developer (#13) - status of the pgrouting (#83) - Message List
Hello, Guys,
Can you outline how far away pgRouting Project can produce a scalable and usable routing solution to achieve the following typical routing goal?
-- Shortest/Quickest path routing for a large data set (e.g. a US state) given a good GIS data
Understand that you can do Shortest routing for a small area. Is it right that the critical item still be to create a large topology in postgis db?
Thanks, Steve
-
Message #319
Hi Steve,
Sure, topology creation takes days sometimes. But it is not the main problem, 'cause you need to do it only once. Shortest path performance for long distances is still an issue.
The performance depends on so many things like network density and size, PostgreSQL settings, the hardware you use etc.
Current pgRouting version works quite well within entire Tokyo road network (which is pretty dense). But it is still impossible to get a result for entire country in reasonable time.
We have plans to implement some solutions for long-distance routing (for example layering or higway hierarchy search), but we don't have enought resources, so it is still in the queue.
If you want to implement it by yourself - please contribute it to pgRouting, 'cause the demand is quite hight.
anton12/14/07 17:09:26 (3 years ago)-
Message #332
Thank you Anton for your info. Can you give estimate of the effort to implement long-distance routing? Can you outline what work is required? Sure I may consider to contribute. But i want some information first.
Thanks, Steve
steve12/23/07 04:39:59 (3 years ago)-
Message #333
Well, there are few ways of how to implement long distance routing. It could be just simple layering (first search for a next edge which leads to the next level; for example, a highway ramp; and then use next, more sparce level until you'll find the closest ramp to the destination point) or something more complicated as HH* algorithm (Highway Hierarchies).
Simple layering may take few weeks to implement and it needs to write SQL wrapper only.
HH* is more effective but more complex and can take few months to implement the algorithm from the scratch.
anton12/24/07 09:36:55 (3 years ago) -
Message #400
Hello, Anton
What is your private email address? About routing via HH, I would like to discuss with you more to have the ball rolling. Yancho and others are interested. I have also contacted some experts in the filed. Some of them have agreed to help.
Thanks, Steve
steve01/28/08 03:25:34 (3 years ago) -
Message #347
Steve, if you want I am interested in helping out on using layers for highways.
What about joining IRC on freenode and contact me, many channels : #qgis #openlayers #mapserver #php #postgresql #postgis #grass .. or else contact me on gmail mpulis [a t] gmail [ d o t ] com .
BTW guys why pgrouting does not have a Freenode IRC channel?
yancho01/07/08 19:37:37 (3 years ago)-
Message #351
Let's see if some pgRouting guys get together there: irc://irc.freenode.net/pgrouting
Nice to see an active community growing!
daniel01/10/08 18:19:04 (3 years ago)
-
-
Message #348
Hi, Yancho
Sure. I will contact you by email. The Trac used by pgRouting appears to be pretty good for tracking the project.
Anyway, I am in the process of contacting more people for contributing the project --Multiple point routing using highway hierarchies. This should be very visible and interesting project. If completed, it will be the solid step forward for breaking through the monopoly of commercial routing providers like ESRI/MapQuest/MapInfo/MicroSoft, enabling services not only to individual users but for thousands of businesses across the world!
Hoping more people will be interested.
Steve at soniyatech dot com
steve01/08/08 08:24:27 (3 years ago)
-
-