[PATCH] Categorize 'Way end node near other highway' warnings.

[maxerickson@…]

In areas with (for example) poorly drawn sidewalks, the results of the 'Way end node near other highway' validation are dominated by footways that are not connected to streets or service roads. Fixing the sidewalks is not always a priority and it is anyway often better to contact the original mapper and encourage them to adjust what they are doing than it is to just fix the problem.

It would useful if the warnings were sorted into categories. I think major/minor would be enough, but it looks like one category per highway type (chosen from the way with the end node) would be easier to implement.

[reichg]

Can we get some examples of this please?

[reichg]

Can you explain how you think these warnings should be categorized? We can try the major/minor outcome you desire, just need to get an idea of how these should be categorized.

Something like this??

Major: Motorway, Trunk, Primary
Minor: Any highway value below Primary?

[maxerickson@…]

To me the useful categories are probably something like pedestrian, bicycle, minor road and major road, with major road being tertiary and above (collector streets and highways where connectivity is more impactful) and minor road being everything else.

For just 2 categories it would be the collector streets and above for major and everything else for minor.

I typically would always fix major road issues and bicycle issues and am less likely to fix minor road and pedestrian issues.

[skyper]

I am often not happy with the sorting of validator results but I did not figure out how the best sorting would look like and it might depend on the test/rule itself and the users workflow. This is a more general problem and better discussed in a new ticket though.

Coming back to this ticket I would start with two categories, motorway to residential/unclassified and the second category everything from service on below.
Additionally, sorting the results could be done by name=*, highway=* and ref=* and even the location could be used if name and ref are missing to get all nearby problems in a group.

[reichg]

When we say categorize, can major/minor simply be a difference in severity (WARNING/ERROR)

[skyper]

Please, no error. How about just splitting into two warnings, one for each category.

[reichg]

Replying to skyper:

Please, no error. How about just splitting into two warnings, one for each category.

Yeah not a problem. The solution ended being very simple I think. Just get the highwayTag from the nearby end node's parent way. I check that tag against a set of priority highway tags -> priorityHighwayTags which consists of MOTORWAY, MOTORWAY_LINK, TRUNK, TRUNK_LINK, PRIMARY, PRIMARY_LINK, SECONDARY, SECONDARY_LINK, TERTIARY, TERTIARY_LINK, RESIDENTIAL, UNCLASSIFIED. If the set contains the highway tag then the warning is Major (just a different message).

[reichg]

Categorizes warning by end node's parent way highway tag.

[reichg]

Does the patch solution satisfy the ticket?

[Don-vip]

Milestone renamed

[GerdP]

I am not convinced this distinction makes much sense.

The patch doesn't work as intended when many issues are found as it changes the message string to "Major" for all following unconnected nodes.
Other minor issues:

• I don't like the idea of using toUpperCase() for tag values.
• Can't we use the existing list CLASSIFIED_HIGHWAYS in Highways.java? It also contains "living_street"?

My understanding is that the "- Major" suffix should be added if both highways are major, not just the one with the unconnected node? So, if a footway ends near a primary or a residential ends near a footway, both are minor issues, right?

[GerdP]

proposed solution

[reichg]

Replying to GerdP:

I am not convinced this distinction makes much sense.

The patch doesn't work as intended when many issues are found as it changes the message string to "Major" for all following unconnected nodes.
Other minor issues:

• I don't like the idea of using toUpperCase() for tag values.
• Can't we use the existing list CLASSIFIED_HIGHWAYS in Highways.java? It also contains "living_street"?

Tested your patch with these changes?

---

My understanding is that the "- Major" suffix should be added if both highways are major, not just the one with the unconnected node? So, if a footway ends near a primary or a residential ends near a footway, both are minor issues, right?

Works for me!