[Tagging] Object rankings & vector tiles (Re: Road classification)

[Daniel Koć]

W dniu 03.09.2015 10:06, Michał Brzozowski napisał(a):
> This (importance ranking) is something I thought recently. How about
> devising a data-driven importance rank? One could priotitize POI that
> have a Wikipedia page or even grab publicly available pageview stats
> from them. There are numerous ways to implement something good enough,
> you just have to be open-minded.

Of course, but it takes a lot of testing to make good ranking scheme for 
given class of objects. I think we're at the beginning of creating such 
complex rules. It also includes area, density and a place context 
(city/rural/mountains/*) awareness.

Is there anybody willing to try?

>  I, like others, think we need to establish a second style that is
> free of all the conflicting requirements (mapper feedback, instant
> updates, a dedicated group of complaining people) that would serve as
> a map for general users, show the good practices of cartography
> (generalization, which may require preprocessing) and respect regional
> conventions. Not being real-time makes it much easier to solve
> problems.

It seems that currently we have no technical infrastructure to run the 
second style yet.

>  As per distribution, vector tiles seem to be the right solution.
> Everybody could set up their own server and render that style in no
> time, with no lengthy and ridiculously RAM-hungry DB imports. This
> facilitates rendering on a cluster of commodity hardware which is much
> cheaper than dedicated servers. Simply grab a weekly (example period)
> vector tile package and copy it to all your machines.

I don't know what are practical problems and real advantages, but I also 
like to have vector OSM tiles. Currently cquest (of French style) is 
planning to test it and maybe marimil (osmapa.pl) can try it a some 
point, but at the very moment we have nothing like this. There was a 
demo here:

http://openstreetmap.us/~migurski/vector-datasource/

but it's down. However Michal Migurski is optimistic about this route:

"I give vector tiles and vector rendering another 1-2 years before it 
tips from weird research and supervised, commercial deployment into wide 
use and hacking."

[ http://mike.teczno.com/notes/sotmus-2015-new-york.html ]

So - who's gonna start it? =}

-- 
"The train is always on time / The trick is to be ready to put your bags 
down" [A. Cohen]