Re: [Tagging] Splitting places and hosted devices in mapping

[Johnparis]

Thanks for the explanation. I see your point.

On Sat, Jun 8, 2019 at 7:10 PM François Lacombe 
wrote:

> Hi Colin,
>
> Le sam. 8 juin 2019 à 18:53, Colin Smale  a écrit :
>
>> If we want to describe the overall transfer function of the whole
>> "building", then we need to describe the input and output characteristics,
>> and some aspects of the path from input to output. Just stating a couple of
>> voltages doesn't do justice to the potential complexity of the overall
>> function of the "building".
>>
>
> That's right and it can be completed with :
> windings=*, phases=*, rating=* and frequency=*
> Some of them can be extended with primary, secondary, tertiary and so on
> according to interfaces
> https://wiki.openstreetmap.org/wiki/Tag:power%3Dtransformer
>
> As it's pretty complex tagging, I'd be in favour to not mix that with
> building/area properties
>
> All the best
>
> François
>
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
>
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Splitting places and hosted devices in mapping

[François Lacombe]

Hi Colin,

Le sam. 8 juin 2019 à 18:53, Colin Smale  a écrit :

> If we want to describe the overall transfer function of the whole
> "building", then we need to describe the input and output characteristics,
> and some aspects of the path from input to output. Just stating a couple of
> voltages doesn't do justice to the potential complexity of the overall
> function of the "building".
>

That's right and it can be completed with :
windings=*, phases=*, rating=* and frequency=*
Some of them can be extended with primary, secondary, tertiary and so on
according to interfaces
https://wiki.openstreetmap.org/wiki/Tag:power%3Dtransformer

As it's pretty complex tagging, I'd be in favour to not mix that with
building/area properties

All the best

François
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Splitting places and hosted devices in mapping

[Colin Smale]

On 2019-06-08 18:08, François Lacombe wrote:

> Specifically on the topic, a converter hall will never have only one feature 
> inside : you'll find the converter, but cooling and auxiliary power stuff 
> also, and all that world should get individual features 
> 
>> By the way, your last example: 
>> No tag specifies the output voltage in your example. As I read other related 
>> pages, I think it should be tagged: 
>> voltage:primary=63000 
>> voltage:secondary=1500
> 
> The example deals with the substation, not with the involved devices (and 
> voltage:primary is a device property)
> power=converter objects will certainly have voltage:primary and 
> voltage:secondary, not the substation (and mixing them would make 
> power=substation and voltage:primary apparently compatible while they aren't, 
> no exceptions) 
> Furthermore, tagging voltage on the substation is a bad idea, only useful to 
> ease renders assuming the size of the facility based on the highest voltage 
> involved.

There are all kinds of converter functions, and changing the voltage is
only one of them. Primary and secondary are only really applicable to
certain types of transformers. Each "converter" has its own transfer
function, and I don't think it would be a good idea to limit the
discussion to a single class of converter. 

Transformers come in shapes and sizes; besides changing voltage, they
can change the number of phases, upset the power factor, and do phase
shifting. Rectifiers are similarly diverse. Motor-generators can do all
of the above, and change the frequency as well. Even if the output
characteristics are equal to the input, there may be useful things going
on, such as decoupling or PF correction. 

If we want to describe the overall transfer function of the whole
"building", then we need to describe the input and output
characteristics, and some aspects of the path from input to output. Just
stating a couple of voltages doesn't do justice to the potential
complexity of the overall function of the "building".___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Splitting places and hosted devices in mapping

[Mateusz Konieczny]

7 Jun 2019, 19:48 by ok...@johnfreed.com:

> I believe the key phrase is "single-use" in the sentence you cite. That is, 
> for instance, if you have a building that is a café (tagged as building=yes) 
> with a node inside it with the details of the café (name=Rick's Cafe; 
> amenity=cafe), then you should move the tags from the node to the area and 
> eliminate the node.
>
It is commonly done but I would not recommend this.

It is causing problems because there is no good way to distinguish whatever 
tags apply
to building, POI or both.

For examples of problems see:
https://github.com/gravitystorm/openstreetmap-carto/issues/3487 

https://github.com/gravitystorm/openstreetmap-carto/issues/3472
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Splitting places and hosted devices in mapping

[François Lacombe]

Le sam. 8 juin 2019 à 17:48, Johnparis  a écrit :

> I disagree with the premise. There is no confict at all between building=*
> and power=*. See https://www.openstreetmap.org/way/142321228 for a
> typical case.
>
Not between building=* and power=* but with power=transformer and
building=* (for instance).
Building is a site/place/facility and power=transformer is a device, just
like power=converter.
power=substation and building=* are perfectly compatible on the other hand.


> This is a very typical situation where there is "stuff inside" the
> building. In this case, it's a single-use building -- a power converter
> building. Yes there are one or more devices inside the building, but that
> doesn't change the building's single use. So no need for guessing where
> nodes might be needed inside the building for the converter device(s). Just
> tag the building itself. (And there doesn't have to be a building -- if
> it's outdoors, just tag the enclosing area.) If you learn more about the
> nature of the converters, you could (for example) have a tag like
> "converter:count=3".
>
I disagree because converters may have different capabilities and figures,
mixing them with the building could lead in a kind of mess.
The global idea of not mixing building and devices unless they are on the
same node really ease the QA and consuming process : you'll get building
and devices separately.
Having rules for single device different from several devices in a building
need to re-qualify building when a device changes whereas the building
hasn't changed in reality.
Mixing devices and buildings make device properties compatible with
building ones while they aren't (and mapping practices are often taken as
truth here). This point is particularly important.
And so on...

Specifically on the topic, a converter hall will never have only one
feature inside : you'll find the converter, but cooling and auxiliary power
stuff also, and all that world should get individual features


> By the way, your last example:
> No tag specifies the output voltage in your example. As I read other
> related pages, I think it should be tagged:
> voltage:primary=63000
> voltage:secondary=1500
>

The example deals with the substation, not with the involved devices (and
voltage:primary is a device property)
power=converter objects will certainly have voltage:primary and
voltage:secondary, not the substation (and mixing them would make
power=substation and voltage:primary apparently compatible while they
aren't, no exceptions)
Furthermore, tagging voltage on the substation is a bad idea, only useful
to ease renders assuming the size of the facility based on the highest
voltage involved.

I know this is a really particular topic, but I try to make it more easy to
handle by making feature separated as to not get the whole bunch at one
time.

All the best

François
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Splitting places and hosted devices in mapping

[Johnparis]

I looked at the actual proposal and I think on the Talk page here is a
problem I see. You write:

Since power =converter
 isn't
compatible with building 
=* (a converter is a device and building isn't), it's less possible and
desirable to use building 
=* + power =converter
 + note
=There should be several
converters in this building
.
How could we feed a reliable QA test with a free text note? Fanfouer
 (talk
) 23:01, 16 May
2019 (UTC)

I disagree with the premise. There is no confict at all between building=*
and power=*. See https://www.openstreetmap.org/way/142321228 for a typical
case.

This is a very typical situation where there is "stuff inside" the
building. In this case, it's a single-use building -- a power converter
building. Yes there are one or more devices inside the building, but that
doesn't change the building's single use. So no need for guessing where
nodes might be needed inside the building for the converter device(s). Just
tag the building itself. (And there doesn't have to be a building -- if
it's outdoors, just tag the enclosing area.) If you learn more about the
nature of the converters, you could (for example) have a tag like
"converter:count=3".

By the way, your last example:

A traction substation in Paris designed with three 63kV AC outdoor
transformers and with DC converters inside the building (we don't know how
many of them). The railway runs with 1500 Volts DC.

No tag specifies the output voltage in your example. As I read other
related pages, I think it should be tagged:
voltage:primary=63000
voltage:secondary=1500

Cheers,

John



On Sat, Jun 8, 2019 at 5:07 PM Johnparis  wrote:

> I agree with Marc that you should never "create nodes at a random position
> with the equipment to avoid the tag for the characteristic". If you place a
> node, it should reflect as closely as possible the actual position,
> although if the position is uncertain, it's typical in OSM to place a node
> rather than an area (closed way). If it's just a sheer guess, however, the
> node should be avoided. And if there are "a lot of these things here", then
> that argues even more strongly for making it a characteristic of the
> enclosing area, rather than a node. Think of hotel rooms -- rather than
> mapping each room (as an area or, shudder, a node), you simply tag the
> hotel with "rooms=35".
>
> As to the example of a bus shelter with a bench, I personally favor making
> it a characteristic of the shelter (as Marc suggests), rather than placing
> a separate node, because the bench would not exist except for the shelter.
> If there is a bench NEAR the shelter, but outside of it, I would make that
> a separate node. But that is my personal preference, I occasionally
> encounter shelters mapped as areas with nodes for benches inside. It's not
> my style, but on the other hand I don't change it if someone else has done
> it. I do add the "bench=yes" tag, however, to the shelter itself, as that's
> useful for people searching for shelters-with-benches.
>
> In general, I try to think of a typical end-user case. What will make OSM
> most helpful for the person who wants to use it? (not: what will make
> things easiest for the mapper?)
>
> John
>
>
>
>
> On Sat, Jun 8, 2019 at 4:46 PM François Lacombe 
> wrote:
>
>> Thank you two for the elaborated answers
>>
>> I agree that currently many mapping practices split sites and devices.
>> Then the idea should spread as to not get in arguments like the one in
>> the proposal.
>>
>> Nevertheless I have comments regarding Marc examples :
>> * " bin/shelter/bench=yes on a bus stop for this equipment is present
>> in the area of the bus stop"
>> Shouldn't we put a node inside the bus-stop shelter area to materialize
>> the bin or the bench?
>> * lit=yes on an highway feature properly indicates kind of "process"
>> without be redundant with individual street lights lighting the road
>> This is exactly with I propose for traction substations, mapping process
>> on sites and devices independently.
>>
>> Things are better than I thought.
>> Then, it would be good to amend the 1 feature=1 osm object page with more
>> clear messages (and examples)
>> All of this discussion only cover sites mapped as areas. We can start by
>> stating that a particular device on the ground should get its own object
>> instead of being moved to enclosing building/site/place ?
>> This is not possible to 

Re: [Tagging] Splitting places and hosted devices in mapping

[François Lacombe]

Le sam. 8 juin 2019 à 17:08, Johnparis  a écrit :

> I agree with Marc that you should never "create nodes at a random position
> with the equipment to avoid the tag for the characteristic". If you place a
> node, it should reflect as closely as possible the actual position,
> although if the position is uncertain, it's typical in OSM to place a node
> rather than an area (closed way). If it's just a sheer guess, however, the
> node should be avoided. And if there are "a lot of these things here", then
> that argues even more strongly for making it a characteristic of the
> enclosing area, rather than a node.
>
I agree too and that's why I proposed the conversion=* tag on substations
perimeters: it avoids anyone to be tempted to add random nodes and provide
sustainable tagging to indicate that a particular process is operated even
if he doesn't know (or doesn't have time to map or whatever) the equipment
involved.


> Think of hotel rooms -- rather than mapping each room (as an area or,
> shudder, a node), you simply tag the hotel with "rooms=35".
>
That's ok because you don't need to add information about each room.
It feels the same for parking capacity (and you assume that each slot or
room are equivalent).

As to the example of a bus shelter with a bench, I personally favor making
> it a characteristic of the shelter (as Marc suggests), rather than placing
> a separate node, because the bench would not exist except for the shelter.
> If there is a bench NEAR the shelter, but outside of it, I would make that
> a separate node. But that is my personal preference, I occasionally
> encounter shelters mapped as areas with nodes for benches inside. It's not
> my style, but on the other hand I don't change it if someone else has done
> it. I do add the "bench=yes" tag, however, to the shelter itself, as that's
> useful for people searching for shelters-with-benches.
>
Agreed because often, bench and shelters are the same piece of furniture.
seat=yes would be better imho.
Moving enclosed equipment on the perimeter should be done in term of
process or function and not in term of devices (i.e don't use the same tag
on the equipment and on the enclosing feature).
Other examples could be "A pump station vs individual pumps" or "A factory
vs individual machines"...

In general, I try to think of a typical end-user case. What will make OSM
> most helpful for the person who wants to use it? (not: what will make
> things easiest for the mapper?)
>
Since the end-use could be any consumer, QA tests analysis, Render
software... we have to provide the most objective data and reduce error
prone situations

François
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Splitting places and hosted devices in mapping

[Johnparis]

I agree with Marc that you should never "create nodes at a random position
with the equipment to avoid the tag for the characteristic". If you place a
node, it should reflect as closely as possible the actual position,
although if the position is uncertain, it's typical in OSM to place a node
rather than an area (closed way). If it's just a sheer guess, however, the
node should be avoided. And if there are "a lot of these things here", then
that argues even more strongly for making it a characteristic of the
enclosing area, rather than a node. Think of hotel rooms -- rather than
mapping each room (as an area or, shudder, a node), you simply tag the
hotel with "rooms=35".

As to the example of a bus shelter with a bench, I personally favor making
it a characteristic of the shelter (as Marc suggests), rather than placing
a separate node, because the bench would not exist except for the shelter.
If there is a bench NEAR the shelter, but outside of it, I would make that
a separate node. But that is my personal preference, I occasionally
encounter shelters mapped as areas with nodes for benches inside. It's not
my style, but on the other hand I don't change it if someone else has done
it. I do add the "bench=yes" tag, however, to the shelter itself, as that's
useful for people searching for shelters-with-benches.

In general, I try to think of a typical end-user case. What will make OSM
most helpful for the person who wants to use it? (not: what will make
things easiest for the mapper?)

John




On Sat, Jun 8, 2019 at 4:46 PM François Lacombe 
wrote:

> Thank you two for the elaborated answers
>
> I agree that currently many mapping practices split sites and devices.
> Then the idea should spread as to not get in arguments like the one in the
> proposal.
>
> Nevertheless I have comments regarding Marc examples :
> * " bin/shelter/bench=yes on a bus stop for this equipment is present
> in the area of the bus stop"
> Shouldn't we put a node inside the bus-stop shelter area to materialize
> the bin or the bench?
> * lit=yes on an highway feature properly indicates kind of "process"
> without be redundant with individual street lights lighting the road
> This is exactly with I propose for traction substations, mapping process
> on sites and devices independently.
>
> Things are better than I thought.
> Then, it would be good to amend the 1 feature=1 osm object page with more
> clear messages (and examples)
> All of this discussion only cover sites mapped as areas. We can start by
> stating that a particular device on the ground should get its own object
> instead of being moved to enclosing building/site/place ?
> This is not possible to distinguish sites from devices at the moment they
> are located on a single node.
>
> All the best
>
> François
>
> Le sam. 8 juin 2019 à 01:06, marc marc  a
> écrit :
>
>> Hello,
>>
>> Le 07.06.19 à 19:08, François Lacombe a écrit :
>> > a need to distinguish sites and devices in our tagging.
>>
>> making a difference between a characteristic of a site and the device
>> providing that characteristic is already what we do for many objects.
>> below are some examples.
>> I find it quite logical and useful to inform therefore that a site
>> has a feature with another tag than the device of this feature.
>> in that sense I find your proposal quite coherent.
>> I completely disagree with the idea of a comment on the proposal
>> to create nodes at a random position with the equipment to avoid
>> the tag for the characteristic. of course,
>> it's better to be able to map each device separately, but not
>> by inventing their position and number just to need one less tag
>>
>> some exemples :
>> bin/shelter/bench=yes on a bus stop for this equipment is present
>> in the area of the bus stop
>> bar=yes on a POI to say that there is a bar in the area of the POI
>> lit=yes on a road to say that there are lighting devices nearby.
>> tactile_paving on the node of a pedestrian crossing to indicate
>> that there is a tactile_paving at the intersection of the pedestrian
>> path and the kerb
>> toilets and toilets:wheelchair on a POI
>>
>> Regards,
>> Marc
>> ___
>> Tagging mailing list
>> Tagging@openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/tagging
>>
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
>
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Splitting places and hosted devices in mapping

[François Lacombe]

Thank you two for the elaborated answers

I agree that currently many mapping practices split sites and devices.
Then the idea should spread as to not get in arguments like the one in the
proposal.

Nevertheless I have comments regarding Marc examples :
* " bin/shelter/bench=yes on a bus stop for this equipment is present
in the area of the bus stop"
Shouldn't we put a node inside the bus-stop shelter area to materialize the
bin or the bench?
* lit=yes on an highway feature properly indicates kind of "process"
without be redundant with individual street lights lighting the road
This is exactly with I propose for traction substations, mapping process on
sites and devices independently.

Things are better than I thought.
Then, it would be good to amend the 1 feature=1 osm object page with more
clear messages (and examples)
All of this discussion only cover sites mapped as areas. We can start by
stating that a particular device on the ground should get its own object
instead of being moved to enclosing building/site/place ?
This is not possible to distinguish sites from devices at the moment they
are located on a single node.

All the best

François

Le sam. 8 juin 2019 à 01:06, marc marc  a écrit :

> Hello,
>
> Le 07.06.19 à 19:08, François Lacombe a écrit :
> > a need to distinguish sites and devices in our tagging.
>
> making a difference between a characteristic of a site and the device
> providing that characteristic is already what we do for many objects.
> below are some examples.
> I find it quite logical and useful to inform therefore that a site
> has a feature with another tag than the device of this feature.
> in that sense I find your proposal quite coherent.
> I completely disagree with the idea of a comment on the proposal
> to create nodes at a random position with the equipment to avoid
> the tag for the characteristic. of course,
> it's better to be able to map each device separately, but not
> by inventing their position and number just to need one less tag
>
> some exemples :
> bin/shelter/bench=yes on a bus stop for this equipment is present
> in the area of the bus stop
> bar=yes on a POI to say that there is a bar in the area of the POI
> lit=yes on a road to say that there are lighting devices nearby.
> tactile_paving on the node of a pedestrian crossing to indicate
> that there is a tactile_paving at the intersection of the pedestrian
> path and the kerb
> toilets and toilets:wheelchair on a POI
>
> Regards,
> Marc
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
>
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Splitting places and hosted devices in mapping

[marc marc]

Hello,

Le 07.06.19 à 19:08, François Lacombe a écrit :
> a need to distinguish sites and devices in our tagging.

making a difference between a characteristic of a site and the device 
providing that characteristic is already what we do for many objects.
below are some examples.
I find it quite logical and useful to inform therefore that a site
has a feature with another tag than the device of this feature.
in that sense I find your proposal quite coherent.
I completely disagree with the idea of a comment on the proposal
to create nodes at a random position with the equipment to avoid
the tag for the characteristic. of course,
it's better to be able to map each device separately, but not
by inventing their position and number just to need one less tag

some exemples :
bin/shelter/bench=yes on a bus stop for this equipment is present
in the area of the bus stop
bar=yes on a POI to say that there is a bar in the area of the POI
lit=yes on a road to say that there are lighting devices nearby.
tactile_paving on the node of a pedestrian crossing to indicate
that there is a tactile_paving at the intersection of the pedestrian 
path and the kerb
toilets and toilets:wheelchair on a POI

Regards,
Marc
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Splitting places and hosted devices in mapping

[Johnparis]

I believe the key phrase is "single-use" in the sentence you cite. That is,
for instance, if you have a building that is a café (tagged as
building=yes) with a node inside it with the details of the café
(name=Rick's Cafe; amenity=cafe), then you should move the tags from the
node to the area and eliminate the node.

I don't think this applies to the case you are looking at. Even if the
building itself might be thought of as "single use", the devices inside
them per se make it "multiple-use" as it is understood on the "One feature,
one element" page. And as with a multiple use building, you should use
"point or area objects representing the locations of what is in them". So I
don't see a conflict there. Just my opinion.

John


On Fri, Jun 7, 2019 at 7:09 PM François Lacombe 
wrote:

> Hi all,
>
> The recent work and talks on a proposal I wrote regarding traction
> substations show a need to distinguish sites and devices in our tagging.
> According to
> https://wiki.openstreetmap.org/wiki/One_feature,_one_OSM_element, I think
> it's a important point, especially when you have complex arrangement of
> indoor features in a single building for instance (malls, technical
> facilities, big buildings...)
>
> The proposal deals with converters (devices) and places/sites (building)
> hosting them, very technical thing in this particular situation. Some
> contributors asked to map one single feature for building and hosted
> devices as an area. To me it would be more valuable to draw the building
> and add nodes inside the perimeter to map the devices (even if there is one
> single main device inside the building)
>
> Proposal :
> https://wiki.openstreetmap.org/wiki/Proposed_features/Traction_substations_extension
> (See Converters hidden in building chapter on corresponding talk page)
>
> This would also be the situation for any furniture in a shop or in a
> library, any amenity in a given public place...
>
> I want to discuss about the sentence here :
> https://wiki.openstreetmap.org/wiki/One_feature,_one_OSM_element
> "An area object representing a single-use building with a point object
> inside it. If you find this, move the tags to the area object and delete
> the point." described as a bad situation.
>
> I find this inconsistent with the purpose of the page principle, not to
> mention it doesn't reflect to what is seen in the situation of a given
> apparatus installed inside a building (and not built in place as a building
> we need to map both situations separately).
> It also makes hard to follow a proper life cycle for both hosted and
> hosting feature when they are mixed, not to mention headaches to get to
> which feature a particular tag refers.
>
> I didn't remember of a similar discussion here, and I'll be pleased to
> read your opinions about this topic.
> This would surely make my proposal better and may allow to complete the 1
> feature=1 osm element page.
>
> All the best
>
> François
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
>
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging