Who drew this street or: A rant about the "history" feature of OSM
Posted by IOOI on 28 September 2014 in English.After browsing that interesting set of strange borders: https://www.flickr.com/photos/amapple/3416483188/in/set-72157616310862857 I had a closer look at those islands and found that strange street project there.
So I used the “history” feature of OSM to find out who did it: osm.org/history#map=14/65.6658/-168.3774
Answer: I could not find out this and I am, one more time, very annoyed with this feature. It is an un-feature because it is implemented in the wrong way. Instead of showing all the change sets which are inside the chosen boundaries it does it the other way around: it shows all the changesets which have the selected area inside themselves. Not only is this wrong it also makes this “feature” completely unusable: The list of those huge, sometimes world wide change sets is huge and so when you’re trying to figure out what went on in a specific area you usually get a list which is spammed by tons of those huge change sets (most often change sets from wheelmap.org, followed by bots and some users who have the strange habit of “fixing” things all over the world). But such huge change sets are not the real problem here. The real problem is that the data is selected the wrong way: change sets that contain the selected area are shown instead of change sets that have changes inside the selected are. Please somebody fix this! It is annoying to no end. And sorry for the rant, I am just somewhat annoyed and haven’t learned to suffer without complaining yet („Lerne leiden, ohne zu klagen, das ist das Einzige, was ich dich lehren kann.“ Kaiser Friedrich III). I don’t want to offend anybody.
regards,
IOOI
Discussion
Comment from Minh Nguyen on 28 September 2014 at 21:37
It isn’t at all intuitive, but the way to get the history of that particular line is to:
In this case, it was added three months ago and never edited since. The line is tagged as a “proposed highway”, with a reference to a Wikipedia article on the Bering Strait crossing.
In my opinion, this line should be deleted. The
highway=proposed
is meant for proposed highways that are in serious consideration, with a well-defined path and some degree of funding, perhaps awaiting feasibility studies, but not yet under construction. It’s just too early to put this proposal on the map; it isn’t even certain whether it’ll be a bridge or a tunnel!Comment from IOOI on 28 September 2014 at 22:15
Minh Nyuyen, thanks for the hint, this really is a very well hidden feature — you just have to know it ;)
Well, osm.org/way/290450974#map=8/65.859/-169.003 is mainly utopia what one realizes when reading the wikipedia article you linked. I think it should be removed.
But for the history feature, I never the less think it should be fixed. Maybe just adding a subselect would help here:
this pseudo code should be sufficient. Now we need somebody who actually fixes this.
Comment from Zverik on 29 September 2014 at 07:26
Take a look at whodidit: http://zverik.osm.rambler.ru/whodidit/?zoom=14&lat=65.6658&lon=-168.3774&layers=BTT It tracks only node changes, so I had to scroll to a first corner of the highway. It shows that it was modified in June, and you then can open the changeset or filter by it, observing other changes from it.
Comment from Sanderd17 on 29 September 2014 at 07:57
There used to be a service called OWL (OSM Watch List), but it doesn’t seem to work anymore: osm.wiki/Owl But everything you propose has already been implemented.
Also, you can easily check the history of an object in various editors. Like in JOSM, you just have to press CTRL+H.
Comment from Minh Nguyen on 30 September 2014 at 07:09
You’ll probably get more attention from the developers if you open an issue at GitHub for your feature request.
Comment from mcld on 30 September 2014 at 07:33
IOOI, I think many of us want the feature you’ve described, and have wanted it for years. However there’s a technical issue holding it back. Changesets have bounding boxes (bboxes) stored in their basic metadata, so it’s very easy to run a database query to find “all changesets that cover this area”. Unfortunately, if you want to find out if a particular changeset actually affects an area, you can’t just look at the changeset metadata - you’d need to retrieve all the nodes/ways/relations affected in the changeset, and recurse down into the objects they contain or no-longer-contain, and then see if any of those lies in the area. Technically possible but a total pain to code properly, and even harder to make efficient, I guess!
The OWL service was a great idea for a better history tab. As far as I can tell they put a load of effort in but then it didn’t progress further. Not sure why. Probably just the amount of volunteer effort available.