In this area https://www.openstreetmap.org/edit#map=17/62.77698/27.50262 some of the roads do not render (latest FF) for no apparent reason (editing in other editors seems just fine).
Reported here: https://forum.openstreetmap.org/viewtopic.php?pid=796045#p796045
simonpoole
All 15 comments
This appears on other places in the world as well. Sometimes the points of the roads and polygons are rendered, sometimes nothing at all. Curious: Some roads from the outside of the actual area are rendered. Changing the zoom level jumps around and does not effectively change the zoom, in the end the background image is shifted.
The iD editor cannot be used currently.
st-anton
on 30 Jul 2020
even more strange - click on the nodes that are showing and things appear?
tastrax
on 30 Jul 2020
Obviously some areas in the world are affected, other not.
st-anton
on 30 Jul 2020
Obviously some areas in the world are affected, other not.
While just idle (not quite) speculation it seems to be triggered by a specific data situation that causes rendering to abort at that point, the console output does report an error, but best leave tracking that down to the devs.
simonpoole
on 30 Jul 2020
@simonpoole Thanks for reporting. I'm seeing console errors like e.uid is undefined and entity n231109719 not found, which are usually pretty serious errors which will exit the rendering code and leave an incomplete map. I'll work on debugging further.
quincylvania
on 30 Jul 2020
Some more examples:
https://www.openstreetmap.org/edit#map=15/51.2277/13.6119
https://www.openstreetmap.org/edit#map=19/51.16365/13.56454
st-anton
on 30 Jul 2020
Looks like this was happening since iD wasn't accounting for the possibility of undefined uid on entities returned by the JSON endpoints of the OSM API. Simple fix for a nasty bug.
quincylvania
on 30 Jul 2020
Ah revenge of the anonymous users :-)
simonpoole
on 30 Jul 2020
Looks like I encountered this editing in Devon a day or so ago. I made a screen grab but hadn't followed it up as I moved to using other editors. I guess the S Devon coastal areas were mapped fairly early on. Glad there's an explanation.
SK53
on 30 Jul 2020
Argh, too bad I missed this, I should’ve known better. 🙄
mmd-osm
on 31 Jul 2020
Are we going to get a new build with this at some point? I held off rolling back yesterday assuming we would get a quick fix but if that's not going to happen I probably will need to roll back to a working version.
tomhughes
on 31 Jul 2020
By the way, both uid and user fields would be absent for anonymous users. Would you need an empty string as fallback for „user“ as well?
Version would never be missing otherwise you couldn’t upload any changes for this object.
mmd-osm
on 31 Jul 2020
I'll release a patch today.
quincylvania
on 31 Jul 2020
@quincylvania : did you read my last post? You didn’t change the user bit before releasing the new version. Just wanted to be sure there’s no further issue here.
mmd-osm
on 31 Jul 2020
did you read my last post? You didn’t change the user bit before releasing the new version.
@mmd-osm Yes, sorry. It should be fine for user to be undefined rather than an empty string since the XML parser worked the same way. For instance, iD properly displays an "Unknown" label in the UI for features last edited by anonymous users.
quincylvania
on 31 Jul 2020
Related issues
tordans
·
3Comments
tordans
·
3Comments
naoliv
·
3Comments
jidanni
·
3Comments
1ec5
·
3Comments
Most helpful comment
@mmd-osm Yes, sorry. It should be fine for
userto be undefined rather than an empty string since the XML parser worked the same way. For instance, iD properly displays an "Unknown" label in the UI for features last edited by anonymous users.