Tag not added
A new node with, containing the tag is added, whereas attaching a tag to an existing node is not added!
#1 Updated by Anonymous over 8 years ago
- File update-test.zip added
Hello, I think this problem does not happen when you add a tag to a node. The problem happens when you add a tag to a way. I think it is because in the temp_node table, the app. does not find the nodes of that way, so the app believes the way is inconsistent. That is in the method:
protected boolean addNodeData(Way way, EntityConsistencyService consistency)
of class WayConstructor.
To test this, I have created a very small .osm file for the initial import and a very small osc file for the changes. I have attached the initial import file to this message, and the osc.gz file is here:
so you have to set https://correo.prodevelop.es/descarga/gvsig as the base folder and do a first daily update using 20110203-20110204.
You will see that the poi for the university "Uni A." is updated but the building "Build. C" is not updated because it is a way.
The added tags are "atag=..." and "btag=..."
#3 Updated by Anonymous over 8 years ago
I wonder if it is a good idea to keep a huge table with [NODE_ID, LON, LAT]. Currently the OSM DB has ~1 billion nodes (1,000,000,000)
If we only need that table to know the coordinates of an ID, we can have:
Node_ID : long (8 bytes)
LON: float is enough (4 bytes)
LAT: float is enough (4 bytes)
16 x 1 billion is about 16 GB for the whole world, which is not a lot. The current DB with OSM-in-a-box would be about 45 GB, while if you use osm2pgsql, the DB can be perhaps 100-150 GB.
#4 Updated by Anonymous over 8 years ago
This was one idea we had, too. This would solve this issue. You can test that by enableing the EntityConsistencyService(run osm2gis without --no-consistency), which will download the missing references and set the tags correctly.
But the EntityConsistencyService isn't really am option, because it is needed too often and you get blocked...
I'm curretly checking, whether keeping all nodes as you proposed would be enought; this would be relatively easy to implement; if we need to access other data(such as raw-way data), the space needed would grow.