Menu-Screen Problem since amiberry v5.7.2 #1401
Comments
|
I just tested Master/debian bookworm/RPi4B -- there seems to be an issue here, but AFAICT it has to do with Fullscreen mode. Could you try something please? Instead of using Fullscreen and setting 800x600, try Full-window (it will default to 800x600 and appear fullscreen), and see if things work properly like that? |
|
As you can see on this unsharp photo the menu screen now appears as a small windows on the workbench. Using RaspiOS 64 Bit.
|
|
Okay, thank you ~ that coincides with my results here (I don't have a RetroPie setup ;) You can use that approach as a work-around perhaps, until this gets looked at? @midwan To recreate: launch amiberry, GUI -> Display -> select Fullscreen (Res. doesn't matter) --> Start //just need kickstart Once kickstart loads, try hitting F12 to pop GUI ...releases mouse, flips and pops emulation window to front again (if you alt-tab at this point, GUI will appear). Same thing on the RPi4 with 5:4 display & I seem to recall you mentioning long ago, the thought of dropping Fullscreen mode altogether had floated to the top once or twice. I tend to agree in a lot of ways, considering Full-window does present itself as fullscreen anyhow, and the fact it pops WM decorations as well as the GUI (with emulation window paused in the background) when F12 is hit, I sort of like =) Food for thought... not sure about the 'wobble' description (might be retro-pie related on top of this issue?) HTH |
|
I can only guess the "wobble" mentioned is regarding the resolution change on the monitor? I'm not sure Fullscreen mode is very useful in this way also, considering the issues it causes. I'll have to think about removing it and only leaving Full-window (which is essentially Borderless window, scaled to full screen), instead. |
|
//...at least this new monitor lets me poke at stuff like this pretty easily ;) Jelly wobbles, it can also be described as 'shakes' ...the latter is more what I see, in that moment of indecision just after hitting F12 ... (as a language, English is very bad for this =). but as you say it might be the display hardware trying to figure out what to sync to....because... ...the actual problem here, is that the GUI window is somehow inexorably linked to the current resolution of the Xdisplay from which it was launched, regardless of what resolution the emulation is started at with ie; Fullscreen/800x600. If in emulation like that and you hit the F12, the GUI window is still in the native Xdisplay resolution...ie; 1280x1024_60 ...so when you hit F12, effectively it's trying to do this.... ...which won't happen without Alt-Tab, Maximize window, because WM (ymmv there) ~ either way you look at it, it's one of the GUI not starting in/switching to Fullscreen mode...OR....when in Fullscreen mode (emulation) and the F12 key is pressed, the GUI should switch to Full-window mode @ the same resolution the emulation is at (not the desktop resolution)....and by the time you get to that point, one has to question if there's any actual usage case, for which Fullscreen mode is absolutely required? I don't know that answer, but if not, having Fullscreen mode is always going to invoke the Xserver to change screenmodes (which is annoying) and not having it.... in essence... fixes more things than it breaks I think (especially if there's no absolute requirement to have any Fullscreen mode) |
This assumption is not plausible to me, as I had no problem with the fullscreen until version 5.7.1. |
|
@Retro1968 -- If as you say you're using |
|
The problem bothers me so much that I have reported it as a bug and will probably fallback to v.5.7.1 until it is fixed. |
Ah right, I missed that part. I'll take a look at the changes between 5.7.1 -> 5.7.2 and see what could cause this, then. |
|
It was the gfx-rewrite, which included the multi-window code from preview. That is, assuming Full-window works fine in these conditions. @Retro1968 can you confirm that is the case, please? |
|
Yeah, I traced it back to 070e193 as well (specifically gui_renderer), but for the life of me I can't find any instance wherein I actually need I did think that (perhaps) the GUI presenting itself somewhat smaller in ...elsewise, I cannot discern a need for Fullscreen mode really. |
I am at the moment very busy an try later. What i do not understand (maybe my english is too bad) that you say the error wont go back. In v5.7.1 everything worked perfect for me. So it must be something you changed form v.5.7.1 to v.5.7.2. problems began with 5.7.2. It should be possible to fix that. :) |
|
@Retro1968 Either we try to find the specific issue with full-screen mode under the scenario you had, or we skip fullscreen altogether and replace it with full-window. The second is easier, which is why I asked if that works OK for you. |
|
I had a deeper look at the goings-ons wrt the WM (XFCE4) ~ what I say above is true ; if you change the window focus settings in XFCE4 to calm it down....this works by setting the WM to give focus to Desktop the moment it appears, before the amiberry GUI window is spawn (in a different screenmode res)... ...in amiberry after hitting F12 results in this... As far as the WM is concerned, the amiberry process has spawned 2 windows (amiberry, amiberry GUI), but it is the case that the 'amiberry' window is Fullscreen 800x600, but the 'amiberry GUI' window is @ 1280x1024. Due to the fact the window focus settings in the WM no longer trigger (the need) for a screen resolution change, the WM itself has inherited the 800x600 resolution (even though by default the desktop res is set to 1280x1024).... ...what I noticed was, when clicking on the amiberry menu-bar icon.... ....the amiberry process itself, is indicating to the WM (bold italics) that the GUI window IS being presented as active ...although clearly, that is not the case ;) Ergo...this then becomes a window promotion thang, m'kay....but if you let the WM do that, it's going to flip screen resolution at the Xserver to do it, and so the amiberry process must promote the GUI window itself...we can do that.... GUI -> Miscellaneous --> GUI Always on top ...with that set, hitting F12 reliably does this... User must remember to use F12/Resume to return to the emulation (not close the window itself =), and it effectively leaves the Xserver in the same screenmode resolution. For mine, that's a pretty gracious work-around here all things considered, but usual caveats apply...ie; RPi4B, debian bookworm, Not sure how this pans out wrt other WMs... but wrt XFCE4 & RaspOS current, this seems to be the case for Fullscreen mode. Perhaps a 'borderless GUI' option (for the GUI window) could be of use here, however if you set the top menu-bar to always auto-hide, when you hit F12 it reliably pops GUI like this... ...that's pretty good IMHO...in fact, isn't that an awesome coincidence the window title bar overlaps the GUI itself in just the right spot...or ...did you see that coming? =) ... so that's 1 part amiberry (GUI Always on top) and 2 parts WM settings (short auto-focus, auto-hide panels) to get it like that ~ works as expected when amiberry is in Ultimately, I don't see a problem here, more a case of changed behavior, as a corollary of the gfx rewrite as it were (didn't we touch upon this at some time...ie; how much was the WM's responsibility, versus how much amiberry had to do instead?)... something like that... or was this a valid case for I'd leave alone =) The HTH |
|
I have now tried amiberry V.5.7.4 2024-08-07. (should i try that version?) I can live with it under RaspiOS 64 bit. Under RetroPi I still have the same problems even with the full Windows mode. F12 can still cause amiberry to crash. When the menu is there, the screen shakes. On both systems I compile with: make -j4 PLATFORM=rpi4-sdl2 |
|
Just tried it with v.6.3.4 preview 2024-08-06 on retropie and got there the same issues as with v.5.7.4 2024-08-07. |
|
I believe you should be compiling with: make -j4 PLATFORM=rpi4-64-sdl2 |
|
I have to correct myself: Certainly I compiled on RaspiOS 64 Bit with: make -j4 PLATFORM=rpi4-64-sdl2 Since RetroPi runs on 32-Bit Linux I compiled there with: make -j4 PLATFORM=rpi4-sdl2 ;-) |
|
This "screen shaking" (retropie) after pressing F12 for menue seems to me to be caused by the menu window and the workbench window "arguing" about which window should be displayed at the front. And that's why one window wants to come to the front and then the other. Also the options under Miscellanious: "Always on Top" an "GUI Always on Top" couldn't fix that. @midwan i guess that you should give the menue window a higher (maybe the highest) priority to stay save always in front as long as the user press F12 to close it. |
|
Thanks for testing ;) Considering that (more or less) we use the same hardware, and that I can't really recreate the shaking or wobbling effect, I'm going to suspect it's being caused by the retro-pie software itself, or the host OS version/libraries in use...ie; I've only tested against debian bookworm, no retro-pie. You'd have to test on that, on your system, using the retro-pie software to launch amiberry, and then compare result against just using amiberry standalone..(launched from an xterm).... ...'wobble' suggests to me the amiberry GUI changing between 720x568 <-> 800x600 constantly ... (so the 'shake' would be a measure of 80x32 pixels)...that's the only way I can visualize 'wobble'. In my stand-alone example above, it's definitely native display resolution (Xserver) being attached to the amiberry GUI (window), in conflict with the requested display resolution used for the emulation process in Fullscreen mode. In theory, if you set your linux desktop resolution to 800x600, then launched amiberry with a config.uae file set for 800x600 Fullscreen, the Xserver shouldn't need to change display resolution, and things will work differently. (not a fix, just to test the situation ;) |
|
I have emulated a whole range of retro systems under RetroPie. Therefore, I will not change any RetroPie system settings. I will be happy to try out what can be set there within Amiberry. However, the problem should be able to be solved if the window for the Amiberry menu is given a higher priority as i mentioned here #1401 (comment). I am pretty sure that @midwan will find a good solution. :) |
|
@Retro1968 ~ I tortured the situation some more, and was able to coax this uncommanded behavior from amiberry, after pressing F12 ... https://www.youtube.com/watch?v=1RCr-TVpyZk ...is that anything like what you're seeing? |
Yes that is nealy the same that happens here. With the different that i use a 4:3 Monitor :) |
Thanks for the video. That seems like an interesting situation. Both windows are open, but SDL2 (under KMSDRM only) seems to be flicking from one to the other then. I think this needs some kind of special handling for KMSDRM, it might even be a bug in SDL2 (not sure). We'll see what we can do. |
|
@Retro1968 ~ thanks for confirming, it took some time testing to be able to find what you were referring to ... the 4:3 displays are in another room connected to my A1200 and voodoo2 PC, so 5:4 had to do (but it does 50Hz modes =) As noted, I had to edit /boot/cmdline.txt to append video=HDMI-A-1:800x600M@60, to get the console into the same mode as the config.uae specified (Fullscreen 800x600), to be able to create the problem. I believe I had 'GUI Always on top' enabled at the time. Doing the same test using the default EDID mode of 1280x1024@60, I had a couple..nay, few instances of both the amiberry GUI and emulator 'disappearing' after hitting F12 to return (Resume) the emulation, only showing the console and no KBD input (mouse was still there but did nothing) ~ had to ssh in to find amiberry consuming 98+% CPU, kill the process, nothing in amiberry.log. This could likely be related...ie; if it's doing this in the vid when console modes are the same, could be it's trying to do the same thing with different modes and ends up in a deadlock? In x11 it's the same but different...as windows don't make a lot of sense on the console ;) I eventually discovered that when SDL creates a window, that also establishes a lock in Xlib, and it's this, not so much the resolution/screenmode difference, that dictates what's going on...ie; if you set desktop display to 800x600@60, it still behaves the same way. You can work around it as per above, using 'GUI Always on top'...and you don't need change the focus settings in the WM, but hiding the panels is worthwhile. I actually believe in x11 this is working properly, it's just that in the x11 case, you must enable the 'GUI Always on top' option, for the F12 key to pop the GUI when using Fullscreen modes. |
midwan
changed the title
|
@giantclambake |
|
@giantclambake |
|
For the record, I cannot recreate the above issue, when using that syntax in general. Does it always trigger for you? Maybe more details are needed. |
|
As the headline says... it started with v5.7.2. |
|
@Retro1968 |
|
Problem was there before the commit, so this is a secondary issue. I can reliably recreate the problem. See attached logfiles... amiberry.log.gz ---> command used You can see in the log that I pressed the F12 key a total of 10 times to toggle between emulation & GUI --> works as expected, quit amiberry. amiberry-bad.log.gz ---> command used You can see in the log that I pressed the F12 key once to pop the GUI --> then pressed F12 again to dismiss GUI and return to emulation. Doesn't happen, leaving me at the console from which cmd was launched (flashing cursor), no KBD input ---> ssh to machine to kill amiberry process. No input from KBD thereafter. So it seems to be something related to the |
|
@giantclambake There must be some other thing that triggers this. |
|
@midwan Turns out the config.uae file being used prompted the behavior -- it does not specify any bootable media, it just boots to the KS-1.3 insert disk hand (kickstart ROM init). I suspect in this emulation state, kbd & mouse aren't actually used (sans reset sequence), and things become unhinged handing these devices back to the emulation? The fact it doesn't display this behavior and works normally without the I would doubt this is typical usage as such ~ user may actually stumble into it, if they load a config.uae file, but they have moved the bootable media to a different location ; in that case it would just do kickstart ROM init and the user would likely end up at this impasse..with KMS....that's about the only way a user could 'accidentally' trip this. |
|
@giantclambake |
…show (#1401) Under KMSDRM, if you started Amiberry from the command line with the -G parameter, and you entered the GUI once, resuming would not show the emulation screen.
…show (#1401) Under KMSDRM, if you started Amiberry from the command line with the -G parameter, and you entered the GUI once, resuming would not show the emulation screen.
|
@giantclambake |
|
@midwan ...in x11 when using I'd call that completed ~ as always, thanks for the fix-ups =) |
|
Footnote: I had tested the above using RPi4/Master, and m93p Tiny/Preview to cover off x86-64 ~ everything worked as expected on both branches/platforms. I just retested on this setup using Nvidia drivers and G-Sync display. Same test in x11, hitting F12 does pop the GUI (GUI Always on top is enabled), however it's scaled entirely wrong, with about half the GUI hidden at the top. If then I click just outside the GUI surface to the left or right, the GUI immediately resizes/updates and is usable as in the tests....that is saying, clicking inside the GUI area at first is ostensibly useless ; you have to click just outside it for the GUI to refresh/resize, before it becomes usable. I'll need dig into it some, to see if it's just an NV driver thing ; I'll hoist another ticket about this if I find anything wrong that's not NV driver related ;) |
So the screen problem is solved. Thanks for that :) However, I experience an amiberry crash when I leave the menu with the "Resume" button under RetroPie with amiberry v5.7.4 (2024-08-24). |
Good idea. Thx for testing Mate! |
|
@midwan If Conversely, if This seems like a different issue to me, and not really related to this ticket....do you concur? |
@Retro1968 ~ does it crash if you use the F12 key instead to resume emulation? |
Yes, even exiting the menu with F12 causes amiberry to crash. |
|
@Retro1968 ~ I can't recreate any crash here... is this on your 32bit or 64bit OS install? (both??) |
|
RetroPie runs on 32-Bit Linux |
|
Didnt checked it on RaspiOS 64 Bit. |
|
The latest commits that fixed #1411 could possibly help with the crashes on RetroPie also. If the crashes were due to the old SDL2 version and KMSDRM, these latest fixes change the behavior to something similar that the older versions had (using a single window, when KMSDRM is detected), which could potentially solve this issue as well. Either way though, RetroPie needs to upgrade from Buster sooner or later, as it's out of support. I suspect that when that happens, since a newer version of SDL2 will come with it, many such issues will resolve themselves. After the next release comes out, I'll spend some time with the teams of RetroPie, Batocera, Knulli and MuOS, to help them bring this latest version to their distros in the best way possible. |
* enhancement: Gayle Null check (merged from WinUAE) * enhancement: Line 0 A1000 and OCS Agnus strobe fix * enhancement: OCS Agnus + OCS Denise config also has line 0 visible at the bottom of screen (like A1000) * Fix base color for Radiobuttons in SaveStates panel (#1293) * doc: added new option in usage help text (fixes #1319) * Update PanelChipset.cpp * bugfix: Always clear the text fields when deleting a config (fixes #1321) * bugfix: Fullscreen toggle didn't quite work (fixes #1322) * bugfix: Set right colors for checkboxes that were missed, so that themes work properly on them * bugfix: Use the right APMODE when toggling full-screen (fixes #1324) * Use the Default toggle screenmode instead of windowed<->fullscreen * feat: added window scaling option in amiberry.conf This option allows you to set the window scaling for the GUI window. The default option is 1.0, which is the normal 100% scaling. The value is a float, so you can dynamically set it to whatever you need, depending on your monitor's resolution. For example, you can set it to: 1.250 = 125% 1.50 = 150% 2.0 = 200% * enhancement: Fix AGA genlock transparency bit being detected as special blanking bits (merged from WinUAE) * refactor: rename gui theme internal variables * feat: implemented GUI foreground color option for themes * bugfix: Quickstart - enabling 2nd floppy drive did not refresh panel * doc: updated help text the --cdimage option does not change configuration, only inserts the CD. The help text did not clearly reflect that * bugfix: Make guisan widgets active/inactive status more visible (fixes #1326) Some widgets were not greyed out when disabled * enhancement: CPU panel had widgets that were not shown as disabled (#1326) The MMU and CPU Data Cache options were not visibly disabled as they should * Remove HighDPI window flag * bugfix: Fixed textbox background color in Create hardfile dialog (fixes #1327) * include <string> in CommonBridgeTemplate.h * refactor: minor tidy in SelectFolderf * Add stdint.h include in SerialIO.h (fixes compile with newer GCC) * sync resource.h with upstream * bugfix: Floppy GUI panel didn't show the select file buttons anymore (fixes #1331) * refactor: Create plugins directory, use it for external plugins (like floppybridge) From now on, external libraries such as the CAPSimg.so and libfloppybridge.so, will live in a separate directory named "plugins". This cleans up the main dir from such files. The new path is added to the GUI (Paths) and config files accordingly. The default value is <start_path>/plugins. The Makefile will now build floppybridge by default - CAPSimg is a submodule, so it needs to be checked out before it can be built. MacOS is a special case, since the libs there should be under the Resources folder. So, we're leaving that as-is for now. * ci: update pipeline with plugin changes * bugfix: only add the module extension if it's actually missing * ci: allow floppybridge to use older CMake versions also (3.13 for buster) * minor tidy * bugfix: Under KMSDRM the GUI would not open a second time (fixes #1333) - The window focus was not grabbed when resuming emulation after the GUI, so any subsequent attempt to re-enter the GUI failed. - Removed single_window_mode option, as it's too buggy and not really worth the trouble. * refactor: Use CD32 model as a fallback for CD auto-detection (fixes #1325) Instead of checking for CDTV, then CD32 and as a fallback going for A1200CD, we will look for CDTV and use CD32 as a fallback always. Most of the titles will be in those two categories anyway, and many might not contain that in the file path, leading to incorrectly trying to boot them (and failing) as an A1200 CD config. * enhancement: improved console handling in writelog * bugfix: if console_logging is requested, ensure it's activated always If console_logging was enabled, but write_logfile was not, then write_log would not show any output. This changes the behavior so that if console_logging is enabled, we always get output. * bugfix: Use WHDLoad custom1-5 values if they are not zero (fixes #1328) Since we always set the default value to zero, this should be safe to use. Apparently some games do not have some custom fields set, but still use them if specified (e.g. BattleIsle) * enhancement: ATAPI Write / Mode Select fix (merged from WinUAE) * enhancement: Serial port undocumented OVRUN behavior emulated. Fast CPU mode + loopback serial port mode fix. Merged from WinUAE * Updated version date * bugfix: the floppy dropdown wasn't updated properly after a selection (fixes #1334) * refactor: Use "Plugins" directory for macOS * Resize and center icon * Leave .so files in Resources for macOS, otherwise dlopen doesn't find them * Try to fix extension .so being added under macOS always * Revert "Resize and center icon" This reverts commit 5386f2b. * enhancement: Set config name automatically for CD images as well (fixes #1336) When Floppies or WHDLoad archives are inserted from the Quickstart panel, the last active config name is updated to the filename used. This did not apply to CD images however. This commit fixes that behavior, so it's consistent across all types of disk images. * feat: Added button to Create new folder in the dialogs (fixes #1329) A new button is now available in the Select File/Folder dialogs: Create Folder This allows the user to create a new directory inside the currently selected one. * bugfix: Fixed CD dropdown did not refresh properly (fixes #1337) When selecting a new CD image, the items in the dropdown did not refresh as expected in the Quickstart panel. This worked correctly in the HD/CD panel. Also synced the behavior a bit between the two panels. * bugfix: Fixed plugins directory name in macOS script * refactor: Remove CreateFolder button from SelectFile (fixes #1338) There's no need to Create a new folder, when you're about to select a File * ci: CMake will now copy libfloppybridge.so into plugins when finished * ci: add the new folders in flatpak as well * Updated version date * build: fix CMake copy of floppybridge lib destination path * ci: avoid relative paths for placing floppybridge library Apparently, this won't work everywhere - flatpak failed with the relative paths * ci: change destination of floppybridge library It should be the top-level source dir, not the floppybridge one * Try to fix the relative path once more * Revert "Try to fix the relative path once more" This reverts commit 9946d56. * Revert "ci: change destination of floppybridge library" This reverts commit af14d19. * ci: change copying of floppybridge lib, do it on the root CMake file instead * ci: Removed Buster platforms from pipeline targets Debian Buster is EOL soon, might as well remove it from the pipelines * Update README.md Minor updates, to indicate dropping support for Buster * Add keywords to .desktop file (#1342) These are also used by Flathub, KDE Discover and Gnome Software * enhancement: Improve WHDLoad config handling * bugfix: load any .uae config after parsing the XML settings, not before (fixes #1343) * ci: copy lib into plugins after the dir is created * bugfix: Fixed crash when saving config files (fixes #1345) * enhancement: When selecting CD32 model, set joystick port to CD32 also (fixes #1346) * Updated version to 6.3.2 * doc: Updated help text in Input panel (fixes #1349) * Update README.md * Update README.md * Change gui_always_on_top to true, by default (fixes #1351) Set the GUI to be always on top by default. This should help with cases where we are in Full screen mode, and sometimes the GUI opens behind the emulation screen. * feat: Implemented Integer scaling (fixes #1352) Added Integer scaling as an option in the Display panel. This saves to the config file, and can be set in amiberry.conf as well. The value 2 can be used in the scaling_method option, to set this as the default. * Package macOS bundles on each build * fix integer scaling was blurry (#1352) When using Integer scaling, the Nearest Neighbor SDL2 hint should be used * only enable GUI always on top, if we have fullscreen by default (#1351) We don't want the GUI to be always on top in Windowed mode also * improve integer scaling (#1352) * Remove obsolete option from amiberry.conf * Revert "improve integer scaling (#1352)" This reverts commit f3db2f2. * Don't set GUI always on top, try to raise the GUI window when it opens, instead (#1351) * build: ensure the floppybridge build is completed, before attempting to copy the lib * feat: implemented Resolution Autoswitch option This option comes from WinUAE, and when enabled, allows the automatic selection between horizontal LowRes/HiRes and vertical Single/Double line modes * enhancement: auto-crop scaling improvements, for integer scaling When we use Integer scaling, we don't want the width/height doubled during autocrop. One exception to that is when we have HiRes horizontally, and single line mode (where we need to double the height, in order to keep the aspect ratio correct) * bring back HighDPI support on windows This was removed as a test, but it didn't make a difference * enhancement: make "System Default" sound card option the default The System Default will now be enabled by default, as it seems to work best for most cases. If you want another sound output device selected, you can always change that. Config files should not be affected anyway, this is just a more sensible default. * bump version to 6.3.3 * enhancement: GUI window improvements - Set window to always on top, if we're running it in Full-window - RaiseWindow later in the process * bugfix: keyboard and mouse was not always grabbed (#1351) When going from GUI->Emulation repeatedly, sometimes the input was not grabbed as expected * bugfix: don't grab mouse/keyboard in the GUI window (fixes #1357) * Update README.md Added Ko-Fi support button * bugfix: VKBD would not show if Auto-Crop was enabled When Auto-Crop was enabled, the texture size was changed. However, VKBD did not update the X/Y coordinates it uses, so the image it showed could happen to be off-screen. * ci: Rename macOS M1 to macOS Apple Silicon * ci: fix apple silicon build * bugfix: Custom controls would not get loaded when starting from cmd line (fixes #1355) When starting Amiberry from the command line, with one of the autoload options, if a config file was found and loaded, the custom controls would never be applied. When doing the same from the GUI, it worked, but only by accident. The jports ID was set to -1 by default_prefs for all ports, in the currprefs structure. This didn't apply to changed_prefs, which had the default values of 0 for those fields. When using the GUI, which always applied settings to changed_prefs, it would work by accident since the ID happened to be zero, which also is the first available joystick in the did struct. Additionally, SDL2 detected a JOYSTICK INSERTED/REMOVED event after startup, which triggered the initialization of joysticks, which in turn loaded the default mapping back on them, overwriting any custom mapping previously loaded. To fix the above, some refactoring was needed: - Disabled the detection of JOYSTICK INSERTED/REMOVED during emulation. This is still triggered when the GUI is open. - Set the first two ports (0 and 1) to their default IDs for Mouse and first Joystick. This is diverging from WinUAE's defaults, but I think it will work out fine for us - Moved the loading of custom controls outside of cfgfile.cpp and into amiberry_input.cpp, where it belongs. This makes it easier to read, since it's not part of that huge parse_host function. - Fixed the loading of custom controls to apply to the uae_prefs struct passed into it, instead of always on currprefs, as it was before. - Added logical checks to ensure the port we're working on is valid (not and id of -1) and within the Joystick ports (value 0-8). We don't want to bother with unmapped ports or mouse ports. - Some more minor refactoring and cleanup was done around those areas as well. * enhancement: Make GUI window follow the same logic as the emulation one (#1351) - If KMSDRM is detected, set the mode to FULL-WINDOW - If a resolution less than 800x600 is detected, go for Full-window as well - Otherwise, we'll follow the settings requested regarding Windowed/Fullscreen/Full-window mode * feat: Added default option to control resolution autoswitch from amiberry.conf This option helps get the best results out of integer scaling, in combination with auto-crop and full-window modes * Updated version date * Change Full-screen toggle behavior - use full-window <-> window (#1351) Let's prefer Full-window instead of Fullscreen, to avoid screenmode changes on the monitor * refactor: return to previous approach: GUI window is always Windowed, except on KMSDRM (#1351) Due to how modern Window Managers handle things, if we open the GUI in full-window while the emulation screen is also opened in full-window, the second one gets minimized. That means that when we resume from the GUI, we don't get back to the emulation screen, even if we call for SDL_RaiseWindow() - it gets ignored. To resolve this, I'm reverting this behavior to the previous approach: The GUI is always Windowed, and opens on top of the emulation screen. The exception is when we're using KMSDRM, which doesn't have the notion of windows. But under KMSDRM we might run into a similar problem as well, so that needs a separate solution. * Update c-cpp.yml Try adding a `brew upgrade` step * Update c-cpp.yml * Update c-cpp.yml * Update c-cpp.yml * bugfix: Fix KMSDRM problems when switching between GUI -> emulation (fixes #1351) The emulation screen would not get the focus always, when resuming from the GUI, if running under KMSDRM. Removed the code that checked if the window had focus, and instead trying to grab it right away now. It still sometimes doesn't get it (not sure why), but at least you can click on it once and resume properly now. * bugfix: when loading custom mapping, skip the check for isrealbutton (#1355) * refactor: Refactored target_cfgfile_load * Updated version date * refactor: remove more checks for isrealbutton from controller mapping (#1355) * refactor: added some more logging in target_cfgfile_load function (#1355) * refactor: accept .uae config files in cmd line, even if they are not the last item (#1355) uae config files and savestates were only accepted from the command line, if they were the last parameters. If we however wanted to include -G after them (to hide the GUI), they would not get loaded. * Updated version date * build: add USE_GPIOD in CMakeLists.txt also * Remove some entries from CMakeSettings.json * bugfix: store joystick GUID, use it to check before re-importing joysticks (#1355) * feat: Auto-detect MT32 titles in whdbooter Auto-detect any titles supporting MT32 MIDI output in WHDBooter, and enable the relevant feature in Amiberry (if the ROMs are installed). This currently works based on the filename (it has to contain the word "MT32"), as it's not included in the XML. Not many title supported MT32, so maybe this is good enough. * Updated version date * Merged latest WinUAE updates Added support for Picmatic Tierras Salvajes and Marbella Vice * Update README.md * bugfix: fixed typo in Sound panel (fixes #1369) * refactor: use string.empty() * enhancement: Added VSync options from WinUAE in Display panel (#1368) Added the WinUAE VSync options for Native and RTG as dropdowns, instead of the single checkbox we had before. The Lagless modes are not implemented, not even sure we can do that in SDL2. The other modes work the same way as in WinUAE. Amiberry still uses V-Sync internally for the SDL_Presenter, so this only changes what is passed internally in the emulation. These options only make sense to enable if the monitor is already set to the right refresh rate - i.e. don't enable if your monitor only does 60Hz and you want to emulate a PAL machine (which needs 50). * bugfix: Fixed changing logfile name would append "amiberry.log" at the end The SelectFile dialog had an incorrect filter setting, which ended up appending the whole default filename instead of only the ".log" extension, if it wasn't found. * refactor: slight refactor of mapping from retroarch function * bugfix: joystick input didn't work with retroarch mapping after latest changes (#1363) * Bumped version to 6.3.4 * Macos folder changes (#1373) * refactor: use separate dir for amiberry.conf and data under macOS (#1372) Moving the contents of the data folder and amiberry.conf, under <user>/Library/Application Support/Amiberry for macOS platforms * Fix capsimg (#1374) * Remove capsimg submodule * Add capsimg sources * Add dependencies in Makefile and CMakeLists.txt to build libcapsimage.so * Fix path in makefile * Add libs to make clean also * Automatically build capsimg as well * Remove extra step to build capsimg from pipeline * bugfix: Fixed new/free mismatch in amiberry_filesys (#1387) * bugfix: Fixed display modes iteration (fixes #1394) * enhancement: Make Paths panel textboxes read-only (fixes #1385) * bugfix: always refresh floppy/quickstart panels after disk changes (fixes #1378) * bugfix: Refresh all panels when WHDLoad action from Quickstart is taken (fixes #1383) * Add FPS Adjustment options to the Display panel (#1399) * enhancement: Add FPS Adjustment options to the Display panel (fixes #1397) * Fix build under macOS * Updated version date * bugfix: Fix Picmatic (100Hz TV only) light gun positioning in lores or shres resolutions * bugfix: Fix duplicated Golem SCSI II / HD3000 name prefix. * bugfix: Always clear intreq2 in audio state 5. * enhancement: add secondary check when loading cfg files, look in configs folder for filename (fixes #1388) * enhancement: resample audio during Push operations (#1397) * bugfix: detect uppercase MT32 filenames also (fixes #1386) (#1400) * Updated XML to latest version * Improve macOS instructions (fixes #1382) * enhancement: fix inconsistency regarding default config name (fixes #1395) * Minor sync with WinUAE No functional changes, just minimizing diffs from WinUAE sources * Another round of syncs to WinUAE * enhancement: GUI - Make read-only textboxes in WHDLoad panel show as such (fixes #1403) Make the read-only text boxes show as such (disabled), since no user input is supported there * Updated version date * Minor sync with WinUAE * enhancement: updated FloppyBridge to v1.6.4 * Updated version date * enhancement: updated game controller DB to latest version * enhancement: updated WHDLoad XML to latest version * bugfix: don't refresh the emulation screen when GUI is open, under KMSDRM (#1401) * bugfix: Don't replace spaces in native_dos_op With the newer version of host-run (v1.6 onwards), this should not be necessary anymore * bugfix: Fixed bug under KMSDRM, where the emulation screen would not show (#1401) Under KMSDRM, if you started Amiberry from the command line with the -G parameter, and you entered the GUI once, resuming would not show the emulation screen. * bugfix: remove newlines from debugger input (#1407) The "fd" command didn't work in the debugger. This was because the "d" was followed in the input buffer by \n rather than \0. The Windows version of console_get replaces the terminating \r with \0. amiberry's version uses fgets, which doesn't do this; it did adjust the length value returned, but the debugger code doesn't look at the length. Make it actually strip the newline characters when adjusting the length. * bugfix: release window grab in the debugger (#1406) When the debugger is entered with Shift-F12, this function is called to release the mouse pointer. The second argument to SDL_SetWindowGrab should be SDL_FALSE to disable the grab; previously, both _acquire and _release had SDL_TRUE, which enables the grab. * enhancement: improve help text in Misc panel (fixes #1408) * Revert "enhancement: resample audio during Push operations (#1397)" This reverts commit a70dd17. * enhancement: auto-select the first item in the Configs list (fixes #1391) If we have items in the Configs list, automatically select the first one. * enhancement: Log command line parameters on startup (fixes #1380) If logging is enabled, also log the full command line used to start up * enhancement: harmonize buttons in HDD GUI panel Change label on Add Directory, to indicate it also supports archives. Harmonize the widths of those buttons. Improve label on CD select image file. * Added CMakePresets.json * enhancement: use string.empty() instead of comparing with "" * bugfix: When using KMSDRM, controller input was lost after entering the GUI once (fixes #1384) This seems to be caused by SDL2 when using KMSDRM only, as it works properly under X. As a workaround, I'm changing the logic to keep only a single window and renderer when KMSDRM is detected. * Updated version date * bugfix: KMSDRM fixes for when using a single window (fixes #1411) Disable integer scaling when opening the GUI, trigger an update of scaling options when resuming * enhancement: separate data from home directories, improve handling (fixes #1073) - Separate DATA from HOME directories - Use special ENV variables to override locations: AMIBERRY_DATA_DIR and AMIBERRY_HOME_DIR respectively - Use the standard Linux locations if the env variables are not set: /usr/share/amiberry/data for the DATA folder, and $HOME/.amiberry for the HOME folder - Fallback to the Portable mode used before, if all the above fail * enhancement: Fix new paths logic for Flatpak also (#1073) * enhancement: improve look of GUI Misc panel scrollbar (fixes #1412) * 1376 add a3000 scsi controller and tape support for amiga unix amix (#1414) * enhancement: added CD and Tape drive buttons Added new buttons in the GUI: HDD panel, for handling CD and Tape drives. * Refactoring of GUI HD Panel - Added TapeDrive option - Refactored GUI panels to look more like WinUAE - Refactored functionality to closer match that of WinUAE * bugfix: Fixes for HDD panel windows * bugfix: Fixed default paths in dialogs * enhancement: Harmonize TapeDrive dialog logic with others Set the default path (harddrives) if no path is configured, before opening the dialog * enhancement: Updated Create HF dialog - The dialog will now only create HDFs, not add them as well * enhancement: fix navigation in Create HDF dialog * enhancement: Added Special Monitors support * enhancement: Added special monitors support in GUI * enhancement: fixed navigation for Tape Drive and other HDD panel dialogs * enhancement: improvements in the Add HardDrive dialog Added controller, unit, type and feature level dropdowns * enhancement: fixed navigation in Panel Harddrive * disable button that's not implemented yet * Updated version date * ci: Remove Manjaro RPI4 target from releases Only AmiKit was using that, and it moved to Debian12 with the new update * refactor: Move "abr" under plugins directory - Handle "abr" as one of the plugins, instead of having it living on a directory in the home dir. - Moved directory, updated paths and macOS scripts accordingly. - get_plugin_path() now returns the actual plugins path * enhancement: expanded DBus support with new events New DBus events added: - Insert Floppy - Insert CD * enhancement: added one more native lib path Added /usr/lib/amiberry as one more native lib search path * enhancement: Added option to use JST instead of WHDLoad binary In case someone needs this, there is now an option to use JST instead of WHDLoad for the Booter. use_jst_instead_of_whd (boolean) The option is saved in the global configuration of amiberry.conf and when enabled, will switch out the startup-sequence references from WHDLoad to JST, including any differences in options. The old AMOS-based booter is not changed with this. * enhancement: add filetype associations under macOS Handle .uae, .adf and .lha files with Amiberry, under macOS * bugfix: Fixed various glitches in SelectFile dialog (fixes #1417) - Do not attempt to scroll if the items are less than 32, as they fit in the screen already - Show the selected item in the text field - Allow existing files to be selected and used, even when the "createNew" flag is set - Fixed path building, it would always append to the existing string - Fixed string assignment in changeDir * enhancement: always show horizontal scrollbar in Selectfile dialog (fixes #1415) * build(deps): bump actions/download-artifact in /.github/workflows (#1419) Bumps [actions/download-artifact](https://github.com/actions/download-artifact) from 4.1.2 to 4.1.7. - [Release notes](https://github.com/actions/download-artifact/releases) - [Commits](actions/download-artifact@v4.1.2...v4.1.7) --- updated-dependencies: - dependency-name: actions/download-artifact dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * enhancement: upgrade libguisan to latest version, fix various bugs Guisan has been getting several updates lately, improving and fixing various bugs. Upgrading to the current version, but keeping my own improvements until I've had time to push them upstream * bugfix: Listboxes would not resize in width (fixes #1415) * enhancement: Guisan: Window widget needs to draw frame in inverted colors * bugfix: Fixed guisan theme colors (fixes #1420) * bugfix: Some Textboxes should not have a frame drawn * enhancement: Guisan: improve widget help text * bugfix: Fixed capitalization in RAM panel (fixes #1422) * enhancement: apply scaling factor to emulation window also (fixes #1421) * enhancement: Guisan updates Added Text class, refactored textbox, textfield etc. * bugfix: Fixed slider background color after guisan updates * bugfix: Fix CPU speed slider scale, when Fastest Possible was selected * enhancement: added .bin, .cdtv and .cd32 as valid rom extensions * bugfix: fixed dropdown and checkbox colors when disabled * bugfix: text widget should clear previous rows when setting context * bugfix: macOS Bundle name should be Amiberry * cleanup: remove CMakeSettings.json (obsolete) This is no longer needed, as we're using CMakePresets.json instead * enhancement: Check if the GUI theme font has the full path also - If the gui_theme.font_name contains the full path to the font, open it from there directly. This means we can now use system fonts as well, from /usr/share/fonts/... - If the filename is not found, then try to open it from the data directory, as before * ci: added Debian files * build: added CMake install section * build: change paths for cmake install * build: added install targets for guisan and floppybridge * build: ensure external libs are built before binary under CMake * build: added CPack support * chore: guisan: harmonize all files to use spaces and same line endings * enhancement: guisan: radiobutton and togglebutton fixes These widgets didn't use the proper disabled color * enhancement: guisan: fix remaining widgets colors These widgets were also missing their disabled color * enhancement: guisan: clang tidy, use const when possible * bugfix: Fixed navigation in Sound panel It wasn't possible to navigate into the Sound panel, since the first control (dropdown at the top) was disabled by default. * enhancement: add support for XDG_CONFIG_HOME (fixes #1425) Added logic to pick up the config directory, from the first match of the following: - $AMIBERY_CONFIG_DIR - $XDG_CONFIG_HOME/amiberry - $HOME/.config/amiberry - $HOME/.amiberry/conf if none of the above is found, the fallback is the current logic (portable mode, <current dir>/conf) * enhancement: Gracefully handle iconv failures If iconv_open() fails for some reason, do not fail completely but instead try to copy the entries verbatim. * bugfix: Refresh RTG panel when CPU changes are modified - The 24-bit addressing option blocks RTG cards from being used. When it's disabled, the RTG panel should refresh, to enable any relevant widgets - Fixed navigation in RTG panel (the VRAM slider could not be accessed) * chore: remove unneeded ifdef * Big bang winuae merge (#1428) * enhancement: Merge more WinUAE code - Added PPC - Added PCEM - Added PCI - Added MAME (A2410 and TM34010) - Added QEMU (various gfx cards, network cards, SCSI, etc) - Enabled relevant ifdefs * enhancement: Added luascript, slirp_uae - Removed unused ersatz - Added luascript (disabled for now, needs an extra library) - Added slirp_uae (disabled for now) - Added sources for separate tool "uaeunp" * enhancement: Added DSP emulation - Added DSP emulation files from WinUAE, with slight modifications (disabled for now) * enhancement: Enabled TabletLibrary, DSP, Draco Not implemented on the GUI yet, but enabled related ifdefs in the code * enhancement: Added SANA2 support Not visible on the GUI yet * enhancement: sync includes and other details with WinUAE * enhancement: added support for more RTG boards * bugfix: added missing arm/arm64 headers for pcem codegen_backend WinUAE didn't have these, so let's grab them from PCem v16 directly * enhancement: Fixed defines in picasso96.cpp * bugfix: Maybe we should actually sleep for a few ms in sleep_millis2 * enhancement: P96 - added ifdefs where Amiberry-specific pieces are added * chore: sync time/rpt headers with WinUAE * chore: reduce the lines of ifdefs in ahi_v2 * enhancement: use SDL_Delay instead of waiting for the cpu_wakeup_event * perf: only call init_row_map if the pixels have changed * enhancement: Added GVP T-Rex II Merged from WinUAE * enhancement: Added DKB WildFire v1.0 and Masoboshi v2.197 Merged from WinUAE * enhancement: Flush trap queues during reset. * bugfix: Added missing S3 Virge PCI byteswap mode Merged from WinUAE * perf: Do not unnecessarily call SetCursor() continuously. * bugfix: Fixed uae_sem_unpost This should be SemPost, not SemWait * enhancement: Allow 16-bit only if D3D9 is selected. Merged from WinUAE - Amiberry uses gfx_api value 2 anyway. * bugfix: Fix SPRxPOS modification just before horizontal match if sprite had lowest horizontal bit set. Merged from WinUAE * enhancement: Both COPxJMP strobe emulation update Merged from WinUAE * enhancement: Serial emulation updates, more accurate transmit timing, receive timing is not anymore instant except in direct mode. Merged from WinUAE * enhancement: OpalVision undocumented horizontal panning emulation address load was ignored if it was included in second control line. * perf: Reduce amount of read_processor_time() calls. Merged from WinUAE * build: updated makefile with new files * enhancement: added PPC options to GUI * enhancement: added MapROM option in GUI * bugfix: GUI RAM Panel did not highlight selected items properly * enhancement: added label in Input panel for Parallel Port joystick adapter The label should make it more clear that the last two ports are for the emulated parallel port joystick adapter * enhancement: GUI listbox inactive color improvement The guisan listbox would use an inactive color of a fixed value. It should use something from the color scheme, instead. * enhancement: added more models and configs in Quickstart * build: set the compiler standard higher in the Makefile * bugfix: fix ambiguous variable names * ci: avoid using named initializers in file, otherwise we run into a GCC bug This code won't compile with older GCC versions, due to a known bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=55227 * ci: fixed voodo file as well (see previous commit reason) * ci: fixed one more file with named initializers * ci: upgrade the minimum CMake version for capsimage Get rid of a warning about older CMake versions being deprecated soon * ci: change SDL_ttf include in sdltruetypefont.hpp * ci: updated more SDL includes in guisan * ci: Cmake builds would fail if started under a subdir (#1430) * ci: Cmake builds would fail if started under a subdir * ci: add cmake files needed for bullseye * ci: updated macOS include directories * ci: fix compiler error under clang/macOS * ci: remove unused linker flag * ci: add cmake in macOS homebrew * chore: fix compiler warnings in calc Well, most of them anyway * Revert "build: set the compiler standard higher in the Makefile" This reverts commit e822fe8. * bugfix: Editing entries in HD GUI panel was broken * ci: ensure Apple ARM builds don't detect both arm32 and arm64 defines * ci: more Cmake files for MacOS * ci: macos fixes, guisan should be static * ci: cleanup makefile macos targets * ci: fix macos apple silicon flags * ci: fix cmake linker flags for macos * ci: more macos Makefile fixes * ci: MacOS fix - remove rt lib from linker in Makefile * ci: MacOS fixes - Remove one more lib from the linker in the Makefile * ci: MacOS Makefile fixes: link to iconv * ci: Fixed typo * chore: make it clear that these flags are for ARM builds only * ci: Remove cmake and wget from MacOS homebrew runner these are already installed in the runner environment anyway * ci: remove autoconf from macos homebrew runner as well * ci: Fix capsimg filename for MacOS * Update make-bundle.sh * Remove unnecessary architecture detection * First imports of toolchain files for cross-compiling * Yet another fix for CMake that will help both OSX and Debian Bullseye * Cmake packaging (#1431) * modify CMakeLists.txt to be able to build packages for MacOS and Linux * Moved pull request template to .github dir * Moved code of conduct file under .github directory * Removed gitmodules file * Moved install extra dirs to only if we're not building an Apple bundle * Added first version of a Deb package option to CPack * Fixed double directory creation during installation/packaging * Remove debian directory We won't be using this, going for CPack instead * Fixed package version * Clean up contents of packages * Fix dependencies and filename of .deb package * Bump version to 6.3.5 * Add desktop file for Deb package * Remove archive generator, the DEB one is good enough * Change pipeline to use CMake for Linux targets Let's generate some .deb files! This is most likely missing stuff, like the distro name... * Fix incorrect architecture names in cross-compiler * Fix 32-bit arm package architecture * ci: Allow 32-bit arm builds * ci: disable the bullseye builds for now Until we figure out a solution with Bullseye + CMake + SDL2_image/SDL2_ttf * ci: fix expected targets * ci: bring back bullseye builds * Remove path from desktop shortcut * Fixed typo in comment * Set the version in the Project, use that for CPack * Updated compiled and linker flags with no-pie properly - There were some differences between the Makefile flags and CMake. Now they are identical - Updated the Findmpg123.cmake file with the latest one from upstream * Include bullseye build targets again * Increase GFX MAX_RTG_Boards to 4 as it looks like Picasso96 code is expecting 4 not the 1 listed. (#1434) --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: Alexander Wilms <Alexander-Wilms@users.noreply.github.com> Co-authored-by: Adam Sampson <ats-github@offog.org> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Solskogen <christer.solskogen@gmail.com> Co-authored-by: BluewizardNet <ryan@bluewizard.net>
* enhancement: OCS Agnus + OCS Denise config also has line 0 visible at the bottom of screen (like A1000) * Fix base color for Radiobuttons in SaveStates panel (#1293) * doc: added new option in usage help text (fixes #1319) * Update PanelChipset.cpp * bugfix: Always clear the text fields when deleting a config (fixes #1321) * bugfix: Fullscreen toggle didn't quite work (fixes #1322) * bugfix: Set right colors for checkboxes that were missed, so that themes work properly on them * bugfix: Use the right APMODE when toggling full-screen (fixes #1324) * Use the Default toggle screenmode instead of windowed<->fullscreen * feat: added window scaling option in amiberry.conf This option allows you to set the window scaling for the GUI window. The default option is 1.0, which is the normal 100% scaling. The value is a float, so you can dynamically set it to whatever you need, depending on your monitor's resolution. For example, you can set it to: 1.250 = 125% 1.50 = 150% 2.0 = 200% * enhancement: Fix AGA genlock transparency bit being detected as special blanking bits (merged from WinUAE) * refactor: rename gui theme internal variables * feat: implemented GUI foreground color option for themes * bugfix: Quickstart - enabling 2nd floppy drive did not refresh panel * doc: updated help text the --cdimage option does not change configuration, only inserts the CD. The help text did not clearly reflect that * bugfix: Make guisan widgets active/inactive status more visible (fixes #1326) Some widgets were not greyed out when disabled * enhancement: CPU panel had widgets that were not shown as disabled (#1326) The MMU and CPU Data Cache options were not visibly disabled as they should * Remove HighDPI window flag * bugfix: Fixed textbox background color in Create hardfile dialog (fixes #1327) * include <string> in CommonBridgeTemplate.h * refactor: minor tidy in SelectFolderf * Add stdint.h include in SerialIO.h (fixes compile with newer GCC) * sync resource.h with upstream * bugfix: Floppy GUI panel didn't show the select file buttons anymore (fixes #1331) * refactor: Create plugins directory, use it for external plugins (like floppybridge) From now on, external libraries such as the CAPSimg.so and libfloppybridge.so, will live in a separate directory named "plugins". This cleans up the main dir from such files. The new path is added to the GUI (Paths) and config files accordingly. The default value is <start_path>/plugins. The Makefile will now build floppybridge by default - CAPSimg is a submodule, so it needs to be checked out before it can be built. MacOS is a special case, since the libs there should be under the Resources folder. So, we're leaving that as-is for now. * ci: update pipeline with plugin changes * bugfix: only add the module extension if it's actually missing * ci: allow floppybridge to use older CMake versions also (3.13 for buster) * minor tidy * bugfix: Under KMSDRM the GUI would not open a second time (fixes #1333) - The window focus was not grabbed when resuming emulation after the GUI, so any subsequent attempt to re-enter the GUI failed. - Removed single_window_mode option, as it's too buggy and not really worth the trouble. * refactor: Use CD32 model as a fallback for CD auto-detection (fixes #1325) Instead of checking for CDTV, then CD32 and as a fallback going for A1200CD, we will look for CDTV and use CD32 as a fallback always. Most of the titles will be in those two categories anyway, and many might not contain that in the file path, leading to incorrectly trying to boot them (and failing) as an A1200 CD config. * enhancement: improved console handling in writelog * bugfix: if console_logging is requested, ensure it's activated always If console_logging was enabled, but write_logfile was not, then write_log would not show any output. This changes the behavior so that if console_logging is enabled, we always get output. * bugfix: Use WHDLoad custom1-5 values if they are not zero (fixes #1328) Since we always set the default value to zero, this should be safe to use. Apparently some games do not have some custom fields set, but still use them if specified (e.g. BattleIsle) * enhancement: ATAPI Write / Mode Select fix (merged from WinUAE) * enhancement: Serial port undocumented OVRUN behavior emulated. Fast CPU mode + loopback serial port mode fix. Merged from WinUAE * Updated version date * bugfix: the floppy dropdown wasn't updated properly after a selection (fixes #1334) * refactor: Use "Plugins" directory for macOS * Resize and center icon * Leave .so files in Resources for macOS, otherwise dlopen doesn't find them * Try to fix extension .so being added under macOS always * Revert "Resize and center icon" This reverts commit 5386f2b. * enhancement: Set config name automatically for CD images as well (fixes #1336) When Floppies or WHDLoad archives are inserted from the Quickstart panel, the last active config name is updated to the filename used. This did not apply to CD images however. This commit fixes that behavior, so it's consistent across all types of disk images. * feat: Added button to Create new folder in the dialogs (fixes #1329) A new button is now available in the Select File/Folder dialogs: Create Folder This allows the user to create a new directory inside the currently selected one. * bugfix: Fixed CD dropdown did not refresh properly (fixes #1337) When selecting a new CD image, the items in the dropdown did not refresh as expected in the Quickstart panel. This worked correctly in the HD/CD panel. Also synced the behavior a bit between the two panels. * bugfix: Fixed plugins directory name in macOS script * refactor: Remove CreateFolder button from SelectFile (fixes #1338) There's no need to Create a new folder, when you're about to select a File * ci: CMake will now copy libfloppybridge.so into plugins when finished * ci: add the new folders in flatpak as well * Updated version date * build: fix CMake copy of floppybridge lib destination path * ci: avoid relative paths for placing floppybridge library Apparently, this won't work everywhere - flatpak failed with the relative paths * ci: change destination of floppybridge library It should be the top-level source dir, not the floppybridge one * Try to fix the relative path once more * Revert "Try to fix the relative path once more" This reverts commit 9946d56. * Revert "ci: change destination of floppybridge library" This reverts commit af14d19. * ci: change copying of floppybridge lib, do it on the root CMake file instead * ci: Removed Buster platforms from pipeline targets Debian Buster is EOL soon, might as well remove it from the pipelines * Update README.md Minor updates, to indicate dropping support for Buster * Add keywords to .desktop file (#1342) These are also used by Flathub, KDE Discover and Gnome Software * enhancement: Improve WHDLoad config handling * bugfix: load any .uae config after parsing the XML settings, not before (fixes #1343) * ci: copy lib into plugins after the dir is created * bugfix: Fixed crash when saving config files (fixes #1345) * enhancement: When selecting CD32 model, set joystick port to CD32 also (fixes #1346) * Updated version to 6.3.2 * doc: Updated help text in Input panel (fixes #1349) * Update README.md * Update README.md * Change gui_always_on_top to true, by default (fixes #1351) Set the GUI to be always on top by default. This should help with cases where we are in Full screen mode, and sometimes the GUI opens behind the emulation screen. * feat: Implemented Integer scaling (fixes #1352) Added Integer scaling as an option in the Display panel. This saves to the config file, and can be set in amiberry.conf as well. The value 2 can be used in the scaling_method option, to set this as the default. * Package macOS bundles on each build * fix integer scaling was blurry (#1352) When using Integer scaling, the Nearest Neighbor SDL2 hint should be used * only enable GUI always on top, if we have fullscreen by default (#1351) We don't want the GUI to be always on top in Windowed mode also * improve integer scaling (#1352) * Remove obsolete option from amiberry.conf * Revert "improve integer scaling (#1352)" This reverts commit f3db2f2. * Don't set GUI always on top, try to raise the GUI window when it opens, instead (#1351) * build: ensure the floppybridge build is completed, before attempting to copy the lib * feat: implemented Resolution Autoswitch option This option comes from WinUAE, and when enabled, allows the automatic selection between horizontal LowRes/HiRes and vertical Single/Double line modes * enhancement: auto-crop scaling improvements, for integer scaling When we use Integer scaling, we don't want the width/height doubled during autocrop. One exception to that is when we have HiRes horizontally, and single line mode (where we need to double the height, in order to keep the aspect ratio correct) * bring back HighDPI support on windows This was removed as a test, but it didn't make a difference * enhancement: make "System Default" sound card option the default The System Default will now be enabled by default, as it seems to work best for most cases. If you want another sound output device selected, you can always change that. Config files should not be affected anyway, this is just a more sensible default. * bump version to 6.3.3 * enhancement: GUI window improvements - Set window to always on top, if we're running it in Full-window - RaiseWindow later in the process * bugfix: keyboard and mouse was not always grabbed (#1351) When going from GUI->Emulation repeatedly, sometimes the input was not grabbed as expected * bugfix: don't grab mouse/keyboard in the GUI window (fixes #1357) * Update README.md Added Ko-Fi support button * bugfix: VKBD would not show if Auto-Crop was enabled When Auto-Crop was enabled, the texture size was changed. However, VKBD did not update the X/Y coordinates it uses, so the image it showed could happen to be off-screen. * ci: Rename macOS M1 to macOS Apple Silicon * ci: fix apple silicon build * bugfix: Custom controls would not get loaded when starting from cmd line (fixes #1355) When starting Amiberry from the command line, with one of the autoload options, if a config file was found and loaded, the custom controls would never be applied. When doing the same from the GUI, it worked, but only by accident. The jports ID was set to -1 by default_prefs for all ports, in the currprefs structure. This didn't apply to changed_prefs, which had the default values of 0 for those fields. When using the GUI, which always applied settings to changed_prefs, it would work by accident since the ID happened to be zero, which also is the first available joystick in the did struct. Additionally, SDL2 detected a JOYSTICK INSERTED/REMOVED event after startup, which triggered the initialization of joysticks, which in turn loaded the default mapping back on them, overwriting any custom mapping previously loaded. To fix the above, some refactoring was needed: - Disabled the detection of JOYSTICK INSERTED/REMOVED during emulation. This is still triggered when the GUI is open. - Set the first two ports (0 and 1) to their default IDs for Mouse and first Joystick. This is diverging from WinUAE's defaults, but I think it will work out fine for us - Moved the loading of custom controls outside of cfgfile.cpp and into amiberry_input.cpp, where it belongs. This makes it easier to read, since it's not part of that huge parse_host function. - Fixed the loading of custom controls to apply to the uae_prefs struct passed into it, instead of always on currprefs, as it was before. - Added logical checks to ensure the port we're working on is valid (not and id of -1) and within the Joystick ports (value 0-8). We don't want to bother with unmapped ports or mouse ports. - Some more minor refactoring and cleanup was done around those areas as well. * enhancement: Make GUI window follow the same logic as the emulation one (#1351) - If KMSDRM is detected, set the mode to FULL-WINDOW - If a resolution less than 800x600 is detected, go for Full-window as well - Otherwise, we'll follow the settings requested regarding Windowed/Fullscreen/Full-window mode * feat: Added default option to control resolution autoswitch from amiberry.conf This option helps get the best results out of integer scaling, in combination with auto-crop and full-window modes * Updated version date * Change Full-screen toggle behavior - use full-window <-> window (#1351) Let's prefer Full-window instead of Fullscreen, to avoid screenmode changes on the monitor * refactor: return to previous approach: GUI window is always Windowed, except on KMSDRM (#1351) Due to how modern Window Managers handle things, if we open the GUI in full-window while the emulation screen is also opened in full-window, the second one gets minimized. That means that when we resume from the GUI, we don't get back to the emulation screen, even if we call for SDL_RaiseWindow() - it gets ignored. To resolve this, I'm reverting this behavior to the previous approach: The GUI is always Windowed, and opens on top of the emulation screen. The exception is when we're using KMSDRM, which doesn't have the notion of windows. But under KMSDRM we might run into a similar problem as well, so that needs a separate solution. * Update c-cpp.yml Try adding a `brew upgrade` step * Update c-cpp.yml * Update c-cpp.yml * Update c-cpp.yml * bugfix: Fix KMSDRM problems when switching between GUI -> emulation (fixes #1351) The emulation screen would not get the focus always, when resuming from the GUI, if running under KMSDRM. Removed the code that checked if the window had focus, and instead trying to grab it right away now. It still sometimes doesn't get it (not sure why), but at least you can click on it once and resume properly now. * bugfix: when loading custom mapping, skip the check for isrealbutton (#1355) * refactor: Refactored target_cfgfile_load * Updated version date * refactor: remove more checks for isrealbutton from controller mapping (#1355) * refactor: added some more logging in target_cfgfile_load function (#1355) * refactor: accept .uae config files in cmd line, even if they are not the last item (#1355) uae config files and savestates were only accepted from the command line, if they were the last parameters. If we however wanted to include -G after them (to hide the GUI), they would not get loaded. * Updated version date * build: add USE_GPIOD in CMakeLists.txt also * Remove some entries from CMakeSettings.json * bugfix: store joystick GUID, use it to check before re-importing joysticks (#1355) * feat: Auto-detect MT32 titles in whdbooter Auto-detect any titles supporting MT32 MIDI output in WHDBooter, and enable the relevant feature in Amiberry (if the ROMs are installed). This currently works based on the filename (it has to contain the word "MT32"), as it's not included in the XML. Not many title supported MT32, so maybe this is good enough. * Updated version date * Merged latest WinUAE updates Added support for Picmatic Tierras Salvajes and Marbella Vice * Update README.md * bugfix: fixed typo in Sound panel (fixes #1369) * refactor: use string.empty() * enhancement: Added VSync options from WinUAE in Display panel (#1368) Added the WinUAE VSync options for Native and RTG as dropdowns, instead of the single checkbox we had before. The Lagless modes are not implemented, not even sure we can do that in SDL2. The other modes work the same way as in WinUAE. Amiberry still uses V-Sync internally for the SDL_Presenter, so this only changes what is passed internally in the emulation. These options only make sense to enable if the monitor is already set to the right refresh rate - i.e. don't enable if your monitor only does 60Hz and you want to emulate a PAL machine (which needs 50). * bugfix: Fixed changing logfile name would append "amiberry.log" at the end The SelectFile dialog had an incorrect filter setting, which ended up appending the whole default filename instead of only the ".log" extension, if it wasn't found. * refactor: slight refactor of mapping from retroarch function * bugfix: joystick input didn't work with retroarch mapping after latest changes (#1363) * Bumped version to 6.3.4 * Macos folder changes (#1373) * refactor: use separate dir for amiberry.conf and data under macOS (#1372) Moving the contents of the data folder and amiberry.conf, under <user>/Library/Application Support/Amiberry for macOS platforms * Fix capsimg (#1374) * Remove capsimg submodule * Add capsimg sources * Add dependencies in Makefile and CMakeLists.txt to build libcapsimage.so * Fix path in makefile * Add libs to make clean also * Automatically build capsimg as well * Remove extra step to build capsimg from pipeline * bugfix: Fixed new/free mismatch in amiberry_filesys (#1387) * bugfix: Fixed display modes iteration (fixes #1394) * enhancement: Make Paths panel textboxes read-only (fixes #1385) * bugfix: always refresh floppy/quickstart panels after disk changes (fixes #1378) * bugfix: Refresh all panels when WHDLoad action from Quickstart is taken (fixes #1383) * Add FPS Adjustment options to the Display panel (#1399) * enhancement: Add FPS Adjustment options to the Display panel (fixes #1397) * Fix build under macOS * Updated version date * bugfix: Fix Picmatic (100Hz TV only) light gun positioning in lores or shres resolutions * bugfix: Fix duplicated Golem SCSI II / HD3000 name prefix. * bugfix: Always clear intreq2 in audio state 5. * enhancement: add secondary check when loading cfg files, look in configs folder for filename (fixes #1388) * enhancement: resample audio during Push operations (#1397) * bugfix: detect uppercase MT32 filenames also (fixes #1386) (#1400) * Updated XML to latest version * Improve macOS instructions (fixes #1382) * enhancement: fix inconsistency regarding default config name (fixes #1395) * Minor sync with WinUAE No functional changes, just minimizing diffs from WinUAE sources * Another round of syncs to WinUAE * enhancement: GUI - Make read-only textboxes in WHDLoad panel show as such (fixes #1403) Make the read-only text boxes show as such (disabled), since no user input is supported there * Updated version date * Minor sync with WinUAE * enhancement: updated FloppyBridge to v1.6.4 * Updated version date * enhancement: updated game controller DB to latest version * enhancement: updated WHDLoad XML to latest version * bugfix: don't refresh the emulation screen when GUI is open, under KMSDRM (#1401) * bugfix: Don't replace spaces in native_dos_op With the newer version of host-run (v1.6 onwards), this should not be necessary anymore * bugfix: Fixed bug under KMSDRM, where the emulation screen would not show (#1401) Under KMSDRM, if you started Amiberry from the command line with the -G parameter, and you entered the GUI once, resuming would not show the emulation screen. * bugfix: remove newlines from debugger input (#1407) The "fd" command didn't work in the debugger. This was because the "d" was followed in the input buffer by \n rather than \0. The Windows version of console_get replaces the terminating \r with \0. amiberry's version uses fgets, which doesn't do this; it did adjust the length value returned, but the debugger code doesn't look at the length. Make it actually strip the newline characters when adjusting the length. * bugfix: release window grab in the debugger (#1406) When the debugger is entered with Shift-F12, this function is called to release the mouse pointer. The second argument to SDL_SetWindowGrab should be SDL_FALSE to disable the grab; previously, both _acquire and _release had SDL_TRUE, which enables the grab. * enhancement: improve help text in Misc panel (fixes #1408) * Revert "enhancement: resample audio during Push operations (#1397)" This reverts commit a70dd17. * enhancement: auto-select the first item in the Configs list (fixes #1391) If we have items in the Configs list, automatically select the first one. * enhancement: Log command line parameters on startup (fixes #1380) If logging is enabled, also log the full command line used to start up * enhancement: harmonize buttons in HDD GUI panel Change label on Add Directory, to indicate it also supports archives. Harmonize the widths of those buttons. Improve label on CD select image file. * Added CMakePresets.json * enhancement: use string.empty() instead of comparing with "" * bugfix: When using KMSDRM, controller input was lost after entering the GUI once (fixes #1384) This seems to be caused by SDL2 when using KMSDRM only, as it works properly under X. As a workaround, I'm changing the logic to keep only a single window and renderer when KMSDRM is detected. * Updated version date * bugfix: KMSDRM fixes for when using a single window (fixes #1411) Disable integer scaling when opening the GUI, trigger an update of scaling options when resuming * enhancement: separate data from home directories, improve handling (fixes #1073) - Separate DATA from HOME directories - Use special ENV variables to override locations: AMIBERRY_DATA_DIR and AMIBERRY_HOME_DIR respectively - Use the standard Linux locations if the env variables are not set: /usr/share/amiberry/data for the DATA folder, and $HOME/.amiberry for the HOME folder - Fallback to the Portable mode used before, if all the above fail * enhancement: Fix new paths logic for Flatpak also (#1073) * enhancement: improve look of GUI Misc panel scrollbar (fixes #1412) * 1376 add a3000 scsi controller and tape support for amiga unix amix (#1414) * enhancement: added CD and Tape drive buttons Added new buttons in the GUI: HDD panel, for handling CD and Tape drives. * Refactoring of GUI HD Panel - Added TapeDrive option - Refactored GUI panels to look more like WinUAE - Refactored functionality to closer match that of WinUAE * bugfix: Fixes for HDD panel windows * bugfix: Fixed default paths in dialogs * enhancement: Harmonize TapeDrive dialog logic with others Set the default path (harddrives) if no path is configured, before opening the dialog * enhancement: Updated Create HF dialog - The dialog will now only create HDFs, not add them as well * enhancement: fix navigation in Create HDF dialog * enhancement: Added Special Monitors support * enhancement: Added special monitors support in GUI * enhancement: fixed navigation for Tape Drive and other HDD panel dialogs * enhancement: improvements in the Add HardDrive dialog Added controller, unit, type and feature level dropdowns * enhancement: fixed navigation in Panel Harddrive * disable button that's not implemented yet * Updated version date * ci: Remove Manjaro RPI4 target from releases Only AmiKit was using that, and it moved to Debian12 with the new update * refactor: Move "abr" under plugins directory - Handle "abr" as one of the plugins, instead of having it living on a directory in the home dir. - Moved directory, updated paths and macOS scripts accordingly. - get_plugin_path() now returns the actual plugins path * enhancement: expanded DBus support with new events New DBus events added: - Insert Floppy - Insert CD * enhancement: added one more native lib path Added /usr/lib/amiberry as one more native lib search path * enhancement: Added option to use JST instead of WHDLoad binary In case someone needs this, there is now an option to use JST instead of WHDLoad for the Booter. use_jst_instead_of_whd (boolean) The option is saved in the global configuration of amiberry.conf and when enabled, will switch out the startup-sequence references from WHDLoad to JST, including any differences in options. The old AMOS-based booter is not changed with this. * enhancement: add filetype associations under macOS Handle .uae, .adf and .lha files with Amiberry, under macOS * bugfix: Fixed various glitches in SelectFile dialog (fixes #1417) - Do not attempt to scroll if the items are less than 32, as they fit in the screen already - Show the selected item in the text field - Allow existing files to be selected and used, even when the "createNew" flag is set - Fixed path building, it would always append to the existing string - Fixed string assignment in changeDir * enhancement: always show horizontal scrollbar in Selectfile dialog (fixes #1415) * build(deps): bump actions/download-artifact in /.github/workflows (#1419) Bumps [actions/download-artifact](https://github.com/actions/download-artifact) from 4.1.2 to 4.1.7. - [Release notes](https://github.com/actions/download-artifact/releases) - [Commits](actions/download-artifact@v4.1.2...v4.1.7) --- updated-dependencies: - dependency-name: actions/download-artifact dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * enhancement: upgrade libguisan to latest version, fix various bugs Guisan has been getting several updates lately, improving and fixing various bugs. Upgrading to the current version, but keeping my own improvements until I've had time to push them upstream * bugfix: Listboxes would not resize in width (fixes #1415) * enhancement: Guisan: Window widget needs to draw frame in inverted colors * bugfix: Fixed guisan theme colors (fixes #1420) * bugfix: Some Textboxes should not have a frame drawn * enhancement: Guisan: improve widget help text * bugfix: Fixed capitalization in RAM panel (fixes #1422) * enhancement: apply scaling factor to emulation window also (fixes #1421) * enhancement: Guisan updates Added Text class, refactored textbox, textfield etc. * bugfix: Fixed slider background color after guisan updates * bugfix: Fix CPU speed slider scale, when Fastest Possible was selected * enhancement: added .bin, .cdtv and .cd32 as valid rom extensions * bugfix: fixed dropdown and checkbox colors when disabled * bugfix: text widget should clear previous rows when setting context * bugfix: macOS Bundle name should be Amiberry * cleanup: remove CMakeSettings.json (obsolete) This is no longer needed, as we're using CMakePresets.json instead * enhancement: Check if the GUI theme font has the full path also - If the gui_theme.font_name contains the full path to the font, open it from there directly. This means we can now use system fonts as well, from /usr/share/fonts/... - If the filename is not found, then try to open it from the data directory, as before * ci: added Debian files * build: added CMake install section * build: change paths for cmake install * build: added install targets for guisan and floppybridge * build: ensure external libs are built before binary under CMake * build: added CPack support * chore: guisan: harmonize all files to use spaces and same line endings * enhancement: guisan: radiobutton and togglebutton fixes These widgets didn't use the proper disabled color * enhancement: guisan: fix remaining widgets colors These widgets were also missing their disabled color * enhancement: guisan: clang tidy, use const when possible * bugfix: Fixed navigation in Sound panel It wasn't possible to navigate into the Sound panel, since the first control (dropdown at the top) was disabled by default. * enhancement: add support for XDG_CONFIG_HOME (fixes #1425) Added logic to pick up the config directory, from the first match of the following: - $AMIBERY_CONFIG_DIR - $XDG_CONFIG_HOME/amiberry - $HOME/.config/amiberry - $HOME/.amiberry/conf if none of the above is found, the fallback is the current logic (portable mode, <current dir>/conf) * enhancement: Gracefully handle iconv failures If iconv_open() fails for some reason, do not fail completely but instead try to copy the entries verbatim. * bugfix: Refresh RTG panel when CPU changes are modified - The 24-bit addressing option blocks RTG cards from being used. When it's disabled, the RTG panel should refresh, to enable any relevant widgets - Fixed navigation in RTG panel (the VRAM slider could not be accessed) * chore: remove unneeded ifdef * Big bang winuae merge (#1428) * enhancement: Merge more WinUAE code - Added PPC - Added PCEM - Added PCI - Added MAME (A2410 and TM34010) - Added QEMU (various gfx cards, network cards, SCSI, etc) - Enabled relevant ifdefs * enhancement: Added luascript, slirp_uae - Removed unused ersatz - Added luascript (disabled for now, needs an extra library) - Added slirp_uae (disabled for now) - Added sources for separate tool "uaeunp" * enhancement: Added DSP emulation - Added DSP emulation files from WinUAE, with slight modifications (disabled for now) * enhancement: Enabled TabletLibrary, DSP, Draco Not implemented on the GUI yet, but enabled related ifdefs in the code * enhancement: Added SANA2 support Not visible on the GUI yet * enhancement: sync includes and other details with WinUAE * enhancement: added support for more RTG boards * bugfix: added missing arm/arm64 headers for pcem codegen_backend WinUAE didn't have these, so let's grab them from PCem v16 directly * enhancement: Fixed defines in picasso96.cpp * bugfix: Maybe we should actually sleep for a few ms in sleep_millis2 * enhancement: P96 - added ifdefs where Amiberry-specific pieces are added * chore: sync time/rpt headers with WinUAE * chore: reduce the lines of ifdefs in ahi_v2 * enhancement: use SDL_Delay instead of waiting for the cpu_wakeup_event * perf: only call init_row_map if the pixels have changed * enhancement: Added GVP T-Rex II Merged from WinUAE * enhancement: Added DKB WildFire v1.0 and Masoboshi v2.197 Merged from WinUAE * enhancement: Flush trap queues during reset. * bugfix: Added missing S3 Virge PCI byteswap mode Merged from WinUAE * perf: Do not unnecessarily call SetCursor() continuously. * bugfix: Fixed uae_sem_unpost This should be SemPost, not SemWait * enhancement: Allow 16-bit only if D3D9 is selected. Merged from WinUAE - Amiberry uses gfx_api value 2 anyway. * bugfix: Fix SPRxPOS modification just before horizontal match if sprite had lowest horizontal bit set. Merged from WinUAE * enhancement: Both COPxJMP strobe emulation update Merged from WinUAE * enhancement: Serial emulation updates, more accurate transmit timing, receive timing is not anymore instant except in direct mode. Merged from WinUAE * enhancement: OpalVision undocumented horizontal panning emulation address load was ignored if it was included in second control line. * perf: Reduce amount of read_processor_time() calls. Merged from WinUAE * build: updated makefile with new files * enhancement: added PPC options to GUI * enhancement: added MapROM option in GUI * bugfix: GUI RAM Panel did not highlight selected items properly * enhancement: added label in Input panel for Parallel Port joystick adapter The label should make it more clear that the last two ports are for the emulated parallel port joystick adapter * enhancement: GUI listbox inactive color improvement The guisan listbox would use an inactive color of a fixed value. It should use something from the color scheme, instead. * enhancement: added more models and configs in Quickstart * build: set the compiler standard higher in the Makefile * bugfix: fix ambiguous variable names * ci: avoid using named initializers in file, otherwise we run into a GCC bug This code won't compile with older GCC versions, due to a known bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=55227 * ci: fixed voodo file as well (see previous commit reason) * ci: fixed one more file with named initializers * ci: upgrade the minimum CMake version for capsimage Get rid of a warning about older CMake versions being deprecated soon * ci: change SDL_ttf include in sdltruetypefont.hpp * ci: updated more SDL includes in guisan * ci: Cmake builds would fail if started under a subdir (#1430) * ci: Cmake builds would fail if started under a subdir * ci: add cmake files needed for bullseye * ci: updated macOS include directories * ci: fix compiler error under clang/macOS * ci: remove unused linker flag * ci: add cmake in macOS homebrew * chore: fix compiler warnings in calc Well, most of them anyway * Revert "build: set the compiler standard higher in the Makefile" This reverts commit e822fe8. * bugfix: Editing entries in HD GUI panel was broken * ci: ensure Apple ARM builds don't detect both arm32 and arm64 defines * ci: more Cmake files for MacOS * ci: macos fixes, guisan should be static * ci: cleanup makefile macos targets * ci: fix macos apple silicon flags * ci: fix cmake linker flags for macos * ci: more macos Makefile fixes * ci: MacOS fix - remove rt lib from linker in Makefile * ci: MacOS fixes - Remove one more lib from the linker in the Makefile * ci: MacOS Makefile fixes: link to iconv * ci: Fixed typo * chore: make it clear that these flags are for ARM builds only * ci: Remove cmake and wget from MacOS homebrew runner these are already installed in the runner environment anyway * ci: remove autoconf from macos homebrew runner as well * ci: Fix capsimg filename for MacOS * Update make-bundle.sh * Remove unnecessary architecture detection * First imports of toolchain files for cross-compiling * Yet another fix for CMake that will help both OSX and Debian Bullseye * Cmake packaging (#1431) * modify CMakeLists.txt to be able to build packages for MacOS and Linux * Moved pull request template to .github dir * Moved code of conduct file under .github directory * Removed gitmodules file * Moved install extra dirs to only if we're not building an Apple bundle * Added first version of a Deb package option to CPack * Fixed double directory creation during installation/packaging * Remove debian directory We won't be using this, going for CPack instead * Fixed package version * Clean up contents of packages * Fix dependencies and filename of .deb package * Bump version to 6.3.5 * Add desktop file for Deb package * Remove archive generator, the DEB one is good enough * Change pipeline to use CMake for Linux targets Let's generate some .deb files! This is most likely missing stuff, like the distro name... * Fix incorrect architecture names in cross-compiler * Fix 32-bit arm package architecture * ci: Allow 32-bit arm builds * ci: disable the bullseye builds for now Until we figure out a solution with Bullseye + CMake + SDL2_image/SDL2_ttf * ci: fix expected targets * ci: bring back bullseye builds * Remove path from desktop shortcut * Fixed typo in comment * Set the version in the Project, use that for CPack * Updated compiled and linker flags with no-pie properly - There were some differences between the Makefile flags and CMake. Now they are identical - Updated the Findmpg123.cmake file with the latest one from upstream * Include bullseye build targets again * Increase GFX MAX_RTG_Boards to 4 as it looks like Picasso96 code is expecting 4 not the 1 listed. (#1434) --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: Alexander Wilms <Alexander-Wilms@users.noreply.github.com> Co-authored-by: Adam Sampson <ats-github@offog.org> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Solskogen <christer.solskogen@gmail.com> Co-authored-by: BluewizardNet <ryan@bluewizard.net> Co-authored-by: Dimitris Panokostas <dimitris.panokostas@norce.io>
System: Raspberry Pi 4B. Problem exists with amiberry on RaspiOS 64 Bit and RetroPie.
Hello Dimitris,
first of all, i thank you for your consistently reliable work for the excellent amiberry project! 👍
I have the following problem since amiberry v5.7.2. I am currently using amiberry v5.7.3.
I use a 4:3 monitor. In amiberry I set a resolution of 800x600.
Since amiberry v5.7.2. I have the problem as soon as I go into the Amiberry menu using the F12 key that the screen keeps "wobbling" there, as if there were difficulties fitting the screen into the 800x600 resolution. I have also experienced under retropie that sometimes F12 led to amiberry crashing immediately. But when you are in the menu with F12, the screen keeps wobbling.
I am attaching my configuration file. Please test with a 4:3 monitor in 800x600.
conf.zip
The text was updated successfully, but these errors were encountered: