[OSM-talk] Student project ideas?

Nic Roets nroets at gmail.com
Tue Jul 31 08:43:06 BST 2007


> routing might be one? Any ideas I can pass on? Can't guarantee anything
> good will come out of them, of course, but it might be useful.

Improving gosmore is an obvious candidate because if they introduce
bugs few people will be affected. Also it's only 1200 lines and link
with very few libraries, so the student will not drown in code.

Here are some things :
1. Numerical analysis : Write a function that calculates or
approximates the actual distance between two points on the Mercator
plane.
2. Routing : (a) Change the current routing code to detect when the
"from" point is on an island and no route to "to" exist.
(b) Improve route planning (get directions) output.
(c) The realtime routing currently warns you too late of an upcoming
turn. Give instructions like "In 100m turn left in Edna Street".
(d) Implement better algorithm for realtime routing.
3. GTK+ / Pango Layout : Improve rendering using "painters algorithm".
I have quite a few ideas here.
4. Data structures : Introduce multiple "Levels of Details"
(mipmapping). Currently the application always renders everything, but
if we have multiple arrays / hashtables we can stop rendering streets
at a certain zoom level.




More information about the talk mailing list