Misaligned Track Joints in TDB after making corrections

[joe_star]

Hello,

Using TSRE 7.002, I have always found that realigning tracks already laid (usually to correct mistakes in elevation / rotation) results in bad track joints being written to the TDB, even if visually the shapes themselves align perfectly. What this manifests as is the jitter / jump of the train when crossing the track joint in the simulator.

I have tried removing the entire track vector, saving, and re-selecting track piece by track, but the result is the same, the vector doesnt seem to align with the track shapes.

Another issue is the "flipping" of the track vectors on the last segment of track before the end of the node, whenever track in between is not changed in the exact order it was laid.

Does anyone have best practices for prevention and remediation of such issues? Does the new test versions 8+ also address this somehow?

Thanks

[eric]

I've seen none of that behavior continue in the new testing versions 8.005, but will caution you to have a backup before trying it out.

My guess on this... Goku had a floating point precision value of 8 decimal places being applied to math calcs, which allowed for some ending position values that weren't matched by the track placement position values allowed for starting position on the next track section.

I reduced this to a precision of 7 decimals, which may or may not have caused those ending positions to round to what the starting positions could be.

This was also being manifested in the writing of TrItem data in the TDB having decimal values like 8.0000001 and 7.9999999 instead of 8.0

[landnrailroader]

I am doing the Anchorage to Fairfanks track for those who requested it and I am using an older version, V0.8.002 and I have only noticed this if I failed to hit the "Z' key once or twice after changing the gradient on a piece. Now what I have noticed is that if you are redoing some track you just laid, a selected piece will not latch on to an existing piece if they are originally further than a meter or so up or down. They just flip off in space somewhere, but you can look at the object pane and notice what the piece is, then just hit delete and select a new piece and place as "new". I am laying a mile or two of track, then visually inspecting to see if there are any of those bumps, or any blue poles showing. Then to be sure I use Track Viewer and start a path at the end of my "work" and let it build as far as it can and if it stops where it should not, well, fix the !@#$%^&*() thing. One thing I have noticed is that there must be a maximum number of nodes that viewer can handle because as I got past about 275 miles from the start (just south of the wye to the Anchorage airport), then the path would stop further and further to the north, but if I then started a new path just above that point, it would build to the end.

Now something else I have noticed with this version I am using that is very good, is if a piece falls to negative infinity, and you move the camera directly over it, until you see the yellow line, most of the time I can just select it and delete. I have laid 330 miles so far and probably have a dozen or so that I will have to edit out of the world files, but most have been easily removed as I mentioned. This route is so long, and I have about 20 miles to go to reach Fairbanks which is a really complex yard, although the town only has a population of 40K or so, and it is the 2nd largest city in Alaska. I will also include the branches. I would hope that more than one person wanted this route - I have no serious interest in it myself, and that some method can be worked out so that more than one individual works on it. It should be possible to do that as long as somebody(s) keeps the original route and adds the updates as they come in.

Jerry Sullivan, P.E. (retired)