Adds barrier=gate regression tests, see #581 #3445
Conversation
| Scenario: Car - barrier=gate should be routed over unless explicitely forbidden | ||
| Then routability should be | ||
| | node/barrier | access | bothw | | ||
| | gate | | x | |
There was a problem hiding this comment.
This is the questionable default.
There was a problem hiding this comment.
So shouldn't it be the other way around? If you want to add the test, it should be as a todo and have the correct default value. Otherwise it might imply to someone that we want it to be this way.
There was a problem hiding this comment.
We want it to be this way. The test is just being explicit about this behavior.
Strictly speaking we're going against the Wiki's recommendation to not route over gates without any access tags. The reason I'm adding this (passing) test is to make a point for #3435.
Where data in the wild trumps the Wiki's recommendation by a huge amount of data points we can no longer call it "mis-tagging". Instead we should support what's there.
There was a problem hiding this comment.
👍 misunderstood the questionable default part here. Thought you where referring to what you added as strange.
|
well. i've seen barrier=gate more than once in places where access is indeed possible, like https://www.openstreetmap.org/edit?node=1227080468#map=19/55.69242/12.55155 |
|
This needs a rebase, current conflicts in master. |
daniel-j-h
force-pushed
the
gate-routability
branch
from
bef74d7 to
94854b5
Compare
See #581 and the recent discussion in #3435.
These tests all pass even though
barrier=*impliesaccess=noper Wiki.But we still route e.g. over
barrier=gatewithout any further access tags.