There's been some recent discussion on this, and still nobody has any definitive answers. Everyone is familiar with the problem, if you don't get send/don't send, you either get:1. Vector/no vector end-end
2. What looks like a succesful rebuild (all tiles visited), but some track sections return "adjacent pieces not loaded"
3. As above, but with a single blue pole in the middle of the track, sometimes followed by same error mesage as #2 if you try to select one of the sections on either side.I haven't been able to do a succesful rebuild since PO&N 4, and assumed it was because of the signals. Since everything ran okay, I didn't worry about it. When Hank Sundermeyer came up with a brand new signal set for the PO&N, I deleted all the old signals from PO&N6, and found a whole bunch (86 IIRC) of orphaned signal object entries in the TDB and TIT files, along with a ridiculous number of "Empty Item" entries (several hundred!). I figured with all the signals gone maybe I should try a rebuild, after several failures I gave up. At one point I found a signal that showed up in the activity editor, but not in the RE, so for that I deleted the section of track and saved it, and the RE offered to delete the signal that I couldn't see.After several months screwing around with this problem:http://www.train-sim.com/dcforum/DCForumID3/23325.htmlI finally uploaded what I had as PO&N 8, with the suggestion that they turn down the visibility slider and save, exit, and reload the game occasionally. I came to the conclusion that too many loose consists in a given area could be cured by spreading things out a bit, so I began tearing up and relaying track. When I went to remove one of the yard ladder switches in Anderson, I got that "adjacent track pieces not loaded" crap, and eventually had to go back to an early version of PO&N7 (never released except as beta) that didn't have the problem (slightly different track configuration).So I tore out all the track to relocate Anderson, the coal mine, Texas Tea, and Chippewa Falls outside of the existing route, and started laying new track. After working a half hour or so laying track, then tearing it up and relaying it differently, I'm getting sudden unexplained crashes from the RE, with send/don't send. It happened frequently enough to make trying to lay track useless, so I figured I better bite the bullet and try that unholy TDB rebuild again.
Figuring that wyes and return loops might be confusing the rebuild, I cut the tracks in all the indicated places by deleting a section of track in each place, so there are no wyes, return loops, or continuous paths possible. Using Route Riter I deleted all the signal shapes from the .w files, and attempted the rebuild. Send/don't send.So now I cut half the .w files out, this time got a succesful rebuild. Deleted the new TDB and TIT files, copied more of the .w files in and ran it again, moving .w files in and out until I isolated which .w files were the problem. Turned out to be three of them, and what they had in common were long curved sections like the a1t2000r20d.s section.Went back to the previous backup, WITHOUT cutting out wyes and loops, this time just eliminated the three offending .w files, and the rebuild completed successfully with all the wyes, loops, and circles in place.So I went to relay the track in those sections, this time using two 10d curves where I had previously used a 20d. Same problem, certain configurations of curves are causing the errors. Here's one oddball example, the connecting tunnel between Hawthorne Melody and Armstrong;
Removing those two tunnel sections allows a succesful rebuild. Thinking possibly a connection to something further along might be the problem, I relayed the track with 100 meter straight tunnel sections and connected the ends back together:
THAT allows a successful rebuild. So in that area at least, I've narrowed it down to two 1000r10d tunnel sections will not allow a rebuild, two 100 meter straight sections will allow it.I'm proceeding cautiously at this point, I have an almost superstitious dread of connecting dynamic track directly to switches now (altho it might not be a factor), but I'm rapidly coming to the conclusion that if the route won't survive a TDB rebuild, I might as well trash it, restore a backup that did survive a rebuild, and lay the track differently.
Printer-friendly page | Edit | Reply | Reply With Quote | Top
RE: TDB rebuild failures, MRRextreme, May-20-04, 02:29 AM, (1) RE: TDB rebuild failures, robertreedy, May-20-04, 05:47 AM, (2) RE: TDB rebuild failures, enocell, May-20-04, 03:28 PM, (3) RE: TDB rebuild failures, sniper297, May-20-04, 07:22 PM, (5) RE: TDB rebuild failures, mivone, May-20-04, 07:17 PM, (4) RE: TDB rebuild failures, landnrailroader, May-20-04, 07:29 PM, (6) RE: TDB rebuild failures, OTTODAD, May-20-04, 09:07 PM, (7) RE: TDB rebuild failures, lfriddle, May-20-04, 11:20 PM, (8) RE: TDB rebuild failures, OTTODAD, May-21-04, 08:26 AM, (9) RE: TDB rebuild failures, sniper297, May-21-04, 11:58 AM, (10) RE: TDB rebuild failures, robertreedy, May-21-04, 01:45 PM, (11) RE: TDB rebuild failures, OTTODAD, May-21-04, 05:16 PM, (12) RE: TDB rebuild failures, OTTODAD, May-22-04, 11:52 AM, (13) RE: TDB rebuild failures, mivone, May-23-04, 11:47 AM, (18) RE: TDB rebuild failures, griffinmt, May-22-04, 12:54 PM, (14) RE: TDB rebuild failures, OTTODAD, May-22-04, 07:02 PM, (15) RE: TDB rebuild failures, sniper297, May-22-04, 08:01 PM, (16) DOH!!, sniper297, May-23-04, 11:27 AM, (17) RE: DOH!!, OTTODAD, May-23-04, 12:27 PM, (19) RE: DOH!!, sniper297, May-23-04, 05:26 PM, (20) RE: DOH!!, OTTODAD, May-23-04, 08:53 PM, (21)
Conferences | Topics | Previous Topic | Next Topic
Gently undulating curves look real sexy, but unfortunately note that the TDB line goes in a straight line. Two curves with one flipped to make an S curve sometimes works, sometimes don't. On this quad I got isolated blue poles at some of the places where the second curve was flipped, altho the RE thinks it made a successful rebuild. In every case where I had a bad tile (four of them total), I removed all these sweeping curves, and following approximately the same survey I relayed the track, making sure there was at least one 10 meter straight section between any reverse curve. That fixed it, I now have all four problem tiles working correctly.
There are several return loops and circular paths (potential, you always have to throw a switch curved to continue on), and 14 wyes in that, and it does rebuild with no errors.So those of you having the problem still, when the RE crashes or gives an error message, look at the numbers in the camera window (-12353 14862, the .w file numbers) and write them down, then investigate that tile plus all the ones connected to it for a back-to-back curve with no intervening straight section. I THINK (still don't know anything for sure) the radius is a factor, I've never seen it with the endpnt10d curve sections, and 500r20d don't seem to have a problem either.
Again, that's been cured for version 9 just by ensuring that any reverse curve sections are seperated by at least one 10 meter straight section. No more sweeping curves flowing into each other, the TDB rebuild don't like them.Having Mike check in here just made me think of the other Mike, Route Riter has gotten to the point where it can do just about everything except build the route for you. The TDB rebuild works (when it DOES work!) by taking the position and qdirection of each track section from the .w files and writing all that TrackNode stuff:TrackNode ( 2
So I took out a track section at that trouble spot and YIPPEE, Re-Build successful with no more complaints about missing anything !!!
O.K., lets put that track piece back in again. You can see below what happened trying so SAVE the Route. The funny thing is that the track piece is showing afterwards but the TDB seems to unaware of it ! 
Tried it once more, but not actually connecting the piece, same effect !
There is a moral somewhere, I am not experienced enough to see, but perhaps you and other experts can ? Take Care and good luck with your future PORT OGDEN & NORTHERN efforts, each one a GEM, Errors or not !!! O t t o
I don't know how Trainmaster or Trainz works, but I seriously doubt either one is coded as clumsy as MSTS. Place a signal on the track, the .w file, .tdb, and .tit all get entries with the info on that one signal, and all three files have to load and process correctly for it to work. Add a level crossing and now you add the .rdb and .rit, so now you have five different files that have to be exactly in synch. If the guys at PI Engineering actually tried to screw it up, as a practical joke, they couldn't deliberately get it as bad as MSTS.Finished laying an area last night, saved, exited, deleted .tdb and .tit, ran the rebuild. "All tiles visited", no errors, so I shut down and went to bed. This morning I went to pick up where I left off, noticed some terrain covering a section of A1t1000r20d.s, so I selected it, hit the Y key, and uncovered it. Then I moved to the end of the track (a few sections over), placed a 100 meter straight section, crash send/don't send. *&%$#! So I deleted the .tdb and .tit, ran the rebuild again. This time I opened the route in the activity editor and scrutinized the mainlines for black dots, sure enough found one way over on the other side of the route. 
Here's a closeup:
The upper dot is correct, cuz that's a switch, but the lower one shouldn't be there. And again, that's two in as many days, the track was fine until I did the rebuild, but after a "successful" rebuild with no error messages I got rogue blue poles. Now I remember why I stopped doing rebuilds as a regular routine. Trying to select that track I got the "adjacent pieces not loaded", so I quit, deleted the .tdb and opened the RE without running the rebuild (with no .tdb file all track sections are merely static objects), deleted all that track, exited and ran the rebuild again. This time I got vector-no vector on the next tile over. Going to that tile I had this:
Mindful of the crossing tile boundaries with big track sections theory, I deleted that and replaced it with a trainload of 100, 50, and 10 meter straight sections. "Vector-no vector". Oh well, didn't like the neighborhood anyway, so I deleted all the track back to 77 and rerouted the whole thing away from that corner. That time it worked, and after CAREFUL examination for stray black dots in the activity editor, I backed it up.Well, gotta make supper for the critters, then get back to this *&%$#! grindstone.
This is a much faster way, if you create a path, the path will automatically end where the blue poles (black dots in the AE) are, and if the end of path dot is in the middle of the mainline, you've just found a place where the TDB rebuild KUJUed up. 
Obviously if you have multiple branching mainlines, you need to "take other exit" at each main divide switch to check each branch.So with that one condition, I can agree with Mike that the TDB rebuild is a good "health checker" for the route, PROVIDED you double check with the AE path editor to make sure the rebuild didn't KUJU up the TDB. Nothing worse than having the doctor tell you you're fit as a fiddle, and two months later your bowstrings snap. Common denominators? I dunno yet, sure as hell would be nice to know exactly what to avoid. Back-to-back S curves with no straight track in between might be a problem, altho I have some areas with long sweeping curves that never cause any trouble. Crossing a tile boundary even in the middle is sometimes a problem, I suspect the angle makes a difference. Where I had track crossing a single tile boundary at a shallow angle and got a rogue blue pole, I rerouted the track to cross the boundary closer to a 90 degree angle, and that cured the problem.
Academic now that you are well under way with the PON-9 but puzzling nevertheless.Trains driving on grass with no tracks visible ? Had a few tiles like that in the N.&.J.C. at one time while adding to it and trying to find a cure for it's TDB Re-Build failure. Take Care, O t t o.
Note the track going under the bridge crosses the tile boundary at a very shallow angle, I think that was the problem. higher up it curves back and re-crosses the same boundary, but at something close to a 45 degree angle, which don't seem to give any trouble."a lot of stuff missing South of Leesburg and seems to be the PON-9 you are working on ?"Yeah, what's missing is the coal mine, which I'm moving outside the existing route. My current theory on this "failed to create SMS object" and "error inserting/removing AI service", sometimes combined with disappearing scenery, is that too many loose consists within the player's circle of visibility overloads this clumsy-ass KUJU game engine.So;
So what's left is laying all the yard tracks, placing all the signals again, then testing with 1000 loose consists to see if this cures THAT problem!