SDL2

@Capehill, Raziel
Just to let you both know: that fix for issue with stretching textures which Capehill find out to be fixed just by replacing "&&" to "||" in the glTexImage() was added by Daniel to the minigl repo, as well as he fix the same issue with glCopyTexImage2D().

@kas1e

Yippieh.

Do you probably know when we can expect a release?

@Raziel
Dunno, but probably (only imho), with Tabor release, as it has many Tabor specific changes

@kas1e

Great to hear, that was a quick correction.

@all
There is new version of Gigalomania over SDL2, plz check if all is ok, so i can upload it to os4depot:
http://kas1e.mikendezign.com/aos4/sdl2/gigalomania.lha

What new:

-- Fixed crash when resuming game from saved state when viewing a design.
-- Add call to SDL_DestroyRenderer and SDL_DestroyWindow, rather than just relying on SDL_Quit (so no VRAM leaks on amigaos4)
-- If playing with nukes disabled, more sectors can now instead build spaceships.

@kas1e

Not played but just tried to iconify/deiconify the game while the audio output was busy by TuneNet, after 2 or 3 attemps i got a crash

-----------

Crash log for task "gigalomania"
Generated by GrimReaper 53.19
Crash occured in module gigalomania at address 0x6BA198E0
Type of crash: DSI (Data Storage Interrupt) exception
Alert number: 0x80000003

Register dump:
GPR (General Purpose Registers):
0: 6BA198D4 327F2730 00000000 00000000 53A7B9C0 020B973C 00000020 000007A4
8: FFFFFFFF 0000001B 00000000 01B4726C 000000C8 39150410 00000000 4484D340
16: 6BAA4650 00000000 46300010 2EA4B020 02280000 02280000 00000000 00000009
24: 5FFF9300 00000002 46444888 00000000 02280000 02280000 2EA4B000 02280000


FPR (Floating Point Registers, NaN = Not a Number):
0: nan 200 35 0
4: 0 1289.8 30.8 0
8: 0 4.5036e+15 1289.8 30.8
12: 1490 66 0 -0
16: 1.32625e-315 0 0 0
20: 0 0 0 1.61895e-319
24: 4.94066e-324 0 2.65249e-315 0
28: 0 1.35808e-312 1.3581e-312 -2.15904e-78

FPSCR (Floating Point Status and Control Register): 0x82004000


SPRs (Special Purpose Registers):
Machine State (msr) : 0x0002F030
Condition (cr) : 0x2E002DA0
Instruction Pointer (ip) : 0x6BA198E0
Xtended Exception (xer) : 0x018233AC
Count (ctr) : 0x00000000
Link (lr) : 0x00000000
DSI Status (dsisr) : 0x01842B34
Data Address (dar) : 0x00000000



680x0 emulated registers:
DATA: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
ADDR: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
FPU0: 0 0 0 0
FPU4: 0 0 0 0



Symbol info:
Instruction pointer 0x6BA198E0 belongs to module "gigalomania" (HUNK/Kickstart)

Stack trace:
module gigalomania at 0x6BA198E0 (section 5 @ 0xA8BC)
module gigalomania at 0x6BA198D4 (section 5 @ 0xA8B0)
module gigalomania at 0x6BA1D730 (section 5 @ 0xE70C)
module gigalomania at 0x6BA2CDB0 (section 5 @ 0x1DD8C)
module gigalomania at 0x6BAA4660 (section 5 @ 0x9563C)
native kernel module newlib.library.kmod+0x000020ac
native kernel module newlib.library.kmod+0x00002d14
native kernel module newlib.library.kmod+0x00002ef0
gigalomania:_start()+0x170 (section 1 @ 0x1920)
native kernel module dos.library.kmod+0x000255c8
native kernel module kernel+0x0003e7b4
native kernel module kernel+0x0003e7fc

PPC disassembly:
6ba198d8: 2f890000 cmpwi cr7,r9,0
6ba198dc: 419e0724 beq- cr7,0x6BA1A000
*6ba198e0: 81290014 lwz r9,20(r9)
6ba198e4: 2f890001 cmpwi cr7,r9,1
6ba198e8: 419e0080 beq- cr7,0x6BA19968

System information:

CPU
Model: AMCC PPC440EP V1.3
CPU speed: 799 MHz
FSB speed: 133 MHz
Extensions:

Machine
Machine name: Sam440EP
Memory: 1048576 KB
Extensions: bus.pci

@samo79
Why i not surprised :) If you want to test something give it to samo79 : on his machine everything crashes always and always no reproducable test-case ?:)

Tried to reproduce it on x5k here,and can't :( Running tunenet to play some audio for a while, and then iconify/deiconfigy about 30 times, without crash.

@kas1e

haha

Try this step by step:

1 - Open TuneNet and play somethings ... maybe a streming radio
2 - Then start the game, then go into any area that play an audio (intro is ok too)
3 - Try to iconify/deiconify the game

If not reproducible at first attempt, try again 2 or 3 times ..

@samo
Still can't.. It can be easyly reproducable on your one as maybe it have less ram (both and gfx and system one)..

Maybe somehow tunenet settings related somehow ?

Also, are you sure there is needs for iconify/deiconify ? I mean, when you iconify/deiconify game, it "pause" (you can heard "put on a hold" voice). So maybe you need to try just pressing "p" many times.

Is it crashes when you deiconify , or when you iconify ?

@samo79
I got a crash one time now, when tunenet runs, i run game, and while game loaded, i move volume slider of tunenet => crash. Kind of strange, but so amiga when you do something in one programm, but another one crashes because of it :))

@samo79
Dunno what to say, its very strange, and happens only when tunenet plays.. Maybe tunent trash some memory ? Are you able to reproduce it without tunenet running ?

@Capehill
Btw, in your quake3 SDL2 port , did you add any other specific changes to the rendering part (or maybe blitting or something?) which can cause that effect:



? I just trying to understand from where start to find out the roots of that distortion. With minigl all is ok through, it is with gl4es (while, quake3 which i port from another repo which use SDL1 , works fine and over minigl and over gl4es).

@Capehill
Is it only file which is related to how create context and handle all the stuff for opengl are SDL_os4opengl.c and SDL_os4openglwrapper.c ? I mean, can it be somewhere else in SDL2 something, done specially for minigl only, which, in turn, can cause that, when used not minigl, but any other version of opengl ?

ps. In latest versions of SDL2 , i also had to fix in SDL_os4video.c, that part:




I had to comment out check on OS4_IsMiniGL(), as you hardcode there 1.3 version, while GL i used with SDL2 are 2.x. I mean as it now, it is like any version of GL which will be used with SDL2 , mean to be minigl and be 1.3.

And probabaly OS4_SetMiniGLFunctions() name, can be rename to OS4_SetOpenGLFunctions() , so it will be logical when one will want to use in SDL2 something else instead of MiniGL

@kas1e

Yeah it was exactly that volume, still somethings broken with TuneNet volume slider or similar, for some reason sometimes it lead the system unstable

@samo79

If we will find some reproducable test case with tunenet's volume slider, we can made a proper bug report. If we can't find reproducable test case, it will be not fixed then :(

@Raziel

I create bug-report about that "white dot where real cursor stays", there :
https://github.com/AmigaPorts/SDL/issues/40

If you can add something there , plz do.

@kas1e

Done

@kas1e

Quote:
Looking at the TuneNet source code I can see that the volume slider is an instance of sliderbar.gadget (i.e. not the standard OS slider.gadget). So the reasonable first step to take is to determine whether the problem is actually caused by TuneNet, or by the sliderbar.gadget class.

Ask Andy to replace the sliderbar by the original slider gadget (the code is still there, commented out) and if the problem disappears, then we know we need to have a close look at the sliderbar.

@trixie

It would be also nice to know whether Gigalomania crash happened in SDL2 code or application code. Seems binary was stripped.

@Capehill
Yeah, i stripped it for release, as without it can be 2-3 times bigger. But can build debug version for tests of course.

Just if samo79 says,that tunenet's new slider sometime cause "unstability of system" just by itself (wihotut anything else running), then.. But will build debug version of gigalomania anyway.

@Capehill
Checked very latest SDL2, and can confirm that issue with "white dot" are fixed. Also all compiles fine over public SDK now as well. Thanks !

@Capehill

Great to see progress in various directions, hope till next release you will have time to implement feature request on ticket #27