Macintosh OS X client 1.5.0 bugs thread

[0ctane]

This is the "official" OS X bugs thread for the 1.5.x client release. I (or someone else) will update this first message as problems are identified and fixed.

 

A MUG (Macintosh User Group) list is here.

 

You can get the game here: Version 1.5.0.1c

Mirror at http://www.eternal-lands.us/downloads/el_osx1501a.dmg

 

Current EL version 1.5.0

Mac version .1c (10.4+ Universal Binary)

 

Note:

If you are looking for it the el.ini file is found in ~/Library/Application Support/Eternal Lands under the "main" directory currently.

 

Note 2:

PLEASE, when you post a bug, include the type of Mac you use and the OS version.

 

Known issues (other than those listed in the ReadMe):

There are problems with reflections. If you want to use reflections, you need to have Framebuffer activated

and

a Shader Quality Setting greater than zero. If not, you will get odd effects where the ground disappears when near water. These are all found in the options window.

 

There is one report of the problems with the mouse pointer. I do not think this affects game-play, but it is still a potential bug.

This was primarily built on a i386 MacBook Pro.

 

Everyone thank Florian for a lot of debugging. Also thank Fedora for getting copy & paste to work. Copy is ctrl+c, paste is crtl+v inside EL (I will look into getting normal Mac behavior for this).

 

Screenshot support added in v1501c. ctrl+p will get a snapshot. These are saved in ~/Library/Application Support/Eternal Lands under the main/screenshots directory.

 

Finally, stay on topic please. This thread is for identifying bugs in the OS X client.

Edited December 6, 2007 by 0ctane

[Fedora]

I compiled the PowerPC client and found two bugs in the code that i think will affect the intel version too:

 

 

- Withdrawing items from storage:

in items.c line 622 (or near, i added some printf here and there to debug)

 

*((Uint16*)(str+1))=storage_items[storage_item_dragged].pos;

should be

*((Uint16*)(str+1))=SDL_SwapLE16(storage_items[storage_item_dragged].pos);

because now the pos field is 16bit

 

 

- Looking at storage item

Same issue, in storage.c at line 376 or so, another SDL_SwapLE16 is needed.

 

 

 

There are also random crashes when sound is enabled.

[Learner]

I compiled the PowerPC client and found two bugs in the code that i think will affect the intel version too:

 

 

- Withdrawing items from storage:

in items.c line 622 (or near, i added some printf here and there to debug)

 

*((Uint16*)(str+1))=storage_items[storage_item_dragged].pos;

should be

*((Uint16*)(str+1))=SDL_SwapLE16(storage_items[storage_item_dragged].pos);

because now the pos field is 16bit

 

 

- Looking at storage item

Same issue, in storage.c at line 376 or so, another SDL_SwapLE16 is needed.

 

 

 

There are also random crashes when sound is enabled.

This won't affect the Intel version, because the bytes are in the proper order.

[Rhapsodist]

-

Edited August 16, 2010 by Winkz

[0ctane]

This page is currently a placeholder. This page will be updated when the OS X client is available. Do not post here yet.

While I appreciate people already finding bugs....

 

If you are interested in helping Florian and me with the OS X client coding, please contact me via PM. We are always looking for more help.

[Enly]

Everything eems to be working fine as far as maps as long as i have shadow mapping and the framebuffer on. Thanks 0ctane. There are random crashes while changing maps (like other people posted). Something else i noticed, changing volume on the inventory sound effects appears to have no effect. Even when you turn it all the way down it is still there.

 

-edit-

 

Copy is ctrl+c, paste is crtl+v inside EL (I will look into getting normal Mac behavior for this).

Is this supposed to work to copy and paste into and from EL with other applications? because if it is it dosn't work for me.

 

-edit2-

Changing font size for the notepad has no effect either.

Also note, I have not read any of the bug threads for windows versions. My apologies if I am reposting universal problems.

Edited November 29, 2007 by Enly

[Fedora]

@Enly: from other apps you have to use normal shortcuts (apple+c, apple+v). Inside EL it is ctrl+c, ctrl+v

 

 

from 0ctane:

 

BTW, to get a screenshot on a mac, Apple+Shift+3. Libpng is not needed.

 

I get wrong screenshots with that combo (missing actors or missing bars). My compiled ppc client has libpng statically linked and ctrl+p works fine, leaving a screenshot under Library/Application Support/Eternal Lands/main/screenshots

 

Is it possible to add libpng in the universal package?

 

Sound is still buggy: left alone it crashes after random amount of time, and often when changing maps.

 

 

The link is broken, this works: http://www.0ctane.net/files/el_osx1501a.dmg

 

 

IBook G4 1,2GHz + 10.4

Edited November 29, 2007 by Fedora

[Enly]

@Enly: from other apps you have to use normal shortcuts (apple+c, apple+v). Inside EL it is ctrl+c, ctrl+v

 

I was doing that, but it wasn't working. It wasn't EL though, something with my own computer but i fixed it. Sorry.

 

Also, I thought I had the reflection problem gone, but i crashed from the map change and when i came back, it was back, and no amount of changing the shadow mapping or framebuffer would help. So i just had to turn reflections off.

-edit-

spelling

Edited November 29, 2007 by Enly

[0ctane]

from 0ctane:

BTW, to get a screenshot on a mac, Apple+Shift+3. Libpng is not needed.

I get wrong screenshots with that combo (missing actors or missing bars). My compiled ppc client has libpng statically linked and ctrl+p works fine, leaving a screenshot under Library/Application Support/Eternal Lands/main/screenshots

Is that with your build or mine?

Is it possible to add libpng in the universal package?

Yes. I can look into it.

Sound is still buggy: left alone it crashes after random amount of time, and often when changing maps.

It could be a conflict with alut. I had to put a hack/fix in since alutLoadMemoryFromFileImage is not available for Macs. While I had no problems yesterday, I just got a crash. Happened in the update_sound method.

The link is broken, this works: http://www.0ctane.net/files/el_osx1501a.dmg

Thanks. I fixed it in the first post.

 

Also, I thought I had the reflection problem gone, but i crashed from the map change and when i came back, it was back, and no amount of changing the shadow mapping or framebuffer would help. So i just had to turn reflections off.

Did you try a shader value greater than zero while having framebuffer on? I tested this out.

 

Frame buffer: on, reflections: on, shader quality: 0 -> problems

Frame buffer: on, reflections: on, shader quality: 1 or 2 -> okay

Seems like shadows and shadow-mapping has no effect on the reflection problem.

[Fedora]

I get wrong screenshots with that combo (missing actors or missing bars). My compiled ppc client has libpng statically linked and ctrl+p works fine, leaving a screenshot under Library/Application Support/Eternal Lands/main/screenshots

 

Is that with your build or mine?

 

with both. And also with 1.4. Every screenshot app relying on system calls to get a screenshot have this problem too. One out of 3 or 4 screenshots is good, the others have problems. Ctrl+p with libpng never fails for me.

 

 

I'll look in the sound code as soon as i get some free time

Edited November 29, 2007 by Fedora

[Enly]

Did you try a shader value greater than zero while having framebuffer on? I tested this out.

 

Frame buffer: on, reflections: on, shader quality: 0 -> problems

Frame buffer: on, reflections: on, shader quality: 1 or 2 -> okay

Seems like shadows and shadow-mapping has no effect on the reflection problem.

 

Hmm, seems to be working now with the higher shader value. But then again it also seemed to be working fine before, and it came back after i crashed. I'll post if I have any more problems with it

Edited November 29, 2007 by Enly

[Fedora]

I compiled my client again adding debug info (_EXTRA_SOUND_DEBUG) and a printf at the beginning and ending of alutLoadMemoryFromFile and alutCreateBufferFromFile.

Then i ran the client on the test server, killed a bear in votd, changed map to dp and crashed. Error log showed no error messages and really weird not even trace of my added printf, as if those functions are never called. Moreover searching through sound.c it seems that alutCreateBufferFromFile is defined but never used.

 

I ran the client again using gdb. It crashed after a while and this is what i get:

 

Program received signal EXC_BAD_ACCESS, Could not access memory.

Reason: KERN_INVALID_ADDRESS at address: 0x0f63f3e0

0xffff8910 in ___memcpy () at /System/Library/Frameworks/System.framework/PrivateHeaders/ppc/cpu_capabilities.h:189

189 in /System/Library/Frameworks/System.framework/PrivateHeaders/ppc/cpu_capabilities.h

 

STACK TRACE

#0 0xffff8910 in ___memcpy () at /System/Library/Frameworks/System.framework/PrivateHeaders/ppc/cpu_capabilities.h:189

#1 0x9420f6d4 in CABufferList::AllocateBuffersAndCopyFrom ()

#2 0x94208a4c in AudioConverterChain::FillBufferFromInputProc ()

#3 0x942081d4 in BufferedAudioConverter::GetInputBytes ()

#4 0x9420802c in CBRConverter::RenderOutput ()

#5 0x94207da0 in BufferedAudioConverter::FillBuffer ()

#6 0x94207f1c in AudioConverterChain::RenderOutput ()

#7 0x94207da0 in BufferedAudioConverter::FillBuffer ()

#8 0x94207c2c in AudioConverterFillComplexBuffer ()

#9 0x0534c87c in OALSource::DoRender ()

#10 0x0534c338 in OALSource::SourceInputProc ()

#11 0x70038524 in AUMixer3DEntry ()

#12 0x700385b4 in AUMixer3DEntry ()

#13 0x7003925c in AUMixer3DEntry ()

#14 0x70039cd0 in AUMixer3DEntry ()

#15 0x700c7200 in dyld_stub__keymgr_get_and_lock_processwide_ptr ()

#16 0x700c8dfc in dyld_stub__keymgr_get_and_lock_processwide_ptr ()

#17 0x700086dc in DefaultOutputAUEntry ()

#18 0x7000f838 in AUHighShelfFilterEntry ()

#19 0x70008dac in DefaultOutputAUEntry ()

#20 0x700c9568 in dyld_stub__keymgr_get_and_lock_processwide_ptr ()

#21 0x700c9298 in dyld_stub__keymgr_get_and_lock_processwide_ptr ()

#22 0x94208dbc in AudioConverterChain::CallInputProc ()

#23 0x94208a0c in AudioConverterChain::FillBufferFromInputProc ()

#24 0x942081d4 in BufferedAudioConverter::GetInputBytes ()

#25 0x9420802c in CBRConverter::RenderOutput ()

#26 0x94207da0 in BufferedAudioConverter::FillBuffer ()

#27 0x94207f1c in AudioConverterChain::RenderOutput ()

#28 0x94207da0 in BufferedAudioConverter::FillBuffer ()

#29 0x94207c2c in AudioConverterFillComplexBuffer ()

#30 0x70008970 in DefaultOutputAUEntry ()

#31 0x700c8dfc in dyld_stub__keymgr_get_and_lock_processwide_ptr ()

#32 0x700086dc in DefaultOutputAUEntry ()

#33 0x70007d78 in DefaultOutputAUEntry ()

#34 0x91466cf4 in IOA_Device::CallIOProcs ()

#35 0x91466a08 in HP_IOThread::PerformIO ()

#36 0x91464a18 in HP_IOThread::WorkLoop ()

#37 0x91464580 in HP_IOThread::ThreadEntry ()

#38 0x914553dc in CAPThread::Entry ()

#39 0x9002bd08 in _pthread_body ()

 

Any clues of what's happening?

[Enly]

I was just going about playing when some red splotches popped up over the map. It was only from certain points with certain camera angles, and only if shadows were on. So I've turned shadows off for now.

[0ctane]

I compiled my client again adding debug info (_EXTRA_SOUND_DEBUG) and a printf at the beginning and ending of alutLoadMemoryFromFile and alutCreateBufferFromFile.

Then i ran the client on the test server, killed a bear in votd, changed map to dp and crashed. Error log showed no error messages and really weird not even trace of my added printf, as if those functions are never called. Moreover searching through sound.c it seems that alutCreateBufferFromFile is defined but never used.

 

I ran the client again using gdb. It crashed after a while and this is what i get:

...

Any clues of what's happening?

Well, the main problem is this: in CVS alutLoadMemoryFromFile is not being called, but rather alutLoadMemoryFromFileImage (assuming you are using NEW_FILE_IO, see sound.c around line 1188). Unfortunately, alutLoadMemoryFromFileImage does not exist on the Mac. When the sound code was being changed over the last few months, the developer assumed that Macs would have all the same OpenAL 1.1 functions as PC and Linux users. Unfortunately, this is not the case. So, the code needs to be adapted for alutLoadMemoryFromFile (which I was able to add in at the end of sound.c) or a alutLoadMemoryFromFileImage needs to be created for Macs. Which would you like to tackle?

 

I was just going about playing when some red splotches popped up over the map. It was only from certain points with certain camera angles, and only if shadows were on. So I've turned shadows off for now.

Try to get a screenshot if possible.

[Enly]

Here is a screenshot, looks like whatever should have been shadows, is red.

 

I had shadow mapping on, and turned that off and that caused it to go away.

 

Edited November 29, 2007 by Enly

[Slavia]

When I try to copy over my map files, it says that I do not not have access because "maps" cannot be modified. Anyone else having this problem?

[0ctane]

When I try to copy over my map files, it says that I do not not have access because "maps" cannot be modified. Anyone else having this problem?

You are trying to edit the files in the disk image. Copy el_osx to your computer, and then you can modify the contents.

[peino]

I'm just expressing frayed nerves atm: Is there any reason beside network traffic that my download manager keeps saying it will take 8-12 hours to download 0ctane's file?

 

I'm on dial-up, but the 70Mb Intel file that didn't work for me only took 4 hours to download. I'm in my second day of just randomly trying different times of day to download this. Should I just keep trying, or is might there be a problem at work here? I just wanted to check with people who understand this stuff way better than I do because it seems so weird.

[Enly]

Its probably just a larger file. The first version Fedora compiled took about 10 min to download. 0ctane's took me about 15 mins. Yeah, those are way shorter time periods, but 0ctane's definately took longer to download.

Edited November 30, 2007 by Enly

[peino]

It's not more than twice as big. No, I think the problem was "traffic" because I finally hit on a time when download manager says it should only take 4.5 hours to download 83.5 MB and though that will keep me up past 3 am, I need to try at least to get it onto my hard drive.

 

EDIT: Nope. Download failed and interrupted. I'll try again some other day. Oh well.

Edited November 30, 2007 by peino

[Fedora]

I did some search about the error i was having and found this

http://www.idevgames.com/forum/archive/ind...hp/t-12051.html

 

I downloaded Outnumbered, taken out the OpenAL framework and put it in /Library/Frameworks (risking to mess up my system?) since there is no OpenAL framework inside el_osx.app. Sound crashes are gone now and everything works fine.

 

I know this is a really dirt hack, but i dont know how to link the client against libraries inside the el_osx.app

 

 

Hope this helps.

 

 

EDIT

 

Spoke too soon. It seems that looping sounds (crowd at sto) have noise when they get repeated, something like hiccuping. After a while it crashes. Crash is prevented if i disable sound during the hiccuping...

Edited November 30, 2007 by Fedora

[kulysoft]

Hey guys!

 

Downloaded 0ctane's universal and now i know that i will be a poor man in the next release However i have a problem (not sure if bug), the mouse pointers doesn't change no matter what i do.

I have a PowerG4, running 10.4.11

 

PS: tried even xslimmer but doesnt helped me.

[Florian]

Spoke too soon. It seems that looping sounds (crowd at sto) have noise when they get repeated, something like hiccuping. After a while it crashes. Crash is prevented if i disable sound during the hiccuping...

I had that a lot during the test phase with the "old" openAL from the 1.4 client.

Solution: use OSX's openAL. Since 10.4 it's new enough.

[Enly]

If i have shadow mapping on, those red shadows come up all over the maps. However for some reason now whenever shadow mapping is off the edges of bodies of water have strange shadows, and the water itself is a dark blue void. If i turn shadow mapping back on, the water and shadows around it are normal, but the red returns.

[Florian]

Switching from windowed to full screen crashed the client:

 

Program received signal EXC_BAD_ACCESS, Could not access memory.

Reason: KERN_PROTECTION_FAILURE at address: 0x00000240

0x17cedc66 in ?? ()