bluap
-
Content count
2985 -
Joined
-
Last visited
Posts posted by bluap
-
-
Looks like the mouse over handler is showing the spell the that the last touch is over. Most windows on Android disable that handler. It should be easy to fix but perhaps too late for the release. Radu has already submitted the update to the play store.
-
1 hour ago, ladrilho said:That worked, maybe I clicked on it by mistake... thank you.
Unrelated question, is there a way to arrange/move items in inventory to move them to the last row for example?
Good that the position restored OK. Normally, the title bar is enabled if the window does off screen (after resize, rescale etc). But that should not happen if a default position is used.
As for the inventory question, there's not an option like that other than moving each item individually. On android the cursor modes are the same as the desktop with some special actions when working with the trade and storage window to make it easier to move items between.
-
14 hours ago, ladrilho said:I'm still testing regularly and I just noticed something
The User Menu is behind the other chat channel UI elements (again this is not blocking, and might be related to my fussing around the font sizes and scaling)
Looks like you have the user menu title bar enabled so it moveable just like any other window. By default it is not movable and automatically places itself at the top of the screen spaced from the end of the chat controls. If you long press and select the "Lock in Standard Position" option, then it will return to the default state.
-
The latest build 1.9.5-20211212.1319 using old data : (same code as the 1.9.6 RC5 build which uses the new data)
- Fix android trade from storage bug - pass correct item position.
-
Added more first time configuration scaling, fonts and window positions.
- This can now be triggered for a new build and will be done when this build is first run.
- Fixes for Android backup - keep the size below threshold buy only backing up necessary files.
- Minor improvements for windows in default positions.
- Use true font width to set password manager size.
- Added option to automatically set the windows scaling factor (on by default for Android).
The 1.9.6 RC5 build uses the Google play store app ID and key so will replace any existing play store version. It can be installed in addition to the 1.9.5 version.
-
2 minutes ago, Grum said:Good find! I wonder though, if saving a single byte is worth the extra code complexity, and if we shouldn't just send two bytes always. Oh well.
I wondered the same but I'm not familiar with the protocol details, I just copied the method used by the non-touch code. I presume if we could, it would have been done using 2 bytes for both cases from the start. The server may do something special so Perhaps Radu can comment. That code was largely copied from the original Android port and I think we could re-factor a little now (one set of functions at least) but not this late in the release
-
I can reproduce the trade bug on Android and have fixed it, I'll publish new version tomorrow.
-
On 01/12/2021 at 8:01 AM, ladrilho said:I had the same "issue" with the UI scaling when I installed the latest version (UI too big).
My phone is a Galaxy s10+ with resolution FDH+ 2280x1080.
The scaling factor is set to 2.90 by default.
Is there something I can do to add more useful information for you?
Is 2.9 what you were getting before clearing your data or after clearing your data? If before, what is it now?
If 2280 is the native display size or the size of the EL window?
The android scaling factor is set for a new install depending on the main window width:
if (window_width<640) new_value = MIN_UI_SCALE; else if (window_width <= 1024) new_value = 1.1; else if (window_width <= 1280) new_value = 1.6; else if (window_width <= 1600) new_value = 1.7; else if (window_width <= 2160) new_value = 2.5; else if (window_width <= 2560) new_value = 2.9; else new_value = MAX_UI_SCALE;
So for you device, 2.9 is what would be expected. At some stage during the development, the client changed from using full screen to windowed with the Android top bar and possibly button bar to the right. Your original size may have been set before that change and the actual window size is now smaller. With the latest build, you can now set the client to run full screen but that will not change the scaling factor.
-
5 hours ago, Nogrod said:Yes, if people have a previous EL app version, they should either uninstall it or they should at least clear out the app data through App Preferences before launching the new version.
Unfortunately, it's a bit more complicated than that. The client uses a list of assets to ensure no old data files are left in place when they should not be. That means it should be safe to just update an already installed version without clearing storage or uninstalling first.
If you do clear storage you loose settings, counters etc. Likewise if you uninstall. In addition if you uninstall then you may have previous settings restored by Android automatically. This is a bit broken currently but I've been trying to fix this.
Also, the play store version and the test version are signed differently so cause conflicts. Though Radu and I can now use a common key to sign builds.
Once we have a new stable version I suggest people uninstall any old version then install the new version. Once installed, clear storage before starting for the first time. From then on, it will be safe to do updates without clearing storage or uninstalling.
-
9 hours ago, Entropy said:Ok, that's great to hear! Bluap, can you do another apk with those changes?
Good news indeed, looks to be a fixed for my wife's phone too. Nicely done!
I've built a RC4 android package with the fix, replacing RC3. As before, this is the new data version signed with the release key.
I've also done a build with the old data, signed with my debug key for those using the package from this thread:
The latest build el-1.9.5-20211130.2302.apk: (also in RC4)
- Fixed not closing the keyboard at login if using SSL.
- Enable full screen toggle (video tab).
- Fix issue for remembered windows positions if resolution changed before client restart (including portrait / landscape changes).
- And of course Grum's fix for crashing when fighting on some GPUs.
-
On 25/11/2021 at 7:50 PM, Grum said:Beats me. The only thing I can think of is that an actor animation command somehow subtly messes up the OpenGL state, which is exposed when drawing the banner. No idea which one, randomly commenting out stuff does not help :/ Either that or a bug in gl4es.
Speaking of which, the fact that there's exactly one version of gl4es that works is somewhat troubling in itself.
I've used a lot of time trying to get more recent versions working. I know Radu suggests getting in touch with the developer but I've not done so so far.
On 25/11/2021 at 7:50 PM, Grum said:I've been trying to bisect, but as Nogrod noticed, that's near impossible since the bisection points refer to the main branch which is missing all the Android stuff. Perhaps we should at some point merge the Android branch back into main, so this can be prevented for future bugs. Does not help us now though
Yep, I was going to propose merging after this release.
On 25/11/2021 at 7:50 PM, Grum said:Edit: we should resolve this soon. My wife wants her phone back
Same here.
-
6 hours ago, ladrilho said:I gave the new android client another try and the only "weird" thing I noticed is that the UI was really big for my screen when starting, almost unusable.
I had to go into the settings and change the User interface scaling factor.
Finding the option to reduce the size of the UI was not easy...
Maybe a future version could have a first time only config window appear to help the user define/confirm the scaling factor, and any other important settings. (I'll add this to the suggestion forum later).
That is strange, do you know the native resolution of your screen? Do you know that the scale was set too? The first time the client runs, it sets the initial scaling factor based on the screen resolution. It should probably be DPI based but I've found that unreliable previously. The simple config window would be a nice addition.
6 hours ago, ladrilho said:Regarding chat, I did miss the local chat separate from the rest.
I was trading with bots and it was a bit weird to have scroll back to see the local chat messages history.
I did not find the option to bring back the separate local chat, I assume it was removed to simplify the options and have less stuff there. (I'll add this to the suggestion forum later).
That's something I included from the original version, I'm not sure why those setting are disabled but I suspect it would OK to enable them. I'll check.
6 hours ago, ladrilho said:The new Keyboard button is quite nice, and really useful.
I did notice that the button is not scaling with the rest of the UI, but that could be a feature.
The button not scaling with the rest of the UI is not causing any usability issues for me.
The "Keyboard" button font is scaled using the chat font size; the button to be the same hight as the input field.
6 hours ago, ladrilho said:I did some harvesting, traveling, bot trading, NPC dialogs, and everything worked.
It does get a bit of time to get used to the android client controls, but if you spend a bit of time on it, the overall experience is quite nice.
Great work!
P.S.: I'll continue to use the client in the coming days and report any issues here.
Good to hear!
-
9 hours ago, Grum said:Just leaving this out here in case someone else wants to look further, because I have to stop now:
The crash seems to be related to Font::draw_ortho_ingame_string(), responsible for drawing the actor names and other banner text. It looks as if trying to draw even a single quad in there will cause a crash at some point, though I have no clue why that would be, or why the crash only seems to occur during or after a fight.
Interesting. I've tried turning off character names from the banner without modifying the client and have not had a crash so far. It might be worth others giving that a try. I note the banner names uses a pointer to the actor in the call the the draw function. Perhaps that's becoming invalid when the creature dies.
Edit: well I fought many more creatures but it did eventually crash. Seems like we're close. Trying no banners at all....
Edit 2: No crashes with no banners enabled. This with nearly a hundred despatched creatures. I'd be really interested to here how others fair.
You can turn off banners by long pressing the banner and selecting the context menu options. To restore the name banners if everything is off you can use "#keypress #K_VIEWNAMES". This needs improving ...
-
The message means that your client_state.json file is corrupt. If you've not edited the file and if other files are corrupt, then it is is probably due to something else such as disk errors. If you edited the file then you can still recover by fixing the error. The JSON format is text but has a specific structure that must be valid.
-
Just now, Aislinn said:The only way I know of to attack an invasion mob that can't attack you (capped below your a/d) is for you to mana drain them.
OK, sorry for the misunderstanding and wasting your time. Burn added bears and the test was for the MD spell. Perhaps you could clear them.
-
6 minutes ago, Aislinn said:Okay done, 300 brown snakes capped at 1. If you want/need more, let me know
Thanks. Perhaps I didn't mean capped. Burn set the creatures so they did not attack but you could attack them. I can't attack these.
-
2 minutes ago, Aislinn said:Yes, how many and where would you like them?
Thank you. Burn did something similar in crystal caves but Naralik would work. A few hundred?
-
BTW, I've been testing in Naralik on the test server and the brown snakes there look to be most likely to cause the crash. Would it be possible to place some invasion brown snakes in a location on test? Caped not to attack of course.
-
4 hours ago, Grum said:I'm guessing that's because that's where the OpenGL queue is flushed. Can you check that if you add a glFinish() before the SDL_GL_SwapWindow(), that the crash then occurs in the glFinish()? (It would be surprising if it's the buffer swap itself that causes problems).
The error message you get is also different from Nogrod's. Yours complains about a protocol error, in Nogrod's case it was warning about a resource deadlock. The underlying cause might well be the same, but I thought I'd point it out.
Not too sure, so far we've only seen this on a single chipset, and I've found more reports about bugs in the adreno driver. Of course, it might well be that it's a driver bug that is now exposed due to changes in gl4es. Nevertheless, it would be nice if we could fix this.
I get both the "protocol" and then, after a few more frames, the "resource deadlock" errors and the crash. I did try the putting a glFinish() before the the SDL_GL_SwapWindow(). Rather than a crash, the game freezes with the scene fixed. However the debug is still being produced so the games is still running. I've never seen anything like that before. If I switch from the game window then back, the game crashes.
<games scene frozen here> 11-19 13:44:00.027 12803 12864 I SDL/APP : draw_scene:156 11-19 13:44:00.027 12803 12864 I SDL/APP : draw_scene:158 11-19 13:44:00.027 12803 12864 I SDL/APP : draw_scene:160 11-19 13:44:00.034 12803 12803 V SDL : onPause() 11-19 13:44:00.049 12803 12803 V SDL : onWindowFocusChanged(): false 11-19 13:44:00.049 12803 12803 V SDL : nativeFocusChanged() 11-19 13:44:00.058 12803 12803 V SDL : surfaceDestroyed() 11-19 13:44:00.058 12803 12803 V SDL : nativePause() 11-19 13:44:00.115 12803 12864 I SDL/APP : Error: file: jni/src/gamewin.c line: 1554 message: OpenGL invalid value 11-19 13:44:00.159 12803 12864 I SDL/APP : draw_scene:156 11-19 13:44:00.166 12803 12864 I SDL/APP : draw_scene:158 11-19 13:44:00.166 12803 12864 I SDL/APP : draw_scene:160 11-19 13:44:00.166 12803 12864 I SDL/APP : Error: file: jni/src/draw_scene.c line: 161 message: OpenGL invalid operation 11-19 13:44:00.166 12803 12864 I SDL/APP : App entered background 11-19 13:44:00.167 12803 12864 I SDL/APP : Info: file: jni/src/json_io.cpp line: 32 message: int JSON_IO_Client_State::Client_State::save(const char *):770 [/storage/emulated/0/Android/data/com.el.elc/files/user/test/client_state.json] 11-19 13:44:00.168 12803 12864 I SDL/APP : Info: file: jni/src/json_io.cpp line: 32 message: int JSON_IO_Quickspells::Quickspells::save(const char *, Uint16 *, size_t):294 [/storage/emulated/0/Android/data/com.el.elc/files/user/test/spells_bluap.json] 11-19 13:44:00.169 12803 12864 I SDL/APP : Info: file: jni/src/json_io.cpp line: 32 message: int JSON_IO_Recipes::Recipes::save(const char *, recipe_entry *, size_t, int, recipe_item *):173 [/storage/emulated/0/Android/data/com.el.elc/files/user/test/recipes_bluap.json] 11-19 13:44:00.173 12803 12864 I SDL/APP : Info: file: jni/src/json_io.cpp line: 32 message: int JSON_IO_Counters::Counters::save(const char *, const char **, const int *, size_t, const struct Counter **):403 [/storage/emulated/0/Android/data/com.el.elc/files/user/test/counters_bluap.json] 11-19 13:44:00.184 12803 12803 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0x400c0907 (IOCTL_KGSL_DEVICE_WAITTIMESTAMP_CTXTID) failed: errno 35 Resource deadlock would occur 11-19 13:44:00.184 12803 12803 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-19 13:44:00.186 12803 12803 V SDL : onStop() 11-19 13:44:00.357 886 886 D SFPerfTracer: layers: (5:9) (StatusBar#0 (0xafdd4000): 9:256770) (com.android.systemui.ImageWallpaper#0 (0xafdc8000): 2:9598) (DimLayerController/Stack=0#0 (0xafc20000): 0:8352)* (Sprite#0 (0xafc4c000): 0:1463)* (NavigationBar#0 (0xafc47000): 10:47689) (DockedStackDim#0 (0xafc4e000): 0:11)* (com.el.elc/com.el.elc.ELCActivity#0 (0xafc5e000): 10:58)* (com.android.systemui/com.android.systemui.recents.RecentsActivity#0 (0xaff26000): 7:9) (thumbnail anim#0 (0xaff38000): 10:11)* 11-19 13:44:02.055 12803 12803 V SDL : onStart() 11-19 13:44:02.055 12803 12803 V SDL : onResume() 11-19 13:44:02.085 12803 12803 V SDL : surfaceCreated() 11-19 13:44:02.086 12803 12803 V SDL : surfaceChanged() 11-19 13:44:02.086 12803 12803 V SDL : Window size: 1920x1080 11-19 13:44:02.087 12803 12803 V SDL : Device size: 1920x1080 11-19 13:44:02.090 12803 12803 I SDL/APP : pixel format wanted SDL_PIXELFORMAT_RGBX8888 (2), got SDL_PIXELFORMAT_RGBX8888 (2) 11-19 13:44:02.092 12803 12803 V SDL : onWindowFocusChanged(): true 11-19 13:44:02.092 12803 12803 V SDL : nativeResume() 11-19 13:44:02.142 2661 7212 I ActivityManager: Process com.el.elc (pid 12803) has died: fore TOP 11-19 13:44:02.142 2661 8811 I WindowManager: WIN DEATH: Window{956445d u0 com.el.elc/com.el.elc.ELCActivity} 11-19 13:44:02.142 886 1118 E SurfaceFlinger: Failed to find layer (SurfaceView - com.el.elc/com.el.elc.ELCActivity#0) in layer parent (no-parent). 11-19 13:44:02.142 886 1118 E SurfaceFlinger: Failed to find layer (Background for - SurfaceView - com.el.elc/com.el.elc.ELCActivity#0) in layer parent (no-parent). 11-19 13:44:02.143 2661 7212 W ActivityManager: Force removing ActivityRecord{723b6c u0 com.el.elc/.ELCActivity t768}: app died, no saved state 11-19 13:44:03.198 886 886 D SFPerfTracer: layers: (9:19) (StatusBar#0 (0xafdd4000): 18:256807) (com.android.systemui.ImageWallpaper#0 (0xafdc8000): 18:9630) (DimLayerController/Stack=0#0 (0xafc20000): 0:8352)* (Sprite#0 (0xafc4c000): 0:1463)* (NavigationBar#0 (0xafc47000): 18:47737) (DockedStackDim#0 (0xafc4e000): 0:11)* (com.el.elc/com.el.elc.ELCActivity#0 (0xafc5e000): 0:69)- (com.android.systemui/com.android.systemui.recents.RecentsActivity#0 (0xaff26000): 0:59)- (thumbnail anim#0 (0xaff38000): 0:23)- (SnapshotStartingWindow for taskId=768#0 (0xaff38000): 0:8)- (thumbnail anim#0 (0xafe4d000): 0:8)- (com.el.elc/com.el.elc.ELCActivity#0 (0xafc5b000): 0:4)- (ScreenshotSurface#0 (0xafc5b000): 18:21) (com.android.launcher3/com.android.launcher3.CustomizationPanelLauncher#0 (0xafc5e000): 18:32)* (com.android.launcher3/com.android.launcher3.CustomizationPanelLauncher#1 (0xaff38000): 18:29) (BlackSurface#0 (0xafe4d000): 18:21) (BlackSurface#0 (0xafc0b000): 18:21) (BlackSurface#0 (0xafe27000): 18:21) (BlackSurface#0 (0xafe4a000): 18:21) -
I've narrowed the crash down to draw_scene.c:156 the crash occurs in the call to SDL_GL_SwapWindow().
diff --git a/draw_scene.c b/draw_scene.c index bd2a5fb2..ec180427 100644 --- a/draw_scene.c +++ b/draw_scene.c @@ -153,7 +153,9 @@ void draw_scene() elwin_mouse = -1; } +SDL_Log("%s:%d\n", __FUNCTION__, __LINE__); SDL_GL_SwapWindow(el_gl_window); +SDL_Log("%s:%d\n", __FUNCTION__, __LINE__); CHECK_GL_ERRORS(); /* stuff to do not every frame, twice a second is fine */
A few frames before the crash we have:
11-19 00:56:15.947 13358 13400 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 71 Protocol error 11-19 00:56:15.947 13358 13400 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot
Then:11-19 00:56:17.167 13358 13400 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 35 Resource deadlock would occur 11-19 00:56:17.167 13358 13400 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-19 00:56:17.167 13358 13400 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 35 Resource deadlock would occur 11-19 00:56:17.167 13358 13400 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-19 00:56:17.167 13358 13400 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 35 Resource deadlock would occur 11-19 00:56:17.167 13358 13400 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-19 00:56:17.168 13358 13400 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 35 Resource deadlock would occur 11-19 00:56:17.168 13358 13400 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-19 00:56:17.168 13358 13400 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 35 Resource deadlock would occur 11-19 00:56:17.168 13358 13400 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-19 00:56:17.213 2661 2681 I ActivityManager: Process com.el.elc (pid 13358) has died: fore TOP 11-19 00:56:17.213 886 2179 E SurfaceFlinger: Failed to find layer (SurfaceView - com.el.elc/com.el.elc.ELCActivity#0) in layer parent (no-parent). 11-19 00:56:17.213 886 2179 E SurfaceFlinger: Failed to find layer (Background for - SurfaceView - com.el.elc/com.el.elc.ELCActivity#0) in layer parent (no-parent). 11-19 00:56:17.213 2661 8195 I WindowManager: WIN DEATH: Window{8226635 u0 com.el.elc/com.el.elc.ELCActivity} 11-19 00:56:17.214 2661 2681 W ActivityManager: Force removing ActivityRecord{67d2337 u0 com.el.elc/.ELCActivity t756}: app died, no saved state 11-19 00:56:17.230 886 886 D SFPerfTracer: layers: (1:7) (StatusBar#0 (0xafdd4000): 0:254362)* (com.android.systemui.ImageWallpaper#0 (0xafdc8000): 0:9340)* (DimLayerController/Stack=0#0 (0xafc20000): 0:8304)* (Sprite#0 (0xafc4c000): 0:1456)* (NavigationBar#0 (0xafc47000): 0:45453)* (DockedStackDim#0 (0xafc4e000): 0:11)* (com.el.elc/com.el.elc.ELCActivity#0 (0xafc0b000): 0:29)*I've goggled around the error messages and rule out the following the Android SDK version and use of full-screen which are something suggested as the cause. Its looks very much like a GPU driver issue.
-
18 hours ago, Nogrod said:Here are some of the creatures I tested.
Thanks. Progress of sorts.
Turns out my wife's phone (a Moto G5 Plus) uses the same type of GPU as yours, an Adreno 506. It also crashes on fighting in a very similar way:
11-18 15:13:49.245 22392 22438 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 71 Protocol error 11-18 15:13:49.245 22392 22438 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-18 15:13:51.043 22392 22438 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 71 Protocol error 11-18 15:13:51.043 22392 22438 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-18 15:13:51.161 22392 22438 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 71 Protocol error 11-18 15:13:51.161 22392 22438 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-18 15:13:51.396 22392 22438 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 35 Resource deadlock would occur 11-18 15:13:51.396 22392 22438 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-18 15:13:51.397 22392 22438 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 35 Resource deadlock would occur 11-18 15:13:51.397 22392 22438 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-18 15:13:51.397 22392 22438 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 35 Resource deadlock would occur 11-18 15:13:51.397 22392 22438 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-18 15:13:51.397 22392 22438 W Adreno-GSL: <gsl_ldd_control:541>: ioctl fd 39 code 0xc040094a (IOCTL_KGSL_GPU_COMMAND) failed: errno 35 Resource deadlock would occur 11-18 15:13:51.397 22392 22438 W Adreno-GSL: <log_gpu_snapshot:450>: panel.gpuSnapshotPath is not set.not generating user snapshot 11-18 15:13:51.439 886 1252 E SurfaceFlinger: Failed to find layer (SurfaceView - com.el.elc/com.el.elc.ELCActivity#0) in layer parent (no-parent). 11-18 15:13:51.440 886 1252 E SurfaceFlinger: Failed to find layer (Background for - SurfaceView - com.el.elc/com.el.elc.ELCActivity#0) in layer parent (no-parent). 11-18 15:13:51.440 2661 10039 I ActivityManager: Process com.el.elc (pid 22392) has died: fore TOP 11-18 15:13:51.440 2661 3567 I WindowManager: WIN DEATH: Window{8ff860c u0 com.el.elc/com.el.elc.ELCActivity} 11-18 15:13:51.440 2661 10039 W ActivityManager: Force removing ActivityRecord{a4ef57 u0 com.el.elc/.ELCActivity t699}: app died, no saved state 11-18 15:13:51.447 886 1252 E SurfaceFlinger: Failed to find layer (com.el.elc/com.el.elc.ELCActivity#0) in layer parent (no-parent). 11-18 15:13:51.450 886 886 D SFPerfTracer: layers: (4:6) (StatusBar#0 (0xafdd4000): 0:245446) (com.android.systemui.ImageWallpaper#0 (0xafdc8000): 0:6851)* (DimLayerController/Stack=0#0 (0xafc20000): 0:8102)* (Sprite#0 (0xafc4c000): 0:1450)* (NavigationBar#0 (0xafc47000): 0:39479) (com.el.elc/com.el.elc.ELCActivity#0 (0xaff38000): 0:30)At least I can now join in the debugging so I'll see what I can find.
-
1 hour ago, Nogrod said:I compiled the app with gles4 repo HEAD set to commit a91f8d1e18362aa26303e84ef4ff776eef0973d2 but I still see the crash when fighting.
I also tried tag v0.9.3 and had to comment out the gl_hint in init.c to get it to compile but I still get the crash.
Shame, but thanks for trying though. Most of the other libraries are updated too but I'd have to check back to find what were used for 1.1.2.
I checked back though you other posts but could not tell if there was a particular creature that always caused the crash. Can you confirm which if there is?
-
Perhaps a more targeted approach would be to try earlier versions of the gles4 library. Given 1.1.2 used an earlier version of that library.
In the gles4 repo commit a91f8d1e18362aa26303e84ef4ff776eef0973d2 looked to be the baseline. This is close to tag v0.9.3. A while ago I tried to updated to never tags the most recent I could get working was v0.9.5, the version used in the current build. I had various issues with newer versions having tried up to v1.1.4. My rough notes:
v0.9.6 click cross weekly coloured
v0.9.7 click cross weekly coloured
v0.9.8 click cross weekly coloured
v1.0.0 click cross weekly coloured
v1.0.4 does not compile
v1.0.6 strippy lines
v1.0.8 strippy lines
v1.1.0 strippy lines
v1.1.2 strippy lines, flashing loc change
v1.1.4 strippy lines, flashing loc change, everything is white
-
On 16/11/2021 at 7:42 PM, Entropy said:Yes, if it's not much trouble please build them again, it is a good idea to thoroughly test them before release.
I've added a RC2 release on github that contains client only builds for Linux and Windows with the latest changes, plus Android packs for 1.9.6 data and 1.9.5 data.
-
31 minutes ago, Entropy said:Could it be because of the Cal3d library version the client is built with?
The Cal3d library used is the same as the desktop client.
9 hours ago, Nogrod said:This did not help with the crashes. I agree, EL isn't that heavy so I don't think the queue would be filling up. I only get this crash on this version of the Android client. On the old Android client v1.1.2 I never saw the crash issue.
The mob that I fight is what crashes the client, not the map I'm on as I first thought. It just happened that when I went to small maps, I was only fighting the mobs that don't cause a crash but after more extensive testing, I found that there are other mobs on the some small maps that always cause a crash.
For some reason, I thought you'd said the v1.1.2 client crashed in the same way, perhaps that was someone else.
The gl4es version I build against is slightly newer than used for v1.1.2, but not the latest. I've been unable to build a working client against the latest gl4es. We could try build against the same gl4es version as 1.1.2. I'll try that later....
The latest build 1.9.5p9-dev-20211116.2246:
- Fixes not displaying the last line for multi-select options.
The issue with scrolling the spin-buttons is gong to be a bigger change. Until then you can just touch the up/down arrows or enter values directly
[Closed] New development version of the Android client
in General Chat
Posted · Edited by bluap
Added 196 updated build · Report reply
It was the mouse over handler. Only the help information was incorrect, nothing as bad as the trade bug but its fixed now anyway. I also fixed another bug when only the initial windows mode use the default window positions. This one effects the desktop client too.
The latest 195 build 1.9.5-20211218.2131 (old 1.9.5 data package), the latest 196 build 1.9.6.0-1 (new data pack)