Depending on which way is the longest, the new way is inserted into the
relation before the existing one. This case must be explicitly handled
for the relation to remain correct.
The operation is disabled when attempting to join ways which don't belong to identical sets of relations. Restriction relations are excluded, because they are already handled with slightly different logic.
Fixes#8674Fixes#8645Fixes#3825Fixes#1512
The pointer events use a different mechanism for counting clicks that doesn't
work when the tests are run in a real browser like Chrome. This change forces
iD to always use mouse events during testing, even in a real browser.
By default, the `rendererBackground` code will choose an initial imagery based on what
sources are available at the given map location. It looks like this code will fallback
to "custom" _before_ "none", and I noticed that in some cases it was trying to fetch
whatever source happened to be stored in the "custom" template in localStorage.
This commit adds an explicit window.location.hash in a few places to encourage the
background layer to be "none". Some tests do change the hash, so this isn't perfect.
- Before it wasn't actually loading the intro graph because assetPath not set
- The test for Three Rivers City Hall was using the wrong nodeid
- All these test errors were being ignored because of the use of .finally(done)
- To actually make an error happen, we can call `done(err)` with the Error
Add utilFetchJson to get around some quirks of d3.json and use it for coreFileFetcher
Load real general English locale strings at the beginning of code tests
- implements a validation work queue, jobs are run during browser idle callbacks
- when merging base entities, don't run validations 2x on both base and head graphs (this was wasteful)
- keep track of resolved issues in a separate set (it's not a simple compare of base/head anymore)
this happens after validation queue is empty and avoids race conditions and inaccurate resolved counts
