⚡⚡⚡ OSM API v1.0⁴: New data model, validation, OSM Premium and more
Publiée par TrickyFoxy le 1 avril 2024 en English.At the last meeting, the OSMF board gave the go-ahead to implement a new version of the OSM API. The full transcript can be found on wiki. Summary:
🧐️️️️️️Validation
Edits that break the geometry or topology will be rejected; for non-critical errors the API will return an HTTP status 267 Doubtful But Okay
🛣 New data model
To simplify the data model, it was decided to abandon the Way type. Relations are already used for multipolygons and can store lists of points, which makes Way meaningless.
This approach will also allow us to describe nonlinear geometries, which takes mapping to a whole new level 🚀
📃 New data format
The obsolete XML format will be replaced by human-readable YAML. This allows you to conveniently edit the map in any text editor and study changesets using tools such as git diff
. JSON format support will remain.
👮♂️Pre-moderation
New users’ edits will be included in the pre-moderation queue. Edits can be approved by any cartographer who has mapped 42 days in the last year. You can avoid the queue by paying for OSMF membership.
💵 Premium Features
In addition to disabling pre-moderation, cartographers who have paid for an OSMF membership will receive priority when accessing DWG and OSMF. The duration of blockages for premium cartographers will be halved.
💎 New cgimap
The current implementation of cgimap, written in C++, does not meet the security requirements, so it will be rewritten in Rust Ruby, which is already used for most of the site.
Discussion
Commentaire de gileri le 1 avril 2024 à 13 h 44
Wow, at last ! API 0.6 was showing its limits, happy to see it become web-scale.
Commentaire de Kamil Kalata le 1 avril 2024 à 14 h 42
When i got rickrolled, I figured this diary entry was written for April Fool’s Day. However, some ideas are not stupid, especially pre-moderation. It is a good concept, but I think queueing new users’ edits would overwhelm the reviewers. Instead, we should keep the status quo (optional review requests and rate limits), but we could apply the idea for blocked users’ contributions. Despite being unreliable, they would be useful. Firstly, they could relieve other contributors from entering data, notably in areas where few editors are present. Secondly, pre-moderated edits might be a penance for blocked ones. Rate limits might apply, of course (I thought of 100 elements per day).
Commentaire de NorthCrab le 1 avril 2024 à 21 h 49
Commentaire de MxxCon le 2 avril 2024 à 2 h 09
I’m strongly against the switch to relations. OSM instead should’ve switched to using vectors. Everything is a node+vector. Then using vectors you can perform matrix operations, which then opens the door for machine learning! Win, win, win! I’ll take my permanent, lifetime seat on osmf board now and you can call me Grand Poobah.
Commentaire de TrickyFoxy le 2 avril 2024 à 13 h 28
And vectors generalize well to three-dimensional space!