Tactics Ogre: Let Us Cling Together - ULUS10565

[sunkaifeng00]

Words goes garbled in r2353……but fps increases Obviously

[Rune]

Still experiencing problems with the garbled text. Using ati card and both 2351 and 2353

[gid15]

Still experiencing problems with the garbled text. Using ati card and both 2351 and 2353

My fix in r2351 was not correct (was missing a code review from Orphis ). Could you try again on r2354?

Thanks!

[digiadventures]

I tested it on nvidia and text is working as it should ( always have been )
I think more important then text is crashing,since game is not playable.
Will Drake's fix for crashing be included ?
It not,why ? Would that brake other games ?

[gid15]

Will Drake's fix for crashing be included ?

The correct behavior of sceKernelExtendThreadStack is still unknown. It needs more investigation...

[digiadventures]

Will Drake's fix for crashing be included ?

The correct behavior of sceKernelExtendThreadStack is still unknown. It needs more investigation...

I am really a newbie to emulation,but could I apply Dreke's fix myself ? I guess I would need source code,not compiled version of emulator ? Where can I get this ? And how do I compile it ???
Any tip would be very welcome,I am sure there are other users who read this and would try it so then can play the game,before sceKernelExtendThreadStack is properly investigated and fixed

[Rune]

Still experiencing problems with the garbled text. Using ati card and both 2351 and 2353

My fix in r2351 was not correct (was missing a code review from Orphis ). Could you try again on r2354?

Thanks!

Dude, I love you

The text is 100% now!

EDIT: I will play the psx version while you guys find out the problem wth the crash in almorica castle. I must say I am experiencing the same problems (like drakko) with the sound cutting off

[sunkaifeng00]

Unluckily,ULJM05753（JPN）still has garbled text in r2356,nvidia
PS:it's not the same as Rune's problem.The japnese words in this game isn’t breaking,just show the wrong ones.

[hyakki]

I am really a newbie to emulation,but could I apply Dreke's fix myself ? I guess I would need source code,not compiled version of emulator ? Where can I get this ? And how do I compile it ???
Any tip would be very welcome,I am sure there are other users who read this and would try it so then can play the game,before sceKernelExtendThreadStack is properly investigated and fixed

you can compile it with ant , netbeans, eclipse ..etc
Removing hleKernelExitDeleteThread just gets past the initial black screen the game will freeze up shortly after, shows a few more dialog text then right when the battle starts the AI won't do anything, so right now its not anywhere near a complete fix.
   

[Drake]

My only reasoning for commenting out those lines was that it appears as though the game expects the thread to continue execution after sceKernelExtendThreadStack was called (there are instructions after the call that wouldn't be reachable otherwise). But it also could have been the case that the calling thread wasn't being exited correctly, or there was something else missing in sceKernelExtendThreadStack. Sorry, I got a little too excited since the game was actually going into battles.

@hyakki
I don't believe the AI bug is related since I can start a new game and get past the castle battle (while loading from a save never seems to work). I've gotten to the battle where you have to save the exorcist. The bug also seems to manifest when the emulator is running slowly, and in other battles.

Btw, a developer might want to check out contextSwitch(): the call to executePendingCallbacks() doesn't appear to do anything (you might also want to check for null threads), and the call to checkThreadCallbacks() will only add callbacks to pending, since they are called from the thread doing the switch, I believe.

Also, in sceDisplay.java, resizedTexFb appears to be using the wrong texture coordinates when it is being drawn to the opengl framebuffer. It uses the coordinates for texFb, which is determined relative to the PSP's display area, when it should be using coordinates relative to the resized dimensions and the dimensions of the texture itself.