iD Support (and other stuff) #12
Conversation
Allows osmosis to be used without allowIncorrectSchemaVersion=yes
(It's an LTS release, no other reason.)
Existing version was outputting correct info, but not at the URL that iD expected.
Also handles changesets containing individual ways by turning them into single-element arrays (change counts are similarly affected).
Necessary due to xml2json conversion.
This removes nodes from output to match OSM.
Updates pinned Node version to 5.3 (5+ required) to facilitate use of Sets and the spread operator (for converting them to arrays). The previous queryBbox implementation failed to load standalone nodes, so I rewrote the DB queries to (largely) match what cgimap (the C++ implementation of the map endpoint) does (preferring to load node/way/relation metadata in a batch). This also simplifies rendering to XML by moving nd / member ordering (defined by sequence_id) into SQL.
(Due to the use of the spread operator.)
Errors within Bluebird promises were proving difficult to debug and since recent Node includes builtin Promises, it wasn't providing a whole lot of value.
|
Hey @mojodna, first off thanks for all the work here! I've done some work off your branch. My goals were to preserve compatibility with original routes in addition to iD/OSM routes ( The code is on a new branch. Could you review and fast-forward into your branch if it still accomplishes your goals? My other request is to please provide documentation and simple test coverage for the additional endpoints you've added, to cover possible future regressions. I'm seeing these:
We use apidocjs. Happy to help as much as I can with both documentation and test coverage. |
|
Sweet. I think I'll have time to incorporate your changes later in the week and then we'll be on the same path once again. |
|
@mojodna great. I'll hold off on merging these changes until you do. |
Avoid returning [undefined].
|
@dereklieu Fast-forwarded, etc, but I haven't written docs / tests just yet. That's next. |
|
(And I'm seeing a bunch of test failures locally, so I want to sort those out before this gets merged.) |
|
@mojodna Cool. Let me know if you have any questions regarding any of the tests. |
This allows tests to be run individually, among other things.
|
Existing tests pass now (locally). API docs are next. |
|
@dereklieu this is ready to merge from my POV. I haven't added new tests, but I'm willing to deal with breakage and introduce them when that happens. I'll also be adding some additional functionality (user info, history info in the db to be able to walk changesets) that will probably touch each of the endpoints I added, so I'll add tests at that point. (I'd like to get us synced up sooner rather than later.) |
|
Thanks @mojodna. The new feature list is minimal at this point, so I'm ok to sync up now and write tests for those later. Can't wait to see the new features/tests you write for that additional functionality. I've spotted some minor fixes, which I'll open a PR for after I merge this. I'll also be doing some work to make sure |
iD Support (and other stuff)
This changes endpoints (and implements new ones) necessary to support the iD editor.
Of the changes, the
mapendpoint is probably the most significant, as it changes up the queries that fetch features (matching cgimap, the OSM C++ implementation of themapendpoint). Previously, nodes that were not part of ways were not being returned.Fixes #11
/cc @olafveerman @dereklieu @danielfdsilva @kamicut