On Sun, 2008-06-15 at 16:32 -0300, Eduardo Habkost wrote: > On Sun, Jun 15, 2008 at 08:53:53PM +0200, Frederik Ramm wrote: > > Hi, > > > >> Do the "ways making up the route" (as defined on Relations/Routes > >> proposal) must be actual highway-tagged ways > > > > Yes. > > > >> or they may be distinct > >> (maybe untagged) ways drawn over the highway ways? > > > > To re-use a popular exclamation of the last few weeks: "Ye gawds no!" ;-) > > Okay. :) > > > > >> I am asking that because on some streets there may be many different bus > >> routes going through it with different entry/exit points. If the ways > >> making up the bus route must be the actual highway ways, those streets > >> will need to be splitted on many tiny little ways, so that each section > >> will belong to a different set of route relations. Did I understand it > >> correctly? > > > > Yes. This is somewhat of a shortcoming but that's the way it is currently > > done. > > This doesn't seem to be a problem only when making bus relations: > streets with lots of tunnels or bridges need to be splitted also. So, > it is a more general issue. I will try to live with it, by now. :) > > > > [...] > > > > Or you simply ignore the fact that the bus route does not use the full > > length of the way; it would still be possible algorithmically to find out > > which part of the way the route uses (except from circular road layouts). > > Won't look nice on today's renderers but who says they cannot be improved? > > That is an interesting idea: maybe I can do this to enter data about > the bus routes more easily and later algorithmically split the ways on > the right places. > > > > > A third option would be inventing a special kind of relation that's called > > a "way section"; it would contain the start node, end node, and way and > > model that part of the way that lies between the two nodes. Then make your > > bus route relation use these way sections as members wherever the bus route > > doesn't use the full way. But this would also be something new and not yet > > supported by renderers. > > I have seen two proposals that would handle the current cases of massive > splitting: "tag section" relations (similar to what you mentioned > above)[1] and "collected ways"[2]. Both seem to sove the same problem > using different approaches. > > At first I thought the collected-ways approach to be better and more > intuitive, but the way-section approach now seems to be better, the more > I think about it. > > (The collected-ways proposal may be useful on other cases, but I > was thinking about the cases of splitting due to bridges/tunnels and > bus routes, only) > > [1] http://wiki.openstreetmap.org/index.php/Relations/Proposed/Segmented_Tag > [2] http://wiki.openstreetmap.org/index.php/Relations/Proposed/Collected_Ways >
The issue here is that both these proposals take us back towards the old segment model. Even towards removing ways, everything just being a collection of nodes. (Let's _not_ talk about that again...) -- Regards, Thomas Wood (Edgemaster) _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk
