[disc]

 midneguy, on 16 September 2014 - 03:01 PM, said:

I counted 20+ names being used for the top of the hierarchy other than "MAIN" as it should be.

20 nodes are already sounds too much in draw calls for an msts locomotive, not to mention other objects.

[midneguy]

 disc, on 17 September 2014 - 05:28 AM, said:

20 nodes are already sounds too much in draw calls for an msts locomotive, not to mention other objects.

I didn't say I counted 20 different nodes in a single model... I counted 20 or more different names that I observed being used for the top-most node ( or matrix, group, etc. ) among many different models in their respective matrix lists.

[copperpen]

Thing is, this single part, normally referred to as MAIN, can in fact have any name as long as it is not a reference part name such as rod, wiper, wheel etc. It is the single part of the model that defines the world positioning and orientation. So, if the program is losing sight of that essential part it becomes impossible to calculate the current world position and orientation. If OR is requiring that every model carry the part name of MAIN for this, it will be quite the task to edit every S file in use.

[midneguy]

I definitely agree.... It would be best long term if OR can be accepting of other top node names other than "MAIN", as long as the structure of the model hierarchy is correct otherwise. When I started modeling, one of the early things I remembered is that the top most level of the hierarchy should always be named "MAIN" and I was really surprised to see just how many examples of models in just this one route did not follow that convention - including from various payware vendors, and even the original MSTS default stock!

I can follow up on my last test last night, which I will say was with OR version 2477. After starting one of the "problem" activities again I had an AI train freeze, so I thought maybe I wasn't onto anything after all. Then it occurred to me that there was one last thing which might have been causing trouble - FRED rear end devices on most of the consists. So I then went back and deleted those from the consists in the activity - the result was that I ran a 5+ hour activity which included around 30 AI services being run throughout. I never had a single consist freeze after the FRED's were removed, and only regular stock and locomotives used. This was after I had gone in and edited the top node in every engine and car shape file to have the top node named "MAIN". So, it does seem pretty clear this was the cause, along with the "unconventional" FRED cars as well.

I've wondered if this has all been related to the other thread discussing disappearing rolling stock "Invisible coaches after last Exp.-Release" so I am curious to see if some of the fixes that have implemented there have an effect on the AI consist problem or not http://www.elvastower.com/forums/public/style_emoticons/default/book2.gif