"Nodes with same name" test in validator is currently useless

[mkoniecz]

It is typical that there are multiple things with the same name, example:

​osmwww:browse/node/2496192091 - hamlet
​osmwww:browse/node/1791686895 - bus station in this hamlet - with the same name

It is typical that there is something more or less important and nearby are bus stations, railway stations, shops etc named after this.

I would suggest changing this test to "Nodes with same name and type" (with "the same type" meaning "are displayed with the same icon by JOSM default style") and exclude types of nodes that usually come in pairs or entire groups and have the same name (for start: amenity=bus_station). Maybe than it will catch something useful.

At least in my region this test is so useless that it is always the first one disabled after resetting validator settings.

slightly related to #3669

Repository Root: http://josm.openstreetmap.de/svn
Build-Date: 2013-12-28 02:34:32
Last Changed Author: simon04
Revision: 6554
Repository UUID: 0c6e7542-c601-0410-84e7-c038aed88b3b
URL: http://josm.openstreetmap.de/svn/trunk
Last Changed Date: 2013-12-28 01:31:47 +0100 (Sat, 28 Dec 2013)
Last Changed Rev: 6554

Identification: JOSM/1.5 (6554 en_GB) Windows 7 32-Bit
Memory Usage: 70 MB / 247 MB (15 MB allocated, but free)
Java version: 1.7.0_45, Oracle Corporation, Java HotSpot(TM) Client VM
Dataset consistency test: No problems found

Plugin: OpeningHoursEditor (30117)
Plugin: buildings_tools (30057)
Plugin: notes (v0.8)

[simon04]

Hm, even after the group by matching presets (attaching patch for reference), this test will detect false-positives: Think of amenities with the same name at different locations/villages, etc.

What about removing this test before trying to find some usefulness?

[mkoniecz]

"at different villages" - this one may not be a problem, I think that typical downloaded area in JOSM is small enough that it will not include multiple villages.

But yes, this test for me seems to be rather useless and I would not be against removing it. On the other hand - somebody considered it good enough to code and commit and not remove it, and I know that some of tests that I suggested are considered useless or even harmful by some people (see #9379) so I prefer to be careful with removing things.

[simon04]

Think of a few neighbour villages w/ 2k inhabitants each. I often work on such a data set.
Also, cities often have several McDonalds' – those shouldn't be detected as well.

I understand that people are against tests that generate warnings for common mapping situations (such as in #9379), but here we mostly remove loads of false-positives.

[stoecker]

Probably the life-time of this test is over. Some years ago we had lots of basic imports which did import identical stuff with different type, ...

Seems that test does not fit anymore and can be removed.

[simon04]

In 6563/josm:

fix #9487 - validator: remove useless NodesWithSameName test