Reading this, it seems to me everyone is pretty close to talking about the same thing. I've knocked a couple of diagrams up in the hope that it might be useful for the discussion and hopefully tease out if we are or if we are not.
<br><br>I did some work on something similar a while back; I did the conversion as a few steps<br><br>1) convert all segments to ways (containing exactly 2 nodes).<br>2) repeatedly:<br> find pairs of segments (A,B), such that last-node(A) == first-node(B) AND tags(A) == tags(B). Add all nodes in B to A, and remove B.
<br><br>I think it is possible to even create an API translation layer such that earlier clients could still be fooled into thinking segments actually existed, by storing (n/s/w mapping) information in tags - this could be done as a proxy.
<br><br>Some questions:<br>1) Areas could just be a specific type, and/or they could have a constraint such that first-node(area)==last-node(area). This seems reasonable to me...<br><br>2) Areas with 'holes' could be stored as an area with 2 child areas, and appropriate tags on those areas to control which was the bit being added and which was the bit being subtracted (the hole) - (I hope that makes sense)
<br><br><br><div><span class="gmail_quote">On 27/07/07, <b class="gmail_sendername">Frederik Ramm</b> <<a href="mailto:frederik@remote.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
frederik@remote.org</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Hi,<br><br>> Over in the current data model thread Frederik declared consensus in
<br>> dropping segments. Is that really the case? I've thought it was<br>> before and been wrong.<br><br>I've been criticized before for declaring consensus where there were<br>still little Gallic villages of resistance but I maintain that a
<br>large proportion of those who have thought about the matter agree<br>that segments should go, and that it is possible.<br><br>> Frederik also noted that it's mostly the case that there is no top<br>> down design in OSM, just whoever does the work. I know some of you
<br>> want that leadership, so I hereby order you all to drop segments.<br>> Please have this on my desk by Monday. :-P<br><br>Only if you let us polish your boots as well ;-)<br><br>> Dropping segments should be approximately trivial if we make a break
<br>> with the past. Here's Steves n-point plan to success:<br><br>Good plan. Reducing triviality, here are some issues:<br><br>1. Branched and disconnected ways can be found and, mostly, remedied<br>even today. Same with tagged segments and unwayed segments. So
<br>instead of breaking them up and then putting them on a Wiki page, we<br>could do that today and at least reduce their number. (OTOH people<br>might not be inclined do something about it until we thoroughly break<br>it.)
<br><br>2. Areas with holes. Don't have a good idea for them yet. Anyone else?<br><br>3. Dropping way and segment history is acceptable to me, but maye we<br>should keep these things on file somewhere in case we need to do
<br>check something.<br><br>4. We would install all this on a testing platform first, to allow<br>JOSM etc. to catch up... wouldn't we?<br><br>> I'd bet money that we could get all this done in a weekend with a few
<br>> key people in a room, much like the dev day in Oxford.<br><br>I'm busy implementing relationships at the moment but after that, I'd<br>even fly over to help.<br><br>Bye<br>Frederik<br><br>--<br>Frederik Ramm ## eMail
<a href="mailto:frederik@remote.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">frederik@remote.org</a> ## N49°00.09' E008°23.33'<br><br><br><br>_______________________________________________
<br>dev mailing list<br><a href="mailto:dev@openstreetmap.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
dev@openstreetmap.org</a><br><a href="http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/dev" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/dev
</a><br></blockquote></div><br>