Kismet Custom Sequences broken with October...no seriously, its borked

[indefiance]

Well I have a custom AI bot sequence using external variables that works perfectly in the September version. I have about 8 of them which are identical with various external path nodes for destinations spawn points and player etc. (See pic) The node works great in the september version, and in fact works great in the October version UNTIL, I mess with their container level's Kismet Sequence by trying to add a new subsequence. I am not even touching the AI nodes custom sequence.

Then some very strange occurences will immediately break those AI sequence nodes, there is NO repair I just have to revert to a previous save prior to adding that subsequence. I have done some testing with it and it seems that the external variables to my Custom AI sequence get 're-arranged' on the external side and then out of place connecting to the wrong thing and it may be they are not working at all. I have tested and found the September version DOES NOT do this, but I can reproduce the error 100% of the time on my map in the October. Adding normal kismet to that master level sequence does not produce the error, only adding another custom sequence to that level at any point produces the bug.

Further strangeness is that when trying to copy paste the custom sequences, all the external variables become re-arranged...including sequences other than this AI sequence which utilize external variables.

I hope this can get resolved. I realize not many people use such an extensive AI built almost entirely in Kismet, but in my case it was necessary. Moving forward we will implement this prototype into Unreal script but this still represents a bug in UDK October since no problem was encountered in the September version and because there is no problem with the sequences at all until I try to add an new sub sequence to that particular level, further clues are that the problem also occurs in some smaller sequences with external variables. I didn't test these extensively though, only to confirm they were also experiencing the strange 're-arrangment' of external node connectors.

Hope I have given enough information, if not just let me know.

Indefiance


External View of my custom AI subsequence:



Internal View of custom AI subsequence: (Yes its a messy one)

[acexito]

I have the same problem any solution?

[Avengus]

I came here to report similar Kismet issues with subsequences and external variables.
Instead I will just chime in here to rally us together.

My Problem:
I have subsequences that have four external variables.
I have multiple copies of these subsequences.
At completely random times, the variables I am passing into the subsequence will come unhooked from the subsequence.
Worse, I don't know this until I double click on the subsequence to enter it and then return to the parent to find them unhooked. Or even worse, it crashes at runtime because the subsequence is expecting to use a variable that is null.

This is a significant source of grief for me since I have about 30 of these each with 4 variables and before every single build, I have to double click and enter every subsequence and hook up any variables that come unhooked. It really slows down development. It wouldn't be so bad but it causes a runtime crash which kills the playtest and requires doing the process over again from the start to make a new build.

I can't provide screenshots because I'm at work, but if needed, I can do that.

[acexito]

thats exactly my problem, i hope anyone can help us, because each time i change the kismet i broke any link without knowing that in some place that i dont know until the game crash

[acexito]

the new version fix that?

[rafaelbudnik]

Same problem here :/

[Spoondog]

A minor version of this happened to me as well - I had a sequence with only one external variable. Whenever I modified something in it, the variable hooked up to the external input would disconnect. I had another sequence almost exactly the same where the problem didn't occur. I think the difference was, in the one with the problem, I had duplicated and/or renamed the external variable used in the sequence. So that might be a clue as to the cause.