Problem with particles V1.12

[Dad72]

Hi,

 

It has been several times that I realize that with version 1.12 with the particle system, after around 3 minutes or 4, the browser (google chrome) crash.

I have this problem only with particles from the version 1.12.

 

Sorry to report bugs, I promise you that I research not. 

 

Thanks.

[GameMonetize]

I cannot reproduce this bug even after 10 minutes on http://www.babylonjs.com/index.html?OMEGA

[Wingnut]

Sadly, I have verified the crash.  (I'd rather give good news than bad, ya know?) But I have interesting news.  Here are three versions of my particle editor... one using babylon 1.10.0, one using babylon 1.11.0 and one using babylon 1.12 beta (/*! Babylon 2014-05-13 */)

 

http://urbanproductions.com/wingy/babylon/d72particles/liveEdit04-110.htm

http://urbanproductions.com/wingy/babylon/d72particles/liveEdit04-111.htm

http://urbanproductions.com/wingy/babylon/d72particles/liveEdit04-112.htm

 

I browsed all three with: Google Chrome 35.0.1916.114 m (downloaded tonight) 

 

ALL THREE VERSIONS crash Chrome browser in under two minutes.  What the heck, eh?  Sorry. 

 

Also, the mouse doesn't seem to work in the 110 version, but that might be because I used hand.minified-1.3.7.js for all three versions, and hand.minified-1.3.7.js might not be compatible with BJS 110. 

 

I tried keeping the console open for the crash.  Nothing is reported there, other than "Inspected Target Crashed".  I tried switching to hand-1.3.8.js.  No change seen.

 

Dad72 seems to be able to find bugs even when they are hidden in the dessert under a kilometer of sand, eh?

 

Dad72... can you test if you are seeing Chrome crashes in all three of the particle URLs listed above? (thx!)  Anyone else who would like to verify or dispute any of this, please test and comment freely, and thank you.

[kolar]

I can reproduce this crash in recent Chrome and Canary versions. There is problem with memory deallocation. In recent Firefox everything works ok. Try: http://jsfiddle.net/kolar/gPt7U/7/ click 'run' and watch how memory goes up and up.

[Dad72]

Your demo DK, use an older version of babylon that works for mee. I speak of the version 1.12. but may be I did not noted in 1.11 if this bug was present. but I know that I hadnot before. I doesn't seem to be the only one to reproduce this bug.
 
I also have the error with your Wingnut demo. 
 
I actually think that it is a matter of memory, because when I start my PC, it does it is not produced immediately,  the duration when occurs the crash seems random.
 
PS: I do not know if it helps, but it would appear after local use that crash. not when used as accommodation online. Demos Wingnut does not crash if use that his only link , but if I'm local I use my editor to check some time after the crash demo wingnut. so it must be linked to a local use WebGL.

[GameMonetize]

Ok this is a bug with latest version of Chrome. If you launch your url with f12 opened and you launch a memory profile there is no problem. Once you close f12 the problem is back.

 

I still have no leak on firefox and IE

 

The funny thing is that Opera which is based on Chrome code has also the leak

[Dad72]

We have to wait an update of Chrome? Are there in the run this memory leak bug? Who can report explaining properly the problem? my English is too bad to explain this.

[GameMonetize]

http://stackoverflow.com/questions/23120851/webgl-deletebuffer-leaking-memory

[Dad72]

I've just updated Chrome, I do not know if this is an update recent or new ? I was a user of Firefox before, so I do not know how Chrome updates (manually or automatically)

 

What is your current version of Chrome Deltakosh or Wingnut?

 

My version is the:  35.0.1916.114 m

 

whatever it is old update or new, the bug is still there.

[Wingnut]

Hi dad72.  I'm using the same as you.  (I told the version number in my previous post)

 

Yeah, I think we will need to be patient and let the Chrome team find their memory leak.

 

@dk - interesting find with the f12 thing.  I have no problems with IE or Firefox, either.  Just Chrome.  Its not like Chrome is an enjoyable browser to use, anyway.  hehe.  It seems like a bucket of bison snot. 

[GameMonetize]

This fills me with joy and hope for IE

[Wingnut]

I agree, DK.  To be frank, I think Microsoft has done a pretty good job with its first webGL-active browser version (IE11).  It has a real fast render-loop speed, and other than some specular color differences between IE11 and FF, it has done well.  Every demo I have built with babylon.js... runs quite well on IE11.  The main problem I have with IE11... is that the animations run much faster than they do in FF.  That's the kind of problem that I consider a GOOD problem.

 

It is all fresh and new, yet.  Some patience is certainly warranted.  In general, I think the whole situation is in good position.  The browsers are acting fairly well, the framework is acting well (and getting some great new powerful features), the users are having tons of fun, the forum is active with great ideas and great projects... we couldn't ask for much better, really.  I am as excited about the babylon.js project as I have always been... and maybe more.  As long as the world doesn't come to an unexpected end, our future looks quite good.

[Dad72]

Hi dad72.  I'm using the same as you.  (I told the version number in my previous post)

Oops, Yes it's true, I had not seen her or I did not pay attention. I must resolve this issue since it is not a bug of babylon, or not ?