Tux

Yes it's probably something easy to fix, I'll check it later...

Already ? Well yeah, it's mainly to finish this branch for good ! So : - Cabal (cabalbl) is finally fixed, graphics and sound. The kind of game which seems easy to emulate and which is tricky. This one is quite different in raine, we don't have at all the same way to handle the samples (but thanks for finding the crazy bitswap for the sound command !). - neogeo/neocd saves didn't contain the 68000 bank, it's fixed (problem hard to locate though !). - a last fix for the sound associations, the one included in the test build earlier. http://raine.1emulation.com/download/latest.html

which is the default in the dipswitches, yeah. Ok I think I'll fix it... (it's quite fun, the broken text layer is broken for 2 reasons : it misses half of its bits (!), and its transparent color is 3 instead of the usual 0 ! It's going to require some very special treatment, I can spend a little time on it... !

Might be a way to handle this by the joystick identification, the one which is displayed in the inputs dialog. The idea is save with the controls which joystick is where (what is its number), and remap if the number has changed when the game is loaded. I'll take some more time to think about it, there might be some cases where it's better to disable automatic inputs saving if a joystick is remapped...

Not something easy to reproduce, it's just luck but I think I found the culprit, some 68k bank, which is extremely rare, and that I forgot to save in the save state... ! To my defense there is an insane number of things to save for this driver, it used already the 16 user slots, + a few extensions, it's the only one to save so many things. Actually I couldn't trigger the bug 1st and when I was about to give up it suddenly happened, next time was at the 2nd time, but then it became extremely hard to do again. This bank changes a lot actually in this area so it's not easy for sure. It's only 1 byte to save, assuming I found the real culprit ! If you want to test this again in a next build you'll have to recreate a new save with the next version, this save here doesn't have the bank info.

Not sure about this, I don't use any fancy inputs here, I just have an old ps3 gamepad which works well but with a special driver in windows though. I guess the inputs would be better with sdl2, but there is a ton of unwanted changes to make to have this, so it's not going to happen soon... Normally you should be able to solve this by switching to custom inputs then save the inputs, say to directinput and x360 and then load the inputs you need when you need them. Sorry for the inconvenience otherwise, can't think of anything else for now. See the assocation thread, it's updated !

Ok, obliged me to cleanup most of the commands for type 2, a lot of them are just ignored in fact. You get your own build (link removed), 64 bits since that's what you use. Normally it works everywhere, retested your 3 save states with it, + kasumi selection. Also : I have a way to emulate the playing speed change, but do you know of any place in a game where the music changes its speed to test this ? It can be slowed down or accelerated, both are possible...

probably, but it takes time... This one worked at a time I think no ? Oh well, I'll have a look then I guess... ! After a quick look, it's a super old driver that nobody touched in ages and it shows ! The text graphics had been badly dumped at the time and so it was using only 2bpp (should have used 4bpp !). I think the game worked better in 8bpp in the very old dos versions though... Anyway everything or almost should be redone in there, automatic rom loading instead of the painful decoding there, using graphic layers, and using the roms of some good dump... ! I never liked this kind of game so I never took the time to look into it. To fix these chars alone, should at least use a new rom for the chars, which will make the old romset incompatible. In short, it's a mess !

I hadn't planned to release this one, at least not now, but heh, I had an interesting fix for mer-curious and he has waited quite long for it, so... anyway here are a few of the changes : - sound associations fixes for kof98/99/garou and aof3, see mer-curious post - big toaplan2 update finally. Changes for the romsets : batridera becomes batriderja (mame name), added batsuguna (which is the old batsugun we had before if I remember correctly) and batsugunsp (which is really a special version and not a stupid clone), bgarreganv which was supposed to be a new version of battle garegga, but good luck to find what's changed inside, also bgareggat2 which is supposed to be "type 2", same comment, and bbakraidja which was the old bbakraid. kgpe becomes kingdmgp, same game but different crc for the rom and official mame name, comes with cheats so it's better. sstriker had been mistakenly removed from the list of drivers so it's back, with a new clone, sstrikerk. Also batrider and bbakraid gain the biggest region switch ever seen in raine with 25 countries inside ! Not 25 languages though of course, it's japenese or english for the language, but it's still impressive ! Also there are some new fixes for batsugun inside to handle the graphical glitches, it should be ok until the end of stage 3 if I remember wel, the big boss here is still too black but it's much better than before anyway ! - peekaboo didn't run in 64 bits, specific problem to the emulation there, it's fixed but there are still emulation problems. - added new clone tetristh which is like tetrist but on some different taito hardware (taito h), but it's exactly the same game though, it's more a curiosity. - new cheats for the new clones, kingdmgp, tetristh, batsugunsp, sstriker (it's not a clone but it was gone !). - you couldn't disable the shaders without quiting in this 0.91 version, fixed now (at least in linux, but probably in windows too). - fixes again region switch for lots of games which got broken when changing something else lately. The list of affected games is big, found out for syvalion. - avoid deleting a rom when downloading with curl : if trying to get a clone which doesn't exist on the server but it is on the disk without its parent, then the clone was deleted ! Now it's not at least ! - makaiden didn't load its correct gfx rom, so it looked exactly like its parent (very old bug since 2010 !). - And some big source changes to make rom structs more readable and a little smaller, it should have been done long ago, but there is always something else to do... ! Yeah I know it makes a big list of changes just for 0.91.11, I wanted to release this later as a 0.92 with more changes inside, but you'll get this now instead ! See the download link at the top of the page (thanks Rob !). In case you miss it : http://raine.1emulation.com/download/latest.html

Yeah ok it's your day of luck, I had a look at this one too. Type 2, more complex sound system, there is even a sound command to change the tempo of the music which is not emulated because we can't change the playing speed of an mp3 ! Anyway the common problem here is the interpretation of a command which takes an argument, if the argument is < 0x20 then it's not an argument, it's another command. I am less confident here than for kof99, but it seems to fix to 3 save states, so we'll give it a try... Ok, I'll try to make a binary then... !

wow, you probably spent some time creating this, a save state stuck in pause, have to press select 1 to get out of pause, not the usual thing. Well I had a quick look and it's indeed something unknown, a code which should start a song and which does nothing. I might take a closer look later, but don't hold your breath. Been fixing a few things in raine lately, but nothing related to sound associations ! edit : fixed, it also affects kof98 and garou. The story : until now I was only testing a 1st array in rom which gave the type of command, 2 was for music. Except that after that, it takes an offset to the music itself, and sometimes the offset is 0 like here, in this case the music command is just ignored. So now I am also testing this array of offsets. Now you could wonder why the main cpu is asking a song which doesn't exist ? Maybe it was something planned at the start and finally they gave up the idea ? Since there are quite a few songs like this and not only one, maybe it was some custom stage music like something faster playing if the hp of one of the players reaches a certain limit, and in the end it didn't sound so good so they just gave up the idea, the call remains but it does nothing. Just a guess, but it could be the reason ! What was your other sound assocation problem ? Might be fixed at the same time by this... !

To be more precise, you actually loose a lot of things if doing that, assuming you succeed : the gui is linked to the graphics engine used by raine, so you loose the opengl driver, and yuv overlays, you keep only the normal blits, what made the dos days. Also even if I admit the old gui was nicer, the new one has a lot of advantages : - it's usable with a joystick only (or a gamepad) a keyboard or of course a mouse, for the old one you mainly needed a mouse although you could succeed to use it with a keyboard by tracking the keyboard shortcuts, not easy but possible, but impossible to use with a joystick only - it's incredibly easier to program, that was my main concern while making it, I wanted something where it's easy to add new dialogs and to update them. There are no coordinates in the new one making everything a lot easier. - it adapts to any resolution using the ttf fonts. The old one was made for a low resolution, but only for a low resolution, if using a high resolution it becomes really small which can create readability problems Also you of course loose all the new dialogs which were created for the new gui only, all these options for negeo/neocd and so on, you loose sdl_sound, so no music for the neocd games. That's about it, but that's quite huge !

Oh, so you would want a windows binary but using the old allegro gui ?!! Well, there is a define in the makefile for SDL, there is SDL = 1 in each section (mingw32, linux...), so comment the one for you for mingw32 and it *should* try to use the allegro gui. In this case you should also comment out HAS_CONSOLE, because it can't build the console in allegro, which means you will loose all the modern cheats, you'll need the old cheat.cfg file. If you use allegro you should use allegro 4.x, allegro is famous to be incompatible between its versions and I strongly doubt you could compile a working version of raine with allegro 5, never tried it, but I am sure it's full of incompatibilities. (and I don't even want to hear about bugs in allegro 5 !) Good luck, you can post about your success here ! Notice that you can do this integer scaling by hand in raine by editing raine.cfg (raine32_sdl.cfg for the windows sdl version) and setting yourself the screen_x and screen_y, that's if you want a precise scaling while still using opengl. Otherwise you can use the "normal blits", then in renderer option choose a scaler (hq2x/3x or scale2x/3x and then in set video mode change to either game resolution, or 2x game resolution. I didn't put 3x here, that's mainly because it's rare in most configs to have a mode matching 3x the game resolution, but it could easily be added. After testing this myself, I find that modern video cards have such good scalers that it's mostly useless to stick to integer scaling for most games at least. That's my taste at least, but of course it should theoretically be better if using integer scaling, even if it's not obvious usually ! (especially if combined with a shader which improves the image quality)

There were quite a few dos binaries released since then. Of course you can compile one yourself if you want, you just need to be patient because afaik djgpp was never integrated in any visual something from microsoft (surprise, surprise !), so you need to get it + all the libs, there are less libs for dos than for linux & windows of course, it does less things too. I cross compile it from linux, there are a few howtos on the web on how to crosscompile for dos from linux, and there is probably something for windows too... I am not sure the famous 3x modes were released as a binary so I can make a new binary for that if you want, I had planned to test this first on a real hardware, but since the old pc here has lost its hard disk, it's much harder than what I had anticipated and I had to give up on this. At least it works when run in a virtual machine ! Nice custom video modes by the way, it's a windows driver which allows to define some modes like this one, or you did it differently ? I did that too in linux back in the day, but since 0.50 it's more about stretching the screens in opengl, it's not the same, but it's not bad too (and you get the shaders too !).

Sorry went too fast to make the binaries, forgot to update the version number for the 64 bits, but if you have the 0.91.10 archive with the exe inside dated october 26th, then you have the right one. I can't tell you about hawk and animal because I don't know the game, but testing with hulk hogan as player 1 (I can recognize him at least !), and animal as player 2 works. I'll try to re-upload the 64 bits exe later, but as I said, it's the right one, just a problem with the version number. (with hh and animal the cheats produce this : poke 1c0607, 0 poke 1c0713, 5 where before they produced poke 1c0607, 5 poke 1c0713, 5 So for me the problem is fixed, now having these chars in the selection is still a hack and so I can't be sure everything works as expected). edit: reuploaded, I didn't bother to change the filename of the archive.

Yes it means the driver has a problem, rubyTexture is the base of this shader, it contains the game bitmap, it's sent for all the frames. I don't get any warning about rubyTexture with my nvidia, so there is definitely something wrong with the way the shader is interpreted by your video driver. 2 choices : just ignore the shaders, you can live without them after all, or find a better video driver... ! And the reason you don't see the message on screen is because it's said it's a warning, the validation succeeds in the end, but if it misses rubyTexture, it misses the main point, the crash could very well be a divide by 0 at this point !

Well it's worth trying a driver update, it shouldn't crash normally even if the video card can't handle it if the driver is well done it just returns an error message which is displayed by raine. 2nd solution : it just doesn't work, doing nothing noticeable on screen. 3rd : yours, never seen before, it crashes -> something is probably wrong somewhere, try to update your driver.

The mails are not rejected, they go to spam. It's very hard to make them accepted nowdays yeah, but the bright side is that almost no spam goes through. Ok don't worry too much, he didn't describe very well what happened, and didn't come back so you can't do too much about it, but thanks for trying !

That's a good one... You forgot the video card ? I never saw shaders crashing the program so far, usually it works or it doesn't that's all, it's a 1st time here. Well it works here of course, but it's very likely something related to your video card. I'll check later in windows, but since it works in linux here it should also in windows. Also you can try to run the emu from the command line and redirect the output, like this : raine > log when it has closed because of your shader check the log file to see if there is something interesting in it, at the end. edit : yeah no problem for me in windows fullscreen with the crt shader, as expected.

Finally I can't test this on real hardware for now, the old computer doesn't have any hard disk anymore and it seems I don't have any working ata hard disk anymore neither ! So game over here, oh well it should work at least, even if it's useful only for these very wide games.

And the linux binaries are updated, I had forgotten that I had done a nice little script which does everything for me for that, you just launch it from the 32 bits or 64 bits directory, give it the version number you want, and it does all the rest (get the sources from git, compile, update the sha256sum and the version number in the PKGBUILD, upload the binary package and the PKGBUILD file, and finally update the html page !), so it's quite fast to do it finally !

No problem, I understood you were pissed by the technical pb in the forum, it's ok !

ok 0.91.10 released with these fixes for windows, I'll try a linux binary tomorrow

Ah I found the forum was very quiet lately, so it was because of a technical error ?!!! Sorry to learn that... Oh well no wonder there are still things to fix about all these shiny new cheats, too much to test, I'll have a look, but I don't know when, it shouldn't take 10 days though ! ok bug confirmed, I hadn't thought about multiple scripts using a choice from a list like that running at the same time, I'll have to find something, it will take a little time... edit again : it's fixed, I'll have to make a new binary for that, and maybe a linux binary too, the biggest change since 0.91.9 except this one is for linux to be able to run its asm code with the recent change which makes the stack not executable by default. Oh well, not an ideal setup to make a binary right now, it will wait until tomorrow. Thanks for the bug report at least !

Yeah I plan to do that too when I get access again to an old 1Ghz pc I have, but for now it's quite far, it will be later. I plan to test with univbe too to check if there are some modes where triple buffer can be enabled, for now the only mode which can work for sure is mode-x 256x240, I am almost sure it's possible in vesa, but maybe univbe is required and I can't test this in an emulator...