Closed Vweber73 closed 10 months ago
sonninnos
commented
10 months ago You can't run stuff from whdload.de because those LHAs don't have game data in them. Look at the file size. They are just installers. Only pre-installed Retroplay LHAs can be launched, as stated in the documents.
Vweber73
commented
10 months ago Oops, sorry! Many thanks
Vweber73
commented
10 months ago Sorry again, but I tried a proper retroplay lha this time, but I got this. Seems the proper kickstart is not renamed properly?
sonninnos
commented
10 months ago Never had that happen. If the core can find the kickstart, so can WHDLoad. The WHDLoad script will copy all required Kickstarts from system on every run.
Also the legacy HDF mode does not support differently named Kickstarts than described in the error, and it shouldn't be used anyway. So just let the core options be the defaults.
Vweber73
commented
10 months ago Thanks, but I don't know what to do then. I am using all core defaults, A1200 preset, but my A500 1.2 and 1.3 KS work perfectlly. The screenshot I posted is using the "Files" option for WHDLload, not "HDF'". Trying HDF gives me the same thing, plus additional error messages in th window, complaining about the KS file size...
Vweber73
commented
10 months ago Here is my files in the system folder... They all work well under PUE...
sonninnos
commented
10 months ago And the core is the latest version with this fix included? https://github.com/libretro/libretro-uae/issues/625
I really can't guess anything more without a log of the WHDLoad first run after deleting saves/WHDLoad.
Vweber73
commented
10 months ago Wow, this issue looks really like mine! I always download the latest version of the core. I check several times a day. I'm definitely up to date.
I deleted both whdload files in Saves, started again, here is the log file. Complaining about size of the KS files...
WHDLoad QuitKey triggered..
Model: 'A1200'
Kickstart: 'kick40068.A1200'
WHDLoad not installing Kickstart 'amiga-os-120.rom' due to incorrect size, 524299 != 262144
WHDLoad not installing Kickstart 'amiga-os-130.rom' due to incorrect size, 524299 != 262144
HD (1) inserted in drive DH0: '/storage/emulated/0/Download/AdventuresOfWillyBeamish_v1.2_1814.lha'
SNDRATE 313.0227.049.920410=3546895.000000
PAL mode V=49.9204Hz H=15625.0879Hz (227x312+1) IDX=10 (PAL) D=0 RTG=0/0
KS ver = 40 (0x28)
KS ver = 40 (0x28)
RTAREA located at 00F00000
TRAP_ENTRY = 00f021de
Known ROM 'KS ROM v3.1 (A1200)' loaded
CPU cycleunit: 128 (0.250)
1872 CPU functions
CPU cycleunit: 128 (0.250)
Building CPU, 45912 opcodes (2 -2 1)
CPU=68020, FPU=0, MMU=0, JIT=0. ~cycle-exact 24-bit
Reset at 00F8F7EC. Chipset mask = 00000007
PAL mode V=50.0804Hz H=15625.0879Hz (227x312+0) IDX=10 (PAL) D=0 RTG=0/0
Autoconfig board list:
Card 01:
'
sonninnos
commented
10 months ago Well how on the earth can those old 256KB Kickstarts be 512KB there..?
Vweber73
commented
10 months ago No idea. These are my cloanto files...
I downloaded several others. Every time, now, I get this....
Vweber73
commented
10 months ago If this helps, here is what is said about cloanto roms on Android:
Amiga "Kickstart ROMs" for Android
On Android systems, Amiga Forever Essentials for Android installs official Amiga ROMs and some additional disk images in a way that is autodetected by Amiga emulator apps for Android (e.g. UAE4Droid, AnUAE4All, UAE4All2, Omega 500). These are the 512 KB system ROM files (256 KB ROMs won't work on many mobile emulators) to run Amiga OS versions 1.2, 1.3, 2.04, 3.0 and 3.1 in emulated computers like the Amiga 500 and Amiga 1200. So if you are searching for "Amiga 500 Kickstart 1.3" or "kick.rom" or "kick13.rom", or for a ROM with MD5 hash 192d6d950d0ed3df8040b788502831c2, Amiga Forever Essentials for Android is the answer that is both legal and easy to use, as it requires no manual configuration work.
sonninnos
commented
10 months ago Makes absolutely no sense. How many Cloanto ROM versions are there? The user in the above issue provided crcs and file sizes for all, and they only had the 11 byte encryption extra.
I would expect that kind of error message precisely with entirely wrong files, which will eventually result if I'd skip the filesize check. And how can they possibly work with UAE either without warning messages. As in are you 100% sure that those double sized versions work when you boot normal A500 with either 1.2 or 1.3, and the log says
[libretro INFO] Known ROM 'KS ROM v1.3 (A500,A1000,A2000)' loaded
Edit: Uhh how can platform have anything to do with actual Kickstarts being used..
Vweber73
commented
10 months ago Sorry, maybe I, wasn't clear, when I said I tried other roms, i didn't mean I tried other Cloanto roms, just that I tried the standard 256k roms that I could find on the net. With those, I get no more size errors, but validity errors as my screenshot shows. I don't know what do you if no kickstart is OK!
I agree with you that platform is irrelevant to kickstart. Very strange!
As requested, I put again the Cloanto 512K 1.3 rom as Amiga-os-130.rom and relaunched the core. It works fine, and here is the end of the log file:
00000000 2048K/4 = 512K ID F32 Kickstart ROM (543DA404)
00200000 8704K/0 = 8704K -- F32
4F694E4E 25F2 25F2 25F2 25F2 25F2 25F2 25F2 25F2
4F694E5E 25F2 25F2 25F2 25F2 25F2 25F2 25F2 25F2
4F694E6E 25F2 25F2 25F2 25F2 25F2 25F2 25F2 25F2
4F694E7E 25F2 25F2 25F2 25F2 25F2 25F2 25F2 25F2
4F694E3E 25F2 25F2 25F2 25F2 25F2 25F2 25F2 25F2
4F694E4E 25F2 25F2 25F2 25F2 25F2 25F2 25F2 25F2
4F694E5E 25F2 25F2 25F2 25F2 25F2 25F2 25F2 25F2
4F694E6E 25F2 25F2 25F2 25F2 25F2 25F2 25F2 25F2
4F694E7E 25F2 25F2 25F2 25F2 25F2 25F2 25F2 25F2
00000000 2048K/4 = 512K ID F32 Kickstart ROM (543DA404)
00200000 8704K/0 = 8704K -- F32
sonninnos
commented
10 months ago And that KS ROM checksum matches with the documents? Why did you cut out the part where it says something about recognizing it?
That log seems like the opposite of "It works fine", since it says "Your Amiga program just did something terribly stupid". Those errors are not at all normal.
Do you have the rom.key there with the files or not?
Vweber73
commented
10 months ago Sorry, I think i messed up with the log, it was not logging anymore when I reverted to Cloanto. Here is the log file, it seems that it thinks it is an Amiga 3000 1.3 rom and accepts it as such... Yes, the rom key is there.
Model: 'A500' Kickstart: 'amiga-os-130.rom' SNDRATE 313.0227.049.920410=3546895.000000 PAL mode V=49.9204Hz H=15625.0879Hz (227x312+1) IDX=10 (PAL) D=0 RTG=0/0 Wait for blitter is not available in 68000/68010 cycle exact modes. KS ver = 34 (0x22) KS ver = 34 (0x22) TRAP_ENTRY = 00f021de Known ROM 'KS ROM v1.3 (A3000)(SK)' loaded CPU cycleunit: 256 (0.500) 1872 CPU functions CPU cycleunit: 256 (0.500) Building CPU, 44274 opcodes (0 -2 1) CPU=68000, FPU=0, MMU=0, JIT=0. prefetch and cycle-exact 24-bit Reset at 00FC6D52. Chipset mask = 00000000 PAL mode V=50.0804Hz H=15625.0879Hz (227x312+0) IDX=10 (PAL) D=0 RTG=0/0 Autoconfig board list: END CPU cycleunit: 256 (0.500) hardreset, memory cleared SNDRATE 312.0227.050.080410=3546895.000000 [Environ]: SET_SYSTEM_AV_INFO: 720x288, Aspect: 1.250, FPS: 50.08, Sample rate: 44100.00 Hz. [Audio]: Set audio input rate to: 44100.00 Hz. [Video]: Timings deviate too much. Will not adjust. (Target = 60.00 Hz, Game = 50.08 Hz) [OpenSL]: Requested audio latency: 128 ms. [OpenSL]: Setting audio latency: Block size = 768, Blocks = 32, Total = 24576 ... [Display]: Found display driver: "gl". [Input]: Game Focus => OFF SNDRATE 312.0227.050.080410=3546895.000000 PAL mode V=50.0804Hz H=15625.0879Hz (227x312+0) IDX=10 (PAL) D=0 RTG=0/0 Illegal instruction: 4e7b at 00FC0564 -> 00FC0582 SERIAL: period=372, baud=9600, hsyncs=14, bits=8, PC=fc2234 SNDRATE 313.0227.049.920410=3546895.000000 PAL mode V=49.9204Hz H=15625.0879Hz (227x312+1) IDX=10 (PAL) D=0 RTG=0/0 [Environ]: SET_SYSTEM_AV_INFO: 720x288, Aspect: 1.250, FPS: 49.92, Sample rate: 44100.00 Hz. [Audio]: Set audio input rate to: 44100.00 Hz. [Video]: Timings deviate too much. Will not adjust. (Target = 60.00 Hz, Game = 49.92 Hz) [OpenSL]: Requested audio latency: 128 ms. [OpenSL]: Setting audio latency: Block size = 768, Blocks = 32, Total = 24576 ... [Display]: Found display driver: "gl". [Input]: Game Focus => OFF
sonninnos
commented
10 months ago Well that is weird. UAE code says the size for that is 256KB:
{ _T("KS ROM v1.3 (A3000)(SK)"), 1, 3, 34, 5, _T("A3000\0"), 262144, 32, 0, 0, ROMTYPE_KICK, 0, 0, NULL,
0xe0f37258, 0xC39BD909,0x4D4E5F4E,0x28C1411F,0x30869504,0x06062E87 },No wonder WHDLoad won't like that kind of version either. In any case best to just get the matching Kickstarts to system, delete the poisoned WHDLoad dir from saves, and enjoy.
Vweber73
commented
10 months ago Thanks! Working know. Your key advice was "delete the poisoned Whdload Dir from saves". I didn't do it at first, so even 256KB roms wouldn't work, which drove me nuts!
I just read that the 512Kb roms are just 2 identical 256Kb roms glued together. It is the way than transom works. So you could support them I think by just something like if (rom13.size == 512) just_take_half_of_it() ; 🙂
sonninnos
commented
10 months ago Yep, the copier won't overwrite files, so first one stays, wrong or not.
So weird that there is anything to copy together with that ROM, so what is the other half other than zero? I can understand combining CD32 KS+extended to one file, but this is alien.
Vweber73
commented
10 months ago Yes, this is nuts, but this what it is. Thanks again!
One little thing, now that I can play 10-disk games without getting crazy. These games have menus, triggered by the ESC key. So I press Select to activate the VKBD, then press ESC, the menu appears, I want to see it well so I press Select again to remove the VKBD, it does that, but it also triggers an Amiga mouse left click, so that the menu entry the mouse pointer is on is selected, against my will. Most irritating. Any way to change this behaviour?
sonninnos
commented
10 months ago Right, I haven't really used it on touchscreen so wasn't aware of it. So basically because touchscreen acts both as a pointer and a mouse, the frontend sends both down at the same. At least the frontend no longer sends RetroPad events and touch events at the same time, which a while ago prevented closing VKBD completely if the overlay Select happened to overlap with the core image.
I can try simulating that behavior somehow and ignore mouse clicks during VKBD toggle.
Vweber73
commented
10 months ago That would be great, many thanks!
sonninnos
commented
10 months ago I can't replicate the behavior by using mouse to touch input overlay Select while having VKBD open and/or over.
How old RetroArch are you running again? This reminded me of a thing that is fixed already: https://github.com/libretro/RetroArch/pull/15346
Vweber73
commented
10 months ago Current stable, 1.15.0
sonninnos
commented
10 months ago Update to nightly then.
Vweber73
commented
10 months ago I did so. This gave me another instance (1.16.0) of Retroarch, which I had to reconfigure. Unfortunately It doesn't change the behaviour: in rise of the dragon, select/esc/select still triggers the mouse event...?
Vweber73
commented
10 months ago I tested more and realize that there is a behaviour difference between version 15 and 16. Both versions trigger a left click mouse event upon pressing Select, when in the main menu of rise of the dragon, opened with ESC. But during the very first menu (skip introduction / play introduction), v15 triggers the click and V16 does not!! Very strange...
Vweber73
commented
10 months ago More differences (Rise of the dragon lha, still) :
In both versions, If the VKBD is on before the Rise of the dragon menu appears, the mouse pointer is on Play, pressing Select again triggers the click.
However, if you trigger the menu, move the mouse pointer away (using the mouse keys from the VKBD), then press select, the menu is still there. If now you set the mouse pointer to play (without clicking), on V15 pressing Select triggers the click, but not on V16, which behaves correctly in this situation, but not on the initial situation! Stranger and stranger...
Vweber73
commented
10 months ago Hi, Any idea why this inconsistant behaviour? Anything that could be fixed are core level rather than Retroarch level? Many thanks in advance, best regards
sonninnos
commented
10 months ago I can't reproduce the issue with my phone. By having the statusbar open there is no mouse clicks going on when toggling the VKBD by pressing Select on the input overlay. Only when hitting some empty area.
Vweber73
commented
10 months ago Have you try rise of the dragon?
I'm talking about this menu. Select does not trigger an event if the menu is already there prior to VKBD, but it the VKBD was on screen before the menu was called (by ESC) then the event is triggered...
sonninnos
commented
10 months ago I can't see how any game could get any input if the input is not shown in the statusbar, so I did not test the game.
sonninnos
commented
10 months ago Hehey, I think I got it. The ghost click happens only if any VKBD key is pressed after toggling. Too bad it still can't be simulated in desktop..
Vweber73
commented
10 months ago I don't get it. The game is getting virtual mouse input, the empy space on the screen acts as a touch mouse. When I touch it the J2 changes to m2 with arrows... When I press on the screen it acts as a left click... This also happens in V15 whenever I press select, and in V16 only if VKBD was there before the menu.
Vweber73
commented
10 months ago Hehey, I think I got it. The ghost click happens only if any VKBD key is pressed after toggling.
Yes that's it! :)
sonninnos
commented
10 months ago It definitely is the Android quick tap press, since it won't happen if the press is longer. Don't even have to touch the VKBD, but just any empty area while the pointer input (VKBD) is active.
Vweber73
commented
10 months ago Well, I am looking at the Select behaviour, and it's weird. In V15 the behaviour is wrong (send a left mouse click) but at least systematic. In V16 the bahaviour is right under some circonstances, and wrong under other circonstances. I don't understand the logic...
sonninnos
commented
10 months ago Obviously the logic isn't intentional. The quick tap prevention gets nulled somehow when pointer input is received between overlay presses. So if you just use the overlay buttons for pressing Esc instead of pressing Esc, there is no ghost click afterwards.
Vweber73
commented
10 months ago Yes, confirmed, if you press ESC using the B button instead of ESC on the VKBD, the problem does not occur! Is it a Retroarch general issue or a PUAE Core specific issue? I also reported it on the general issue you pointed me to.
Many thanks and best regards
sonninnos
commented
10 months ago As already stated, the fault is clearly in the Android input driver. It translates touch events to mouse events. The core gets and obeys the input it receives. If the core was at fault per se, I could replicate the behavior with desktop using the same overlay method.
Since I can't test Android changes easily, I can try adding some circumvention to ignore inputs more aggressively when toggling VKBD. It already waits for joystick input to be released, but ignores mouse.
Vweber73
commented
10 months ago Many thanks, great, that would be much appreciated. I think then it would also work on V15, as I prefer to stick with stable versions... Cheers
Vweber73
commented
10 months ago I tried the new core, thanks a lot for trying! But I see no behaviour change, neither in V15 nor in V16...
Vweber73
commented
10 months ago Ouch the last core is a catastrophy! On V15 it takes ages to launch, then there is no on screen elements, no select key,... although the game itself seems to work. On V16 Retroarch completely freezes on launch!!
Vweber73
commented
10 months ago Sorry, don't know what happened, but the cores are now working, on both V15 and V16. However no change compared to previous ones as to the click behaviour...
Vweber73
commented
10 months ago I tried the latest core. Still the same as before... With one small difference. The left click is still registered after Select, but with one second delay. I initially thought the problem was solved... Until the click finally occurred.
sonninnos
commented
10 months ago This is getting pretty unbelieveable and insane.. Every time I increase the amount of frames the core skips ALL input after VKBD is toggled off, the ghost mouse click just keeps on getting triggered later, which makes absolutely no sense.
When the quick tap triggers normally, there is a clear delay sure, but not that long.
Vweber73
commented
10 months ago Ouch ! So what to do ? Cheers
Le mar. 5 sept. 2023 Ă 18:50, sonninnos @.***> a Ă©crit :
This is getting pretty unbelieveable and insane.. Every time I increase the amount of frames the core skips ALL input after VKBD is toggled off, the ghost mouse click just keeps on getting triggered later, which makes absolutely no sense.
When the quick tap triggers normally, there is a clear delay sure, but not that long.
— Reply to this email directly, view it on GitHub https://github.com/libretro/libretro-uae/issues/630#issuecomment-1706969837, or unsubscribe https://github.com/notifications/unsubscribe-auth/AWBK3J6NWUCE7MWBCAIONODXY5J5TANCNFSM6AAAAAA4GE3LTY . You are receiving this because you authored the thread.Message ID: @.***>
Vweber73
commented
10 months ago Wow, we are making progress! With the latest core: -V16 works fully as expected, -V15 works well except when using fast forward mode. Pressing the FF key registers a left click, and pressing Select during FF mode registers too. We are getting there!
sonninnos
commented
10 months ago Ok, I believe we are there then, since old versions are bygones..
Vweber73
commented
10 months ago Well V16 is still not official! :) Many thanks anyway for your fast and quality work Cheers
Vweber73
commented
9 months ago Hi, Just upgraded to V16, which is now official. Everything works great, save for a very minor issue: when you press the up and down triangles to hide/show the Retroarch control, this triggers a mouse left button event. No big deal as you don't do that very often, but still strange... Cheers
Hi, Forgive me is this is trivial, I am new to WHDLoad and would like to give a try. So far no luck. I tried to load both .lha files from whdload.de for Heart of China and Willy Beamish. I used the A1200 preset (3.1, 2Mb/8bMb). I enabled Whdload in the Core. If I choose "files", I get a readme file, then a window with empty "options" parameters, and a choose box with "set", "cancel", "ignore". Whatever I do, I got a recoverable alert, then back to the readme file. I choose "hdf", then in the Workbench window I get repeated error messages about the 600 then 1200 kickstart having a bad signature. I am using Cloanto Amiga Forever Kickstart files, legally purchased on the store, could this be the problem ? I have no problem whatsoever with kickstart files on everything non-WHDLoad...
Thanks in advance and cheers,