Zugg Software :: View topic

SubAdmin Joined: 18 Nov 2001 Posts: 5182

I sent 2 feedbacks on the bugs along the way to this. Those are most likely causing some bad memory corruption. In 2 of 3 tries while I was refining the procedure I had the chance to send a crash dump, the third time went to Access Violation.

Steps
The usual launch, close, open
Paste Script in command line---
#ONINPUT "RiftCap" {^ir$} {} "" {notrig}
#COND {^Glancing into the Rift, you see:$} {Rift="";RiftTemp=""} {case|notrig}
#COND {} {#IF (%match(%line,"%dh")) {#STATE RiftCap 4;#SET RiftCap 4 1} {RiftTemp=%concat(@RiftTemp,"|",%subchar(%insert("|",%insert("|",%line,56),28),"][","="))}} {looplines|param=30|prompt|notrig}
#COND {*} {#SHOW Failed to match prompt during rift capture, adjust param higher or correct other error} {notrig}
#COND {} {#FORALL @RiftTemp {#ADDKEY Rift {%trim(%word("%i",2,"="))} {%trim(%word("%i",1,"="))}}} {manual|notrig}
------
Create new Package
Drag triggers starting with the RiftCap trigger and working down through the child states to the new package
Click on the new package tab
Drag the Glancing trigger into the RiftCap trigger
Drag the Glancing trigger from the place where it was incorrectly put into the RiftCap trigger

Posted: Mon Dec 04, 2006 8:41 pm

Unbelieveable. Of course, the main problem is that when you drag the RiftCap trigger to the new package, it doesn't move the substates in the first place. And this is the same bug that I've already spent something like 20 hours on. It *was* working...don't know why it's not working again. Very frustrated.

Anyway, once it fails to move the substates correctly, then they are going to be screwed up, so moving them to the package and then trying to move them back into the main trigger isn't going to work.

Posted: Mon Dec 04, 2006 8:54 pm

Definitely data corruption going on with trigger states. In fact, it's not even the package editor drag/drop that's causing it. It seems that trigger states are generally screwed up in 1.18 again. Sorry about that...I'll try to fix it soon. Assuming these damn trigger states don't kill me first.

Posted: Mon Dec 04, 2006 9:21 pm

Turns out the problem was that the #COND statement wasn't creating trigger states correctly. It wasn't properly setting the Parent field that the database needed. So triggers created within the package editor would move to another package just fine. But triggers created from the command line containing conditions added with #cond wouldn't move to another package.

Thanks for finding this one Vijilante...it was really obscure but very important to fix.

I just hope I haven't caused any side-effects in multistate triggers with these fixes (damn...now why did I have to say that :)

Magician Joined: 13 Oct 2000 Posts: 463

Natural pessimism.

Posted: Mon Dec 04, 2006 11:29 pm

Yeah, I'm a pessimistic perfectionist. Nasty combination ;)

SubAdmin Joined: 18 Nov 2001 Posts: 5182

Hopefully with all of that fixed I will finally be able to redo my full import and start over with a whole new set of package divisions. I actually found a few of my packages from 1.17 to be corrupted beyond repair, and others had trigger states lost. Just glad it is gone.

Hopefully you were able to capture the untrapped exception before fixing the corruption issues. It would be nice to never have an Access Violation message come up in the public version.