<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; ">Hi Stephen<div><br><div><div>On Feb 19, 2010, at 11:08 AM, Stephen Woodbridge wrote:</div><br><blockquote type="cite"><div>pgRouting does not care about this and works fine with either setup.<br></div></blockquote><div><br></div>Good to know thats. At least i can skip this then!</div><div><br><blockquote type="cite"><div>Maybe you should describe what you are doing and what the results and issues are.<br><br>Which of the routing methods are you using?<br>what are the results that you think are wrong?<br><br>pgRouting has some peculiar behaviors that can be explained an worked around if we know more about what you are actually doing.<font class="Apple-style-span" color="#000000"><font class="Apple-style-span" color="#144FAE"><br></font></font></div></blockquote><br></div><div>Ok, the short overview is I am trying to match up sections of road in our navteq based data with sections of road reported to us from another system using Teleatlas data. They can provide a TMC code, start lat/lon, end lat/lon, distance, and road direction (probably general direction not actual segment direction). The problem is the TMC codes are not a reliable way to match to ours, so I am attempting to take the start end points they provide, located the road segments (links in navteq vocab) and "walk" the road from the start to the end to get all links in between. It's usually a small number of links (like a handful and a half mile), and when larger its in open road areas (still 35 links or less, up to 10 miles).</div><div><br></div><div>I will need to run some tests again to give you exact info since I worked on this last week and moved on, but I ran both the dijkstra and A* algorithms on all the data we could be sent. There were differences between them and when I started to spot check, I found really crazy stuff like the algorithm returning seemingly random links over a wide area. I tried re-running assign_vertex_id using a tolerance of tolerance of 0.00001 (I don't recall exaclty what I used the first time, but was following one of the tutorials) and it seemed better by still got whacky results. My understanding is that this is about a 1 to 1.5m tolerance so maybe I need to increase it? Again I don't recall what I used before but I think it was bigger the first time.</div><div><br></div><div>One problem I know I have which might be a cause of problems (but not the random links issue, nor a pgrouting or postgis issue) is that it seems I need to improve my algorithm to detect the nearest road. I have points given to me that fall smack between the two roads so I have to determine the road direction and the link direction itself isn't necessary going to be the same as the road direction. I can't find a reliable way to find the road direction in the navteq data.</div><div><br></div><div>I can dig up some specific examples over the weekend if it will help my changes of you helping me :)</div><div><br></div><div>TIA,</div><div>charles</div></div></body></html>