If you have any trouble registering for the forum, please email travmap@teresco.org for help.
0 Members and 1 Guest are viewing this topic.
Might even be open to a discussion about potentially removing that datacheck altogether.
Quote from: michih on October 07, 2019, 10:27:50 amI was not yet on board when LONG_SEGMENT errors were declared being "always a true error" and I don't see any point why they should be.Thinking about this some more.This was really just to aid in the old CHM map-drawing engine, and is not really relevant any more for TM, with the switch to interactive maps.The idea was to avoid long segments that would not have been plotted due to crossing completely over a given viewport / bounding box without having any points within it.For example, the Portland map would have looked more or less like this one (generated with https://github.com/yakra/tmtools/tree/master/canvas),except that the small segment of ME35 crossing over the SW corner was left out.So we can see that even with the datacheck included, we still weren't always guaranteed proper results.Less of a big deal because there's little to no need for the small city/area maps, with the interactive Leaflet maps. A region map will by definition contain every highway segment in that region. Less of a big deal because even if we don't see TM routes in a neighboring region when looking at a region map, we at least have OSM & other tilesets to get our bearings & some context.Even if we did add routes within a bounding box for a given region, I wouldn't cry about missing a route segment from my MA map in a faraway corner of CT or NY.Doing such, or re-adding CHM style maps seems unlikely. Especially with the variation in viewport sizes due to different screen resolutions / window sizes, it'd be too complicated code for too little gain.TLDR; I'm fine with leaving these marked FP as they currently are.Might even be open to a discussion about potentially removing that datacheck altogether.
I was not yet on board when LONG_SEGMENT errors were declared being "always a true error" and I don't see any point why they should be.