Crash upon hitting PLAY

[jamills6377]

Perhaps the way I worded it was misleading - I meant that I was happy with what I was doing and accomplishing and was content to stick with LOR but that I recently I couldn't accomplish what I wanted to with their current software and I began looking around - thus coming across xLights which did allow me to accomplish what I had hoped for EXCEPT for this blasted crashing problem!

I can understand that the butterfly effect would be pretty darned intense in terms of data....but you saw the sequence I was testing with - and it was nothing like a butterfly effect. I was testing with a singing face - just the face - no additional effects involved at all....and it crashed almost immediately.

So what do we do next - is there anything I can do to test anything further?

And, what if I had used 4 - Pixie4s instead of one Pixie16? Would that really be any different - they would still be all the same effects on all the same pixels - just multiple controller addresses all on the same network, right? So the end result would have been the same??

I also take it that you can't tell at all what is going on with their Enhanced protocol - to try to replicate it in any way???

So - because the 11th hour is very nearly upon me with this year's show - do I just give up on using xLights? I currently have all my sequencing done in xLights for my show - and no way to get that show up on the pixels! I can't even get the export of the clipboard of just the matrix to copy into LOR S5 properly (the matrix portion of the sequencing being the part that xLights the really does for me those things I want - I could do the rest with LOR if I had to). I'm afraid there will be no Halloween show at all and my Christmas show will be without all the bells and whistles I had hoped for if I have to now start over again in LOR.

[Gilrock]

I don't know if it will help your crash but I was having a blue screen crash with that LOR output when turning off output to lights with a lot of channels.  I implemented a fix that is now preventing that crash.  What I do is inspect the serial transmit queue and if its still not empty from the previous frame then I skip the current frame.  That keeps it from stacking up a ton of data.  I also added log messages that will let us know how many bytes you are transmitting and will record when frames are skipped.  Keith might get it released tonight.

[jamills6377]

Ok - cool! I'll watch for that new release and download and test as soon as I can! I do appreciate all of your hard work on this.

I guess the one question I would have about this is - if it's dropping frames isn't that going to make any sort of animation attempt lok jumpy? Of course, jumpy is better than a crash, of course!

[keithsw1111]

Jumpy or laggy.  They are the only two options when the output can’t keep up.


Sent from my iPhone using Tapatalk

[jamills6377]

Simple stuff worked like before - but as soon as any load was put on the system - used Monster Mash again because it's the same one I"ve tested with in the past - as soon as the Faces effect cam up it crashed...played the initial part of the sequence OK - no obvious jumping or lagging - but as soon as it was time or the face, it crashed.
Attached the sequence  and log files.

[Gilrock]

Well I can't get mine to crash like that so I don't know what to look at.  It looks like you have 4 of the outputs running and I only have 1 so that's one difference.  Those Faces effects may not look that complex but according to the log each output was averaging 4,000 bytes per frame and it was doing the skipping properly.  Pretty much every other frame is going to skip at that size because it would take 1.5 seconds to send 20 frames of 4,000 bytes over a 512Kbaud output.  So if you are running 20 FPS it can't keep up.  It's crashing inside the function that loads the data packet for that output so maybe there is a problem in that code for some outlying condition but its not easy to find unless its happening to me.

[jamills6377]

Is there anything else I can run, or log, or do to try to get you more data to work with. I'm not at all computer illiterate so can do pretty much whatever you need - keeping in mind that I quit programming before learning much o what you guys are programming in - yet if I look at your code or logs I can read it enough to know what is going on - just couldn't begin to write the code rom scratch myself.

Maybe I should look at that language a bit more closely - I understand the principles just haven't even used it -- have done object oriented web based stuff in the last couple o years I worked (now retired)...having started back in the large IBM mainframe days with COBOL then learning Visual Basic and SQL and web design and PHP, etc.....so perhaps if you tell me what to do to help, I could do something more with this problem to help resolve it.

[Gilrock]

Even if we stop the crashing I'm not sure you'll be happy with the lag you're gonna have using that many channels.  I can try putting a bunch of log messages inside the function to try to narrow down where in the function it's crashing.  It usually takes several iterations and since its during the week you won't get updates from me very often.  Why again are you not able to just use LOR for everything this year?  If you have your channels matches up can't you just use the Convert Dialog and export to an LMS file?

[jamills6377]

Well - you guys convert to S4 - and because of number of pixels I have basically been forced into S5.....so I can't just export/convert.
When I try just exporting a clipboard it doesn't work, when I try importing an exported avi al o it isn't in there.
That is what I have another thread running here in the forums about the clipboards exporting.

[Gilrock]

So if you had LMS files in S4 they were useless and unable to be opened in S5?

[jamills6377]

When S5 tried to open them they would not convert because there was no exported "configuration" data included in the file so S5 didn't know what to do with any of it....or at least that is what LOR tech support suggested was the problem - the convert process ran for 4-5 hours and I finally forced the program to quit - then asked for tech support help on that issue and that was what I was told - no configuration information included in the file.

[Gilrock]

So no way to also have S4 on your machine?  I'm pretty sure S4 would open the file and then you can import a saved channel configuration from another file.  Then you would have a file that should work in S5.

I can try to add some log messages and send you a new xlights.exe tonight.

[Gilrock]

You know you don't need to write code but if you were able to get everything setup and building in Visual Studio you could run the code with a debug build and when it crashed hopefully be able to see in the debugger which line its crashing on.  Then if we were in Zoom we could inspect where there might be a null pointer or array boundary violation.  There are ReadMe's in the code base that have the installation/compiling instructions.  First step would be to install GIT for Windows and figure out how clone the xLights GIT repository.

[jamills6377]

I'll try opening in S4 - I haven't tried that because I thought I'd have to set up my whole display all over again in S4 - and S4 doesn't support 100-pixel strings which most o mine are - hence the need to run in S5- and that seemed daunting to me or some reason.

I could try importing it into S4 because i have two Windows machines (one still has S4 on it)...see what happens.

I don't have Visual Studio on my Windows machine, at least it doesn't come up in my "Start Menu" programs listing. I'll have to look at the Microsot site and see what's out there.

Also, not familiar with what GIT is - but I can learn (and typically can learn new things in programming, etc. quickly). I'll google Git and see what comes up.

[jamills6377]

OK - I found GIThub and created an account - searched and found smeighan/xlights.
But have not done anything further yet.
I see the Clone or Download - but am not sure...should I do that on my S5 or my S4 Windows machine? My lights actually run from my S5 machine...so I suspect it would need to be there...but I typically sequence and mess around on my Mac (or the S4 machine) then copy it over to test it.