From: Bill R. <ro...@gm...> - 2011-10-07 05:43:54
|
On 2011-10-07, at 4:08 , Erik Vos wrote: >>> Using git bisect I am able to track the offending commit (at least as > far as 1830 goes) to commit 0c69cf, which introduced the Coalfields variant. > Digging further, the cities on tile #59 were at positions 052 and 352 > before this commit, but at positions 152 and 452 following the commit. >>> > > This puzzles me. As far as I can see, the position values 152 and 452 for > tile #59 exist since the introduction of these codes 6 years ago. This may all be a waste of time, if the problem is simply in the display code, or can be easily fixed in the display code, but git claims that tile #59 changed at commit 0c69cf on 13 Feb 2011. Specifically, the following three consecutive commits (of those commits that affect the file tiles/Tiles.xml ) are relevant for what I describe below: 30628c - 14 Mar 2011 - Extra off-map tiles for 18VA (and others) 0c69cf - 13 Feb 2011 - Initial commit for 1830 Coalfields 98c5dd - 18 Dec 2010 - 1889 fixes: When comparing it is easier for me to take diffs between 30628c and 98c5dd, as the file in 0c69cf doesn't seem to have any linebreaks in it that diff will understand. By visual inspection the code for tile #59 is identical in both 0c69cf and 30628c (up to linebreaks). When I do the following in the tiles/ subfolder of rails, I get the following output: rosgen@thales tiles $ git checkout 30628c5 HEAD is now at 30628c5... Extra off-map tiles for 18VA (and others) rosgen@thales tiles $ git diff 98c5dd -- Tiles.xml | grep -A 7 id\=\"59\" <Tile colour="green" id="59" name="59"> - <Station id="city1" position="052" slots="1" type="City" value="40"/> - <Station id="city2" position="352" slots="1" type="City" value="40"/> + <Station id="city1" position="152" slots="1" type="City" value="40"/> + <Station id="city2" position="452" slots="1" type="City" value="40"/> <Track from="city1" gauge="normal" to="side1"/> <Track from="city2" gauge="normal" to="side3"/> </Tile> I can find no reason for this change. The tile dictionary and the code used to generate Tiles.xml (in utils/ ) shows no changes that look like they could cause this to happen. When I hand-edit the Tiles.xml file for 1830 to revert to 052/352 in place of 152/452, the tokens once more line up with the stops on the tile, but this is not a good solution. Bill |