Layer: Existing Infrastructure

[mem48]

Variables in the DB

Table: roads

osmid INT NULL, - OSM ID number (non unique due to segment splitting)
name VARCHAR(255) NULL, - road name (truncated to 255 characters)
ref VARCHAR(20) NULL, - road reference e.g. M25 (truncated to 20 character)
highway VARCHAR(20) NULL, - road type
junction VARCHAR(20) NULL, - junction (mostly roundabout or no)
elevation VARCHAR(8) NULL, - bridge / tunnel / no
maxspeed SMALLINT NULL, - in mph as number (note different to OSM where kph is default)
segregated VARCHAR(3) NULL, - yes / no
rtid INT NULL, - ID to link to road type table

Table: roadtypes (see https://github.com/cyipt/cyipt-bigdata/blob/master/forDB/roadtypes.csv )

rtid INT PRIMARY KEY, - unique ID
roadtype VARCHAR(100) NULL, - description of road, and cycling permissions
onewaysummary VARCHAR(40) NULL, summary of one way rules
sidewalk VARCHAR(20) NULL, - left / right / both / no / not applicable
cyclewayleft VARCHAR(20) NULL, - lane / track / share_busway / no
lanespsvforward SMALLINT NULL, - number of bus (public service vehicles) lanes
lanesforward SMALLINT NULL, - number of general use lanes
lanesbackward SMALLINT NULL, - number of general use lanes
lanespsvbackward SMALLINT NULL, - number of bus (public service vehicles) lanes
cyclewayright VARCHAR(20) NULL - lane / track / share_busway / no

[mem48]

Suggested UI

1. Cycle Infrastructure map, filtered for: cyclewayleft or cyclewayright is not no OR certain types of highway e.g. roadtype == Segregated Cycleway | Segregated Shared Path | Cycleway | Living Street

this is basically the current view except it removed roadtype == Shared Paths, which is basically all the blue paths that show up all over the place.

I'd suggest a colour scheme with a dashed line based on cyclewayleft and cyclewayright right so that "lane no" looks the same as "no lane" then solid colours for the roadtypes

2. Road speed map, when no speed is provided in the OSM CyIPT guesses the speed, so it would be useful to see what CyIPT is using and would tie in well wit the edit the osm button

3. Footways Layer - using the sidewalk tag, again CyIPT guesses when no data is provided so would be useful to see what CyIPT is using, also relevant to road width calcualtions

[jf1]

This layer has great potential to enable people to comprehend and improve OSM. I considered it over the weekend, before reading @mem48's comments here. My proposal below could be used to combine your suggestions for 1. Cycle Infrastructure map and 2. Footways Layer, or the proposed categories and colours below could be displayed on two different map views (or three: road-only; footway-only; combined).

I'm not sure what "solid colours for the roadtypes" means exactly. Please note that my proposal expands the spec to introduce two (red, orange) or more (+grey, black) colours for roads without infra, which would result in all roads being drawn.

I agree dashed lines would be useful and wonder if two-colour dashed lines are possible by drawing two lines with complimentary dashOffset values. This would enable eg track lane to be shown accurately. Once two-colour dashed lines are possible then this key scheme could be implemented:

My braindump of types and proposed colours is in this google sheet.
15 existing types would use 3 colours instead of the current 8, then the two other existing types (no infra) would show road category using one of the 4 colours mentioned above. I believe all of my Type values can be linked to your roadtypes.csv.
Screenshot from this spreadsheet below - you're welcome to insert a table into this issue or elsewhere on github if either are possible.

If thick and thin lines can be displayed on the same map then I'd prefer to see roads and off-highway provision together. Additionally, I think circles would be a good way to highlight access points and crossings.

nb. I'm very keen for CyIPT to use a Red-Yellow-Blue scale beyond this proposal for the existing infra layer, even if you prefer different shades. Adam Reynolds and I trialled various options, considering colour-blindness and user-friendliness, before settling on this. "Blue is best, red is bad" seems intuitive. We then added black for 'worse than red' and I think it's practical to also use a shade of grey, eg for 'no value specified' items.
We found the following sites useful:
http://colorbrewer2.org/#type=diverging&scheme=RdYlBu&n=7
http://www.color-blindness.com/coblis-color-blindness-simulator/

ps. The CycleStreets Cycleability ratings layer has similar potential to enable better comprehension of and contributions to OSM data, and I feel it would similarly benefit from an intuitive set of buckets and colours. I'll create an issue there that links to this one.

[mvl22]

Popups now dependent on the value of the layer parameter; implemented in 13225d9 and new upstream library support in cyclestreets/Leaflet.LayerViewer@378fc2c.

[mvl22]

Once two-colour dashed lines are possible then this key scheme could be implemented

This is mostly in place now (the simplified version with 7 values).

The dashOffset concept will need quite a bit more work, and I've not fully checked feasibility yet.

[mem48]

Great work, the cycle infrastructure is good enough for the workshop, the speed limits and footways still need work. I've noticed that CyIPT is overly pessimistic about the presence of footways, which is fixable but not before the workshop

[mem48]

Also, I think the API filters remove the possibility of Minor/Major roads every showing up on the map. So you could remove them from the legend

[jf1]

Thanks @mvl22 for the work and notification.

I think the API filters remove the possibility of Minor/Major roads every showing up on the map. So you could remove them from the legend

If they are suppressed for now then I hope they'll be easy to reinstate later. Seeing how tools like PCT, CyIPT and cyclestreets understand our local highways network is crucial if we're to have confidence in their output, and to enable us to fix OSM errors or raise issues about any strange behaviour by the tools.

I agree this layer is good enough for the workshop, however you may wish to address the below before or after it...

Straight lines (which to me look very strange) at zoom 15 or lower:

Accurate geometries at zoom levels 16 or higher:

[mvl22]

Also, I think the API filters remove the possibility of Minor/Major roads every showing up on the map.

Yes, my understanding is that this is a map of anything with cycle infrastructure, so those don't show.

So you could remove them from the legend

OK.

[mvl22]

Straight lines (which to me look very strange) at zoom 15 or lower:

@mem48 Can you look at that? Pretty sure this is not the API mechanics, as that's just relaying what's been put in the database.

Looking at the example I suspect the curvature of that path means it's going outside a matching buffer.

[mvl22]

the speed limits and footways still need work

Can anyone suggest a colour scheme for both of these?

[mvl22]

Can anyone suggest a colour scheme for both of these?

I've put one in - this stuff is now in place.

[mem48]

Straight lines (which to me look very strange) at zoom 15 or lower:

This is the SQL calls to st_simplify reducing the amount of data returned when making large requests. So it is a feature, not a bug. However, it is causing some confusion so perhaps needs revising. I will look a producing a table of zoom level to simplification which produces more pleasing results.

[mem48]

Can anyone suggest a colour scheme for both of these?

Speed limits seem to be missing colours for 10, 50, 60 mph
Perhaps blue for 10 mph (as probably an of road area) and shades of red for the 50/60 mph

[mem48]

For footways, missing the right option (uncommon but possible) I'd suggest using the same colour as left as the key concept is only on one side of the road and the OSM left/right side concept is not immediately intuitive.

[jf1]

Shared-use footways are being displayed as 'Track' which wasn't the intention in my suggestion. Not sure if this is deliberate or not but I feel these are two very different types of provision that CyIPT should be clearly differentiating.

FWIW I think it's a shame the major/minor road classification isn't shown anymore as it was a useful way to help comprehend why CyIPT/PCT etc produce certain results, potentially highlighting issues to address with OSM data, the tools or the LA's road categorisation.

ps. For now I'd suggest footways on one side are displayed in the same way as when both are present. Once the dashOffset technique can be implemented then any single-sided footway could be represented as a dashed line. This would be consistent with displaying any single-sided infrastructure as a dashed line.

[mvl22]

Speed limits seem to be missing colours for 10, 50, 60 mph

I took the view that this gradation is more than is needed for cycling. Max speeds 0-20 are good, 30 verging on problematic, and more than that too fast.

More inflection points can easily be added at:
https://github.com/cyipt/cyipt-website/blob/master/js/cyipt.js#L207-L212

[mvl22]

Shared-use footways are being displayed as 'Track'

OSM has no concept of shared-use footways, unfortunately. We only know that something is a footway or a cycleway, each of which may have foot/bicycle=yes/no, and the issue of what is a cycleway vs a footway that can be cycled on is disputed.

[jf1]

Thanks for that clarification @mvl22 - perhaps something to touch on on Friday and consider thereafter.