: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
>> Subject: Re: [mkgmap-dev] Turn Restrictions using three ways - design guide
>>     for OSM mappers?
>>
>> Hi blc,
>>
>> there are already normal restrictions [1]  which look correct to me, so as I said before this one is obsolete. I think it should be removed.
>> Besides that I would not add restrictions without local knowledge or other allowed sources.
>>
>> Gerd
>> [1] https://www.openstreetmap.org/relation/3843893
>> https://www.openstreetmap.org/relation/3843894
>>
>> ________________________________________
>> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von blc <blc at mail.vanade.com>
>> Gesendet: Samstag, 5. Oktober 2019 08:53
>> An: Development list for mkgmap
>> Betreff: Re: [mkgmap-dev] Turn Restrictions using three ways - design guide for OSM mappers?
>>
>> Gerd,
>>
>> Thanks for the reply.
>>
>> So it looks like it's still being handled, but would you say that these
>> should be changed in OSM?
>>
>> It seems a bit strange that if you're on way A, you must travel through
>> way B and get to way C, but indeed it is true that if you weren't allowed
>> to make any turn at the point between A and B (and B and C), you'd get the
>> same result -- is this the prefered way of denoting such?
>>
>> For this particular example in OSM I suspect the mapper did not
>> want to allow right turns at the intersection (even if it's not
>> illegal) and hence wrote the restriction as an only left way-way-way
>> instead of a way-point-way no right turn, perhaps because of either a sign
>> or the paintings on the road and you can't make an "only left turn" on the
>> first intersection of the dual carriageway because that's the wrong
>> direction.
>>
>> How should this particular intersection be restricted from travel to not
>> emit warnings?  Adding that no right turn at the first intersection
>> would probably have the effect, but I've seen a lot of these way-way-ways
>> around (mostly dealing with complex dual carriageway intersections between
>> multiple roads) and wonder if it's worth "fixing" them, or should these
>> warnings be simply ignored for the most part?
>>
>> Thanks!
>>
>> On Sat, 5 Oct 2019, Gerd Petermann wrote:
>>
>>> Date: Sat, 5 Oct 2019 05:43:56 +0000
>>> From: Gerd Petermann <gpetermann_muenchen at hotmail.com>
>>> Reply-To: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
>>> To: "mkgmap-dev at lists.mkgmap.org.uk" <mkgmap-dev at lists.mkgmap.org.uk>
>>> Subject: Re: [mkgmap-dev] Turn Restrictions using three ways - design guide
>>>     for OSM mappers?
>>>
>>> Hi blc,
>>>
>>> the code that produces these warnings is this:
>>>                if (valid && !viaWays.isEmpty() && restriction.startsWith("only")){
>>>                        log.warn(messagePrefix, "check: 'via' way(s) are used in",restriction,"restriction");
>>>                }
>>>
>>> So, mkgmap considers them valid, but dubious. I think that's what they are. The restriction says something like
>>> "when you want to travel from way A via way B to way C you MUST travel from A via B to C"
>>> What kind of restriction is that? In my eyes, the given example is completely obsolete.
>>> On the other hand, a "no-" restriction with via way(s) means
>>> It is not allowed to go from A to C via B. This cannot be expressed with a single via node.
>>>
>>> Hope that helps?
>>>
>>> Gerd
>>>
>>> ________________________________________
>>> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von blc <blc+mkgmap at mail.vanade.com>
>>> Gesendet: Samstag, 5. Oktober 2019 06:31
>>> An: mkgmap-dev at lists.mkgmap.org.uk
>>> Betreff: [mkgmap-dev] Turn Restrictions using three ways - design guide for     OSM mappers?
>>>
>>> Hello, I thank all who have been working on this neat program to allow our
>>> otherwise old Garmins sit in the dust heap when we can't afford to
>>> subscribe to new maps.
>>>
>>> I've been trying to improve the quality of OSM by fixing the errors
>>> that mkgmap emits, which a lot of times mirrors what's seen in
>>> KeepRight.  However there's one variant of turn restriction I've noticed
>>> that warns in mkgmap but do not show up in KeepRight (and iD seems to
>>> understand this type of turn restriction) - the way-way-way type
>>> restriction where three connected ways are in series for non no-u-turn
>>> restrictions.
>>>
>>>
>>> example:
>>>
>>> Turn restriction (only_left_turn) 2256354 (at
>>> https://www.openstreetmap.org/?mlat=47.777585&mlon=-122.319488&zoom=17)
>>> check: 'via' way(s) are used in only_left_turn restriction
>>>
>>> The way-way-way type is the proper method for restricting u-turns
>>> on dual carriageway roads which is understood by mkgmap.  On the
>>> other hand, iD and KeepRight it seems to be valid to do way-way-way
>>> instead of way-POINT-way for no/only left/right turn restrictions, no/only
>>> straight on restrictions, etc.  I've seen a lot of the non no-u-turn
>>> way-way-way restrictions in the USA.
>>>
>>> These type of non no-u-turn restrictions seems to cause a warning in
>>> mkgmap and probably not translating them.  My question is that should
>>> these be supported in mkgmap, or should these be fixed in OSM so that they
>>> are simple way-via-way despite iD and KeepRight seem to claim them
>>> valid?  Or perhaps way-way-way is deprecated but still supported by OSM
>>> but should be changed to way-point-way?
>>>
>>> way-point-way = relation
>>> from: some-street-way
>>> via: some-intersection-point
>>> to: some-street-way
>>> (this is the most common type of turn restriction)
>>>
>>> way-way-way = relation
>>> from: some-street-way
>>> via: some-street-way
>>> to: some-street-way
>>> (this is necessary specifically for dual carriageway u-turn restriction,
>>> but it's used for other types as well which mkgmap complains about.)
>>>
>>> Thanks for shedding some light on the discrepancy here!  Note: I'm
>>> currently depending on OpenMapChest data for mkgmap runs as my computer
>>> and internet connection are not large or fast enough for the quantity of
>>> data I'd like to work with.
>>> _______________________________________________
>>> mkgmap-dev mailing list
>>> mkgmap-dev at lists.mkgmap.org.uk
>>> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>>> _______________________________________________
>>> mkgmap-dev mailing list
>>> mkgmap-dev at lists.mkgmap.org.uk
>>> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>>>
>>
>> WARNING: All HTML emails get auto deleted.  DO NOT SEND HTML MAIL.
>> WARNING: Emails with large typo counts/spelling errors will also be deleted.
>> _______________________________________________
>> mkgmap-dev mailing list
>> mkgmap-dev at lists.mkgmap.org.uk
>> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>> _______________________________________________
>> mkgmap-dev mailing list
>> mkgmap-dev at lists.mkgmap.org.uk
>> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>>
>
> WARNING: All HTML emails get auto deleted.  DO NOT SEND HTML MAIL.
> WARNING: Emails with large typo counts/spelling errors will also be deleted.
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>

WARNING: All HTML emails get auto deleted.  DO NOT SEND HTML MAIL.
WARNING: Emails with large typo counts/spelling errors will also be deleted.
_______________________________________________
mkgmap-dev mailing list
mkgmap-dev at lists.mkgmap.org.uk
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
mkgmap-dev at lists.mkgmap.org.uk
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev