[mkgmap-dev] is_in() function for point on line

Tue Jun 9 13:10:30 BST 2020

Hi Joris,

with the latest patch poi-tagged-v3.patch I see this:
Node generated from 241008394 [dbg1=true, highway=track, mkgmap:from-node:barrier=gate, mkgmap:line2poi=true, mkgmap:line2poitype=inner, mkgmap:way-length=503, name=Gillispad] from-node

My options:
--add-pois-to-lines=mid   --route --style-file=c:\mystyles\def
where def is a copy of the default style
containing these extra rules
mkgmap:from-node:highway = turning_circle & highway = track             [0x641a resolution 24]
mkgmap:from-node:highway = turning_circle & highway != track            [0x641b resolution 24]
mkgmap:from-node:barrier = gate & highway = track        {set dbg1 = true ; echotags 'from-node'}

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Joris Bo <jorisbo at hotmail.com>
Gesendet: Dienstag, 9. Juni 2020 13:29
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] is_in() function for point on line

Hi

The 'highway = turning_circle' still works.
Options used:   --add-pois-to-lines & --link-pois-to-ways
mkgmap r4525 + poi-tagged-v3.jar

I can catch it with:    mkgmap:from-node:highway = turning_circle & highway = track     [0x641a resolution 24]

But another example (i wrote 2 times a more detailed mail with pictures, which I guess does not come through)
        barrier = gate on a highway = track can't be catched

        No response for
                mkgmap:from-node:barrier = gate & highway = track       {set dbg1 = true ; echotags 'from-node'}

        if I address the node by osmid() it can be found but then there are no values of the track itself available
                barrier = gate & osmid() = 2486838092                   {set dbg2 = true ; echotags 'by-id'}

If I echotags the track (where the barrier is part of) then there is a value:  mkgmap:way-has-pois=true
Looks like for the barrier no new node is created.

Test area
Node: barrier = gate     https://www.openstreetmap.org/node/2486838092
Highway = track  https://www.openstreetmap.org/way/241008394
I hoped to find the values of the track available with the barrier to be able to control the access tags afterwards based on what I find on the track (route_from_relation=foot) , in my case

Kind regards Joris

-----Oorspronkelijk bericht-----
Van: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> Namens Gerd Petermann
Verzonden: dinsdag 9 juni 2020 12:03
Aan: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
Onderwerp: Re: [mkgmap-dev] is_in() function for point on line

Hi all,
I've attached a new patch. The parameter tagged-node is not needed anymore. The corresponding flag is set when the points style uses the mkgmap:from-nodes: prefix. I've changed the code so that only those tags which are used with this prefix are copied, the special tag mkgmap:copied-as-poi=true is only to the original node added if at least one of its tags was copied with the mkgmap:from-nodes:  prefix.

Reg. performance: Without the additional logic mkgmap requires a lot more memory, so I think it is better to reduce the number of generated POI and the number of added tags.
Binary is here:
http://files.mkgmap.org.uk/detail/481

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Gerd Petermann <gpetermann_muenchen at hotmail.com>
Gesendet: Dienstag, 9. Juni 2020 10:24
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] is_in() function for point on line

Hi all,

forgot to add the download link for the binary:
http://files.mkgmap.org.uk/detail/480

I've now tested that  --add-pois-to-lines=mid,tagged-node tagged-node works as expected, but documentation still needs improvements.
If e.g. --add-pois-to-lines=mid,tagged-node is used the option tagged-node has no effect unless the points style contains a corresponding rules like
mkgmap:from-node:highway = turning_circle & highway = track             [0x641a resolution 24]
is found.
I've implemented it like this to reduce the memory footprint because I thought of those nodes which have unintresting tags like source=*. I forgot that those tags are already removed before POIGeneratorHook is executed. I'll do some more testing now...

Gerd

Gerd

________________________________________
Von: Gerd Petermann <gpetermann_muenchen at hotmail.com>
Gesendet: Dienstag, 9. Juni 2020 09:43
An: Development list for mkgmap
Betreff: AW: [mkgmap-dev] is_in() function for point on line

Hi Mike, Hi Joris

while looking at patch addpois-doc.patch I noticed that poi-tagged.patch has a paste+copy error. The new option "tagged" did not work.
Attached is a corrected version which implements tagged-node and changes/adds documentation.

@Mike: Don't know why I used "other" instead of "inner". My bad. We cannot just change the doc, we also have to support the corrected option.

Reg. "creation of POIS from the lines style file" : It might be possible to add new logic which would add a point object while a line is processed. Up to now the order is "relations, points, lines[+polygons], so POIs are processed before lines. If we change that we have to find an alternative for the dead-end-check processing. Problem : That wouldn't allow to remove the existing code, it just adds complexity with the risk of breaking existing installations.
Without the legacy code I would like the idea.

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Mike Baggaley <mike at tvage.co.uk>
Gesendet: Montag, 8. Juni 2020 23:45
An: 'Development list for mkgmap'
Betreff: Re: [mkgmap-dev] is_in() function for point on line

Hi Gerd, I've tried out the --add-pois-to-lines=mid option and it seems to work well, giving me something like a 5% performance improvement. There are a couple of errors in the documentation - it doesn't explain what 'other' does, and there's no option for 'inner' (should 'other' read 'inner' in the option?).  Also it says that the default is to create all possible POI - I assume this is the default if the option is provided with no value and that the default is for no POIs to be added.

You haven't mentioned whether the patch also allows the creation of POIS from the lines style file - I'm assuming it is just the command line.

Thanks for the quick implementation,
Mike

-----Original Message-----
From: Gerd Petermann [mailto:gpetermann_muenchen at hotmail.com]
Sent: 03 June 2020 08:40
To: Mike Baggaley <mike at tvage.co.uk>; 'Development list for mkgmap' <mkgmap-dev at lists.mkgmap.org.uk>
Subject: Re: [mkgmap-dev] is_in() function for point on line

Hi Joris, Hi Mike,

I've implemented the suggested improvements with r4507. Please suggest improvements for the documentation changes.
I did not yet document the new prefix mkgmap:from-node: in the style because I found no simple description. Feel free to suggest something.
Reg. the implementation:
I decided to ignore all tags with the prefix mkgmap: , e,g. those from the LocationHook like mkgmap:admin_level2. It might happen that the values for the node are different to those copied from the way, but I don't assume that anybody wants to add extra rules for that.
Maybe I should change it the other way around so that the generated POI gets the "exact" values from the node (without the mkgmap:from-node: prefix?

The code should also work for POIs added with the --add-pois-to-areas option when the generated POI is using the location of a node because of the --pois-to-areas-placement option.

Gerd

________________________________________
Von: Mike Baggaley <mike at tvage.co.uk>
Gesendet: Montag, 1. Juni 2020 12:34
An: 'Gerd Petermann'; 'Development list for mkgmap'
Betreff: RE: [mkgmap-dev] is_in() function for point on line

Hi Gerd,

I have --add-pois-to-lines in my command, but right near the beginning of my points file I have

mkgmap:line2poitype=start | mkgmap:line2poitype=inner | mkgmap:line2poitype=end {deletealltags} # only centre of line wanted

I wonder whether there could be some efficiency improvement by extending the --add-pois-to-lines option to be something like

--add-pois-to-lines[=start|end|inner|mid[,...]]

We would then reduce the number of unwanted POIs created.

I actually only use this for placing bridge and tunnel names at the middle of the way and replacing line gates with point gates, so being able to do this in the lines file of my style sheet would be even better:

tunnel=* | bridge=* | barrier=gate {set mkgmap:addpoistoline=mid}

This assumes that generating POIs happens after the line stylesheet is processed.

Regards,
Mike

-----Original Message-----
From: Gerd Petermann [mailto:gpetermann_muenchen at hotmail.com]
Sent: 01 June 2020 10:50
To: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
Subject: Re: [mkgmap-dev] is_in() function for point on line

Hi Joris,

probably my ideas don't work. Your case is very special because you render the POI only when it is part of a way. The example with entrance=main is probably not a good one as entrance=main & is_in(building,mall,on) [...] already works.
I've coded a better patch so that your example will also work when option --link-pois-to-ways is not used. Up to now the patch depends on that.

I do not yet see how mkgmap can avoid the generation of all useless POI without adding a lot of code to analyse the rules in the points file, but I still think about this. Problem is that we have the generated tags like mkgmap:admin_level10 or mkgmap:residential and I fear to break something that uses tricky rules with just these tags.

> Or we could work the other way around, in stead of render all poi's for all objects and not use them, the mapmaker specifies the explicit objects the poi's are needed.
Yes, might be an option. An exclude list might also work.

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Joris Bo <jorisbo at hotmail.com>
Gesendet: Montag, 1. Juni 2020 10:51
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] is_in() function for point on line

Hallo Gerd,

This works perfect ! (points - file)
mkgmap:from-node:highway = turning_circle & highway = track             [0x641a resolution 24]
mkgmap:from-node:highway = turning_circle & highway != track            [0x641b resolution 24]

Your suggestion will indeed also feature your 'entrance=main' example which is actually a very useful one to avoid cluttering The "add-poi-to-lines" option is still usefull I think ?
For example
- to give a bicycle route a name-poi
- render a symbol for the line leisure = slipway
- render a cable car symbol for aerialway = gondola
- a warning symbol at the highway = via_ferrata
- or incline symbols at the start of steep mountain highways?

If those could be solved without rendering thousands of poi's for complex multipolygon lines that would be great of course.
Or we could work the other way around, in stead of render all poi's for all objects and not use them, the mapmaker specifies the explicit objects the poi's are needed.

Kind regards
Joris

-----Oorspronkelijk bericht-----
Van: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> Namens Gerd Petermann
Verzonden: zondag 31 mei 2020 11:38
Aan: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
Onderwerp: Re: [mkgmap-dev] is_in() function for point on line

Hi Joris,

up to now we only mentioned nodes with a highway=* tag used in ways with a highway=* tag.
I've coded a rather simple patch for that, but while I did that I wondered why this should be limited to highways.
I decided to use the prefix "mkgmap:from-node:". Tags with prefix mkgmap:  are not copied.

My understanding is that you want a function which tells you if a tagged node is used in one or more ways with a given tag.
For example, you may want to render a node with entrance=main, but only when it is the entrance to a building=mall.
This will not work, because only nodes with the tags barrier or highway are treated special.

I think it is possible to implement this as well. We probably just have to check the points rules and look for uses of "mkgmap:from-node:".
Maybe this would also allow to get rid of the computational overhead produced by the add-poi-to-lines option. The current implementation of this option is really ugly as it computes lots of points which are never used to generate any map object. Think of a complex natural=wood polygon (no other tags). If the points file doesn't contain a rule for natural=wood there is no need to generate the POIs.

I've uploaded a binary compiled with the attached patch here:
http://files.mkgmap.org.uk/download/473/mkgmap.jar

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Joris Bo <jorisbo at hotmail.com>
Gesendet: Sonntag, 31. Mai 2020 09:06
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] is_in() function for point on line

Hi Gerd

Thanks for the answer, always nice if it makes sense again. Then it was worth the testing 😊
For me that would be perfect, the number of cases this will occur is limited so that it's no problem to explicitly test against this replace value. There are probably more situations around highways where this could occur like highway = bus_stop, highway = speedcamera, and traffic signals.

Kind regards,
Joris

-----Oorspronkelijk bericht-----
Van: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> Namens Gerd Petermann
Verzonden: zaterdag 30 mei 2020 18:39
Aan: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
Onderwerp: Re: [mkgmap-dev] is_in() function for point on line

Hi Joris,
yes, it is a special case. The generated POI gets the tags from the way, so highway=turning_circle is replaced by highway=track.
The node 7131748927 appears only with the tag highway=turning_circle (and maybe the address tags)

Not sure what to do. Maybe store the replaced tag(s) with a prefix, e.g. mkgmap:replaced:highway=turning_circle in this case?

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Joris Bo <jorisbo at hotmail.com>
Gesendet: Samstag, 30. Mai 2020 18:11
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] is_in() function for point on line

Hi Gerd

Could it be possible that in this case the highway tag of the line is ignored because the point turning_circle is of type highway as well?

It's like there is a point {highway = turning_circle} on a line {highway =  track} It looks like I cannot catch it with the mkgmap:line2poitype and mkgmap:line2poi.

Somewhere else I already use
leisure = slipway & mkgmap:area2poi != true & mkgmap:line2poitype = mid  and that works fine

Kind regards,
Joris

-----Oorspronkelijk bericht-----
Van: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> Namens Gerd Petermann
Verzonden: zaterdag 30 mei 2020 17:51
Aan: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
Onderwerp: Re: [mkgmap-dev] is_in() function for point on line

Hi Joris,

is_in() only works with polygons. You can use --add-pois-to-lines, but be aware that it produces a lot of points for which you probablly don't want to add an object to the map. See the help for the details.

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Joris Bo <jorisbo at hotmail.com>
Gesendet: Samstag, 30. Mai 2020 17:40
An: Development list for mkgmap
Betreff: [mkgmap-dev] is_in() function for point on line

Hi

Does the is_in() function should also work for a point on a line in stead of in a polygon?

highway = turning_circle & is_in(highway, track, in_or_on) = true      [0x641a resolution 24]

It returns false for this situation where the turningcircle is attached to a highway = track.
(mkgmap:cache_is_in_POINT_highway_track_in_or_on=false)

Would there be another way to know if the point is attached to a highway = track rather then a for example highway = residential?
I use add-pois-to-lines, and link-pois-to-ways but as far as I can see, that doesn't give me more information.

https://www.openstreetmap.org/#map=18/49.65888/5.45845
[cid:image002.jpg at 01D636A9.5643B8A0]

Kind regards
Joris

_______________________________________________
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
_______________________________________________
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

_______________________________________________
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
_______________________________________________
mkgmap-dev mailing list
mkgmap-dev at lists.mkgmap.org.uk
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev