When i minimize a server(or even a client) for a few minutes(around 10-15minutes), the rendering in the window ceases.
It also stops rendering the title bar of the window...although the X at the top right corner works...but you cant see it(i just guessed where it is and clicked)
AUTOPAUSE TRUE or FALSE?
FALSE
All my projects are affected like this.
The new one in beta section does too, noaa goes project. (http://www.glbasic.com/forum/index.php?topic=9952.0)
You have to minimize it for a while.
When i come back to the program, it wants to link to the shortcut, instead of restoring the window.
If i click the virtual window above the taskbar icon, the window comes back up, but its all black, including the title and minimize/maximize/exit buttons.
-Windows 7 64bit
amd a10 trinity w/8 gig ram, and an ssd....system is fast, and clean with not much installed except the important updates and security stuff...no junk at all.
more info....the program is still running....it is still responding to buttons, i could tell because i have a key that makes it go fullscreen.
it is still downloading images on intervals too...seems to be fine, except rendering.
only happens to glbasic binaries..other programs run fine, and restore rendering when clicked into
it's not responding? Or just not redrawing?
When the window comes back up...black screen, even the top title bar of the window.
I have some ideas of how this might be happening..
a) internet browser plugins or something, video players? I think youtube.
b) directx programs.
c) theres a moth on the transistor.
ps i have all the latest drivers for os and video. i made sure.
pps. this doesnt happen if i run the same programs in linux using wine as the loader....the programs run indefinetly, very good i might add(although i havent tested 3d yet...soon)
There was a post about this somewhere, which I cant find.
Essentially, what happens is that a Windows program uses an extra 0.1Mb per second when minimised.
Got a video of it here : http://www.dailymotion.com/video/x24t6n1_glbasic-memory-usage_tech
Property all textures have wiped out and needs to been reload them again. This can been happens on some graphics drivers.
Have you checked the size of the texture after a resume, if they got wiped?
Good idea...
errr the screen turns black, so does the title bar for windows.
Doubt this is anything to do with loading images.
Fonts certainly aren't removed from memory, and I doubt sprites would either - that would only happen on a screen resolution change.
more info here...
it is not something else in windows interfering...
I minimized the program, and ran it for a while.
And i did NOT run anything else at all.
Still same issue...black window after a few minutes...about 5-10 minutes.
Yes, the memory leak does need fixing
fuzzy70 experiened a blackout too, in windows 7, he reported this in the beta thread of noaagoes.
On this same note, he also tested in windows 8.1 , and didnt have this effect.
That is odd, as I can replicate the problem too - although only for debug mode. It's possible that the problem exists with release mode but for previous systems...
I think it might be opengl.
I rather would bet on the window/context creating/handling stuff. But since we dont know the code, that are just blind guesses.
I can reproduce that memory leak. But I can't locate it. I even installed "Visual Leak Detector" - nothing. I have no idea where it happens. I think it's even in SHOWSCREEN.
:(
Multiple surfaces aren't being created are they ?
No. It's a memory isseu. Not graphics memory. I tried gDebugger as well. Nothing. Spent the while evening yesterdsy. :(
Gesendet von meinem GT-N7100 mit Tapatalk
hopefully not a sdl issue?
It doesn't sound like it...
I deleted my last msg....
I take back my remarks...i did NOT find a leak in the console....i just shut down the agent, and the memory stayed the same...
The leak im seeing is in my antivirus program, unfortunetly.
If you need help running tests of some sort, just let me know, my computer is free for testing long term tests.
Is there a simple (only a few lines long) test program, that always leaks for sure under some specific conditions?
The test programs I provided on the first page will always cause the problem (after a while) in debug mode.
Sorry I must be blind, but I can not find any test program posted by you in this thread. Also, did you try to make it as few lines/commands as possible, so its easier to track it down?
Ah yes, the test program was elsewhere :)
Run this code in debug mode and minimise the window. Watch the memory use continually increase :
WHILE TRUE
SHOWSCREEN
WEND
Eventually it will come to a point where the window wont be able to display anything.
@Kitty
Lets solve this? I want to help...its important, for all of us. (we cant detect our own leaks using task man)
The used memory in the RAM, stays used, even after shutdown of the program, observed in task manager win7.
From what i see in all the memory leak threads, this is a new clue, for starters.
I know this is a sore subject for you, but i am confident we can solve this together, the mountains you have climbed already are proof!
I have more info for you gernot........you mentioned you couuldnt find the leak using visual leak detector.
I research that program and found some restrictions which may be a clue for you!!!!!!!!
Quote
If more than one copy of the same C Runtime DLL is loaded in the process at the same time, then some leaks may go undetected (note that loading more than one copy of the C Runtime DLL at the same time is probably a bad idea to begin with).
Quote
Memory allocations made through calls to functions loaded from a DLL using delayed loading may not be detected.
I dont know whether it is used to Windows 10 or this machine is using the demo version, but the problem doesn't seem to be happening now.
How long did you run a test with just showscreen?
i l0se a gig of ram in an like 4 hours.
It was the for the length of the demo - memory actually goes down halfway through...
Interesting, is the memory being wasted until reboot?
Just fishing you to get some more clues. :whip:
I dont understand it either - I did a reboot to see if a program was causing the memory release, but no... I cant use the full version as I dont have access to my computer unfortunately.
Its possible that Windows 10 memory allocation/release is better (I always thought there was a problem with Windows 8 in that regard), or graphics drivers are better or DirectX 9 has been improved.
Or it could be all 3.
This is with the 12.312 version.
Is this happening with all GLB apps? Or, IIRC, was it happening on the weather pic app Hemlos did?
Sometimes I leave my GLB engines running for many hours >4 to see if something gets stuck logic wise and I
never found any significant memory leak.
So I have my map generator that I can leave running for many hours, would that count as a test to figure something out for this?
If so, how could I proceed to be usefull?
I've tried the DDgui example program and that is stable at 28.1Mb in debug mode (which is 1 Mb more than release mode).
Well....im about to test v12 update and v14 too. :help:
Im not going to test new update for v12...because its bugged setscreen and so is v14.
HOWEVER, in windows 7...v14 fixed the memory leak. WOO!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Not only did this fix the leak, it cleans corrupted ram from a pulluted version of my program that ran earlier today during this boot session!
OPENGL SERVERS GOOOOOOOOOOOOOOOOOOOOOOOOO