[OSM-dev] GSoC: OpenStreetMap carto

[Paul Norman]

On 3/20/2015 1:17 PM, Sophie Haynes wrote:
> I was looking at the GSoC project ideas, and I am interested in the
> proposal for fixing issues with the OpenStreetMap cartography. I rely
> quite heavily on applications that use OSM, so I would love to support
> it in return, as well as gain understanding as to how it works.
OpenStreetMap Carto is a very active project, and about 1.5 of the 
maintainers were considering mentoring, myself included, and we could 
work with up to two students if there are proposals in different enough 
areas.

One difference between osm-carto and other projects it that also 
involves some non-code aspects, like icon design. I'd want a project 
that consists primarily of code.
> I noticed the required skills included PostgreSQL and CartoCSS. While
> I have not studied these languages specifically; I have learned mySQL
> and CSS which I believe are quite similar. I am more than willing to
> learn any skills I don't have, would this be okay?
MySQL and CSS are similar enough that you should be able to pick up any 
needed differences fairly quickly. CartoCSS is actually derived from less.

For the purposes of this project, we're constrained in that we need to 
use the standard Mapnik + PostGIS rendering stack, and released versions 
of Mapnik. This means that some of the advanced labeling methods being 
proposed in papers aren't useful for us, or would needed to be 
implemented in PostgreSQL/PostGIS
> Any feedback on what work I could do to help for the project or
> general advice would be greatly appreciated!
The description on the wiki is a bit vague and the student will need to 
flesh out what they want to do. It's less of a ready-made project than 
some of the others.

If looking at issues I'd look at an area with a number of issues of a 
broad over-arching theme. Some examples with issue numbers are are

- Roads styling (there is already someone proposing this) (#110, #831, 
#102, #547, 1326)

- hstore conversion (I may pick this up, as it involves a *lot* of 
benchmarking)

- Route shields from relations and pictorial route shields (#596, #508)

- Density dependent rendering. A gas station in the middle of the 
Australian outback is interesting at low zooms. One in the middle of LA 
isn't. There's no obvious solution, and getting transitions between 
areas is going to be tricky. For this one, I'd want to see some kind of 
plan of attack to have confidence it would get done.

- Label placement. There's always label placement work that can be done, 
but this would need scoping out

- Conversion from land shapefiles to ocean shapefiles. Probably needs 
more work in it to fill it out to a full summer's worth of work.

To get started I would suggest loading some data 
(https://switch2osm.org/loading-osm-data/) and getting started using 
Kosmtik (https://github.com/kosmtik/kosmtik). It is easiest to work on 
Linux, FreeBSD, or OS X.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20150320/2e15d59c/attachment-0001.html>