[OSM-talk] Bug in iD?
A user of the iD editor has added a new way along an existing way, with the two ways sharing the same nodes. In so doing, he has replaced all the nodes of the existing way with new nodes. The old nodes have been deleted. The new nodes are in exactly the same positions as the old nodes. If an old node had tags, the tags are reproduced on the new node. If an old node was a member of a relation, the new node replaces the old node in the relation, with the same position and role. These things would be difficult to do in JOSM. See, for example https://www.openstreetmap.org/changeset/37338712 - 144 nodes deleted https://www.openstreetmap.org/way/280996572/history versions 1 and 2 https://www.openstreetmap.org/way/155392067/history versions 1 and 2 As a result, a good deal of history has been lost. (I have since reinstated some of the old nodes.) I am not familiar with iD, so I am asking, is this 1. A bug in iD, 2. Inadvertent action by the user, or 3. Deliberate action by the user? (The user is French; his name exists in both English and French. If the user needs to be contacted, it may be necessary to raise this on talk-fr.) ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Bug in iD?
It's slightly unclear who's done what here. Did he create or just extend 'Canal du Midi à Vélo' https://www.openstreetmap.org/way/341531911 ? I'd be more alarmed by the fact there are two separate ways overlaid than loosing some nodes Is there a specific reason the nodes are so dense? Are they all required? Amending the geometry of such ways is very awkward & time consuming. Dave F. On 17/03/2016 22:44, Adrian wrote: A user of the iD editor has added a new way along an existing way, with the two ways sharing the same nodes. In so doing, he has replaced all the nodes of the existing way with new nodes. The old nodes have been deleted. The new nodes are in exactly the same positions as the old nodes. If an old node had tags, the tags are reproduced on the new node. If an old node was a member of a relation, the new node replaces the old node in the relation, with the same position and role. These things would be difficult to do in JOSM. See, for example https://www.openstreetmap.org/changeset/37338712 - 144 nodes deleted https://www.openstreetmap.org/way/280996572/history versions 1 and 2 https://www.openstreetmap.org/way/155392067/history versions 1 and 2 As a result, a good deal of history has been lost. (I have since reinstated some of the old nodes.) I am not familiar with iD, so I am asking, is this 1. A bug in iD, 2. Inadvertent action by the user, or 3. Deliberate action by the user? (The user is French; his name exists in both English and French. If the user needs to be contacted, it may be necessary to raise this on talk-fr.) ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk --- This email has been checked for viruses by Avast antivirus software. https://www.avast.com/antivirus ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Bug in iD?
On 17/03/2016 22:44, Adrian wrote: A user of the iD editor has added a new way along an existing way, with the two ways sharing the same nodes. In so doing, he has replaced all the nodes of the existing way with new nodes. The old nodes have been deleted. The new nodes are in exactly the same positions as the old nodes. If an old node had tags, the tags are reproduced on the new node. If an old node was a member of a relation, the new node replaces the old node in the relation, with the same position and role. These things would be difficult to do in JOSM. See, for example https://www.openstreetmap.org/changeset/37338712 - 144 nodes deleted https://www.openstreetmap.org/way/280996572/history versions 1 and 2 https://www.openstreetmap.org/way/155392067/history versions 1 and 2 It's not a problem that I've heard of with iD before. It might be worth mentioning it to the iD developer over at https://github.com/openstreetmap/iD/issues , but without some idea of how it was created I suspect it's going to be difficult for them to do anything. What I'd suggest is to try talking to the mapper via a changeset discussion comment, politely explaining what the problem is and what you did to fix it, and asking how the data ended up as it did in the first place. I'd add the comment to the discussion in both English and French (if you don't speak French machine-translated between those languages won't be perfect but will I'm sure be good enough; if you're worried that the results might be wrong machine-translate it back to English to double-check). You may not get a reply, but at least you've tried to find out :) Cheers, Andy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Bug in iD?
@ Dave F Almost all of the work on way 341531911 has been done by the user in question. It has taken me a while to look into the details. I agree that what the user has done, also raises issues of quality and good practice. These issues will need to be discussed with the user. But before starting a discussion with the user, it is necessary to know whether 1. to call him out for deliberately misusing iD, 2. to explain to him how he is inadvertently mishandling iD, or 3. to say nothing about the replacement of nodes, if it is a bug in iD. Also, before sending in a bug report, it is necessary to know whether it is a bug. That is why I asked the question about iD. It is recommended to try to preserve history https://wiki.openstreetmap.org/wiki/Good_practice#Keep_the_history To answer your question about the history of way 341531911 in a bit more detail:- The history of the way is complicated so I am simplifying considerably. User Géovélo split an older way, and the part split off became version 1 of way 341531911. Géovélo did this to add the part split off, to two cycle route relations. Version 1, 13 nodes, 280 m, running east 2015-04-29 User J... (I avoid using his full name) split the way at its second node, so the bulk of the original way continues to exist under a new id. After an editing session on 2015-08-21 he arrived at this: Version 5, 36 new nodes and 2 pre-existing nodes, 250 m, running south Most of the way follows an older way. Over that stretch, the older way has had all its nodes replaced. J... altered the lists of members of four cycle route relations and left one of them broken. And so on, with edits on 2016-01-01, 2016-01-16 and 2016-02-21, extending the way over existing ways, reaching Version 12, 775 nodes of which 8 pre-existing, 6.9km Almost all the nodes of the overlaid ways have been replaced. Finally, I fixed up 1.6km of roads that the way follows, producing Version 13, 771 nodes of which 95 pre-existing and 1 new, 6.9km 2016-03-15 The user's mapping raises a large number of issues: o Two highways overlaid. There should be one highway; the mapper should decide what is its principal characteristic or use and tag accordingly, then add any appropriate tags for additional characteristics or uses. o He has mapped a cycleway along roads where there isn't a cycle track or a cycle lane. o He has mapped a route which includes one-way sections (roundabouts and dual carriageways). How is he going to account for the opposite direction of travel? o Traced from misaligned imagery. o Excessive density of points in some parts of the way. o A flourish or hook at the end of the way, which does not reflect anything on the ground, and the end node tagged 'to be continued' (no note or fixme). The hook raises errors in QA tools (intersecting ways on the same level with no node in common). o In two places the way is overlaid with a second short cycleway, one a bridge, one a tunnel. o The way is a member of two cycle route relations. J... has left both relations broken. o What the user is mapping with this way, should actually be mapped with a cycle route relation. Way 341531911 should be deleted except for two short sections where it does not overlay any other way. This may not be easy to deal with. Adrian ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Bug in iD?
@ Andy Townsend Thank you for the good advice. Fortunately I do speak French. @ all I would be interested to have further views on the replacement of nodes in iD. Adrian ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk