Talk:Wikimedia Maps/2021 modernization plan

To my knowledge, WMF still uses vector tiles structure as defined by MapBox -- which geo feature appears at which zoom level, feature naming, etc. (this is different from the vector tile structure itself, which is open source). MapBox has asserted its copyright to that structure, and sent a number of Cease And Desist letters to small map providers who they tried to use the open source code created by Wikimedia, forcing them them to establish the alternative openmaptiles.org -- a popular fully open source alternative to MapBox specification. Mapbox chose not to bring legal charges against WMF. While vector tiles were rendered server side, this issue only affected those who wanted to reuse the source code. Now the tiles will be brought to the browser, it will further advertise MapBox's proprietary vector tiles to anyone wanting to participate in maps development or styling. I believe this is a violation of the open source values declared by the foundation. Is there a plan to migrate away from the proprietary maps data format?

I assume you're referring to OSM2VectorTiles – are you sure OSM Bright is using that schema?

@Mxn from my (possibly flawed) recollection of the WMF maps project, the vector tiles data schema was targeting OSM Bright style as published by Mapbox, with some simplifications (removal of the features that were not important to Wikipedia maps). Thus despite all the SQL optimizations recently done by WMF, the resulting data schema appears to be similar to the original one. Take a look at all the .mss files in [https://github.com/kartotherian/osm-bright.tm2source osm-bright.tm2source . I think the OSM2VectorTiles went further along that path adding the features that Wikipedia chose not to implement at that time.

P.S. Big kudos for moving to imposm3! I hope at some point OMT and WMF work could be more in line, as there are numerous people contributing and optimizing OMT SQL.

ping @MSantos (WMF) and @SKim (WMF), did u guys see this ?

@TheDJ somehow I get the feeling this issue is being ignored on purpose. Any acknowledgment by WMF staff would prevent them from claiming ignorance later on. Of course the head in the sand approach keeps the costs building up until the proverbial ... hits the fan, rather than dealing with the problem early on and just solve it. A move to OpenMapTiles would be good - its initial goal was specifically to solve this exact issue. Disclaimer: I did participate a lot in OMT work, but I'm OK if WMF decides to move to another project, I'm just not aware of any other ones with similar goals.