Error opening sequence after 2016.0x upgrade

[CJHacker]

I've got a sequence I've been working on it 4.3.02 and saw there was a new update the last couple of days.  When I update to 2016.02 (tried on 2016.01 as well), every time I try to open this one particular sequence, it crashes right away.  I've uninstalled the new versions, reinstalled 4.3.02 and can still work and edit just fine in 4.3.02 so I'm guessing I've got something in my sequence that doesn't play nice with the new version?  Debug zip attached.

Thanks for looking!
CJ

[Gilrock]

I get this to crash also.  Trying to dig into it.  It doesn't immediately show me where its crashing so could take some time.  Saw one or two others report the same thing.  I told em they should call the new version beta...lol.  There was a ton of restructuring but it was working fine for all the sequence the developers and Sean tested with so we thought it was ok.

[CJHacker]

No problem, I can chug along on 4.3.02 for a while.  I tested out 4 others that I was working on and they're all fine, so far it's just this one I have issues with (the main one I'm working on of course).  haha Thanks for taking a look!

[thedementedelf]

Getting the same error message. Here's what is interesting. When I installed 2016.2 I picked a sequence at random and it loaded.  It is the only one out of 5 that will load. Everything else crashes. I did get one crash report and I will attach to this.

[Dazzled]

Hi Gil and probably Dan
Had a crash on the Mac build today using 2016.2, not sure if it is related to these issues.
Here is some background to what I was doing.
1/ Opened an old sequence in 2016.2 and played around with snow flake effect to checkout improvements and also used the new Tendril effect did not save any changes and closed seq with no problems.
2/ opened a couple more seq and just played to make sure all was ok and had no problems until about the fourth seq when just a few seconds into playing it popped up the crash window but kept playing the entire seq without any problem to the end at which time the program shut down.
3/ Reopened program and and started working through playing about 12 seq by opening and saving the seq then playing it and had no more problems at all.
What ever the problem was it seems like it is not repeating now, I have attached the crash report as it may help you with this problem please let me know if you need anything else.
By the way I have a problem with the increased speed at which we can now save as it has reduced the time I can spend to rehydrate with some cool liquid refreshments.
Stuart.

[Gilrock]

I'm not the only developer so don't let me discourage you from posting debug reports but I personally can't look at all of them.  Since I have a repeatable crash from cjhacker's files I figure if I or another developer can get that fixed it may fix it for all of you and we'll need you to retest your stuff when we think we've fixed that one.

I spent 3 hours last night and the crash still had me stumped.  Unfortunately its getting into areas of the code that I wasn't familiar with like how we create and store our model data so I was having to learn how our code works in that area.

[dkulp]

I just stuck up a 2016.2b for the Mac that I'm hoping fixes some of the crashing.   I think there were a couple of pathways where a data variable wasn't being set before it was used that is now fixed.   Definitely should fix the crashes on imports, but there were some other model group things that may have hit the same issue.

[thedementedelf]

Gil...thanks for doing what you are doing.

[sean]

with the large restructuring of version 2016.xx, you should stay on 4.2.3 is you are actively sequencing, if you want to try out the new code then download.
Important Note: If you go back to 4.2.3 from 2016.2 , be sure you uninstall first. The old self installs from 4.2.3 will NOT remove the wxmsm*.dll libraries. You dont want our newer library with your older 4.2.3. So 1) Uninstall xlights, 2) download 4.2.3 exe, 3) install.

[Gilrock]

I pulled the latest checkin and it just seemed to have moved cjhacker's crash into the phase of opening the program instead of opening the sequence.  Most crashes will show me where it's crashing in the debugger but this one doesn't show anything so that's why it's giving me trouble to find.  It also doesn't seem to crash in the exact same spot which is also troubling.

[Gilrock]

By the way cjhacker,
Did you do any manual editing in the xlights_rgbeffects.xml file?  It has the following line for a model group and it should have a StartChannel field so not sure how that would have gotten written to that record.  That's also way to big of a channel number to be valid.  I thought that was going to fix it when I deleted it but it didn't.  I think our program just ignores the unexpected field.

<modelGroup StartChannel="275741436" name="Icicle Group" models="Icicles" layout="minimalGrid" GridSize="400" selected="1"/>

[Dazzled]

Hi Gil I didn't really intend for you personally to look at my report I thought I would include it as it may have been of some help for Dan just to save starting another post about this problem.
I really appreciate the amount of time and effort you and the rest of the developers devote to producing this fantastic program. I have no idea how all of you manage to have a job and work on Xlights as well as have time for a personal life and family.Thanks again.

Dan I have tried the 2016.2b release and have performed most tasks within the program and it seems to be working ok for me, thanks for such a quick fix.

Stuart

[Gilrock]

Dan made some changes that appear to have fixed this crash.  It's not obvious to us why it fixes the crash because the code changes appear to be functionally equivalent to the previous code.   Almost looks like a compiler issue where rearranging the code fixes the object code the compiler produces.

[CJHacker]

By the way cjhacker,
Did you do any manual editing in the xlights_rgbeffects.xml file?  It has the following line for a model group and it should have a StartChannel field so not sure how that would have gotten written to that record.  That's also way to big of a channel number to be valid.  I thought that was going to fix it when I deleted it but it didn't.  I think our program just ignores the unexpected field.

<modelGroup StartChannel="275741436" name="Icicle Group" models="Icicles" layout="minimalGrid" GridSize="400" selected="1"/>

Hey Gil,
Nope, I haven't manually edited any files that I recall.  Do I need to go in an manually edit that huge number to be something else?  or maybe just re-create the group?  I'm not sure why that would be such a huge number.

CJ

[Gilrock]

It probably won't matter but I just manually deleted that attribute from the file and it works fine.