search

area515 / Photonic3D

Control software for resin 3D printers
http://photonic3d.com
GNU General Public License v3.0
131 stars 115 forks source link

[bug] : Dev build : Java heap space error #210

Closed kloknibor closed 8 years ago

kloknibor commented 8 years ago

Hi,

I started the same model quite a few times for testing purposes, lowered my exposure time to 10S and used the following code for Z lift :

G1 Z6.0 F100.0 ; 12200 G1 Z-5.9 F100.0 ; 14200

and set it back later to

G1 Z6.0 F30.0 ; 12200 G1 Z-5.9 F30.0 ; 14200

The file which worked previously now returns an heap space error... Also an new uploaded file creates an heap space error! I suspect that something is not cleaned well... Here are the logs :

LogBundle (3).zip.pdf

kloknibor commented 8 years ago

PS : not sure if this has to do anything do with #111 but might be ;)!

WesGilster commented 8 years ago

Can you attach the stl file that would load?

It doesn't really have anything to do with cleaning. It's the same stl loading bug that' I've had trouble recreating. For some reason I read an incorrect "HUGE" number of faces into Number_faces and explode the memory in one statement: Temp_Info = new byte[50*Number_faces]; // Each face has 50 bytes of data

I need your file to determine why that format does that sometimes.

WesGilster commented 8 years ago

Dup of #108. Still need that stl file though.

kloknibor commented 8 years ago

I'm not the owner of the file so I'll send it to you in private instead of posting it here for the world to see ;)

WesGilster commented 8 years ago

I was thinking about working on this, but the stl file that you sent me slices just fine. I've only tried this on Windows. I'll try to reproduce this on Raspberry Pi, but if I'm unsuccessful, I might have to have you give me the steps to reproduce it.

WesGilster commented 8 years ago

Tried on Raspberry Pi and couldn't reproduce. Do you have further steps to reproduce this?

WesGilster commented 8 years ago

This bug kindof prompted me to do some memory profiling today and I found that I had a pretty nasty memory leak. I used a 50mb stl model for my profiling.

Notice the second time I attempted to print that model, the jvm didn't even collect the first print job. image

After I fixed the bug, it dropped well below where the STL file was originally loaded. So we are looking pretty clean now. image

WesGilster commented 8 years ago

Fixed in Dev-006