Hi,
I just finished reading the 59 pages of this thread to be able to resolve my problems. The first one seems to be a known XDK bug, but I will still explain it here.
My bug seems to be similar to this one : https_://crosswalk-project.org/jira/browse/XWALK-2655 (delete "_" as I'm not able to post links)
My game runs perfectly while previewing over wifi on Chrome for Android, on my two devices : Samsung Galaxy Young GT-S6310N and Samsung Galaxy Tab3.
In the emulator of XDK, it works fine.
When exporting with Crosswalk for Android, then playing the APK on the devices :
1) On the tablet, I can see the title screen, then it crashes most of the time. I'm back on the menu screen of the tablet, then if I check the running apps, I can still see my app running there. When I click on it, I'm back to the C2's loading screen, which sometimes freezes, or it either crashes the game again, or succeed to launch the game.
2) On the mobile phone it's the same, except that I'm not sure the game started even once. It crashes nearly always.
Hi Rable, I definetely think you have encountered the X-Walk bug 2655 while testing on your tablet. However I think the crashes on your Galaxy Young are most likely caused by the weekness of your device.
Today Xingnan Wang from X-Walk project has updated the issue with this statement:
[quote:3a24rauw]The bug is a regression of chromium after M37, that could be reproduced on both Content Shell and Crosswalk.
It only happens when loading large mount of images(300+ images in the case) via file scheme. The allocation of shared memory for caching the url request(file://) will be randomly failed and then result the error of image loading in blink. Continue to root cause.
I'm trying a workaround encoding all the images in Base64 and embedding them directly in c2runtime.js using the great beufreecasse tool that you can find here: https://www.scirra.com/forum/random-freezes-on-windows-phone-8-device_t80070?start=20
This will avoid the flood of url requests and hopefully solve the problem until chromium team will fix this bug.
I'll keep you informed if this is a viable solution.