Activity for Strobe
-
Strobe committed [r46075] on Code
Add SDL equivalents of the ds3 & Logitech windows mappings. Add a gtk3 thec64 one too. Comment the Logitech one so it's easier to tweak to a users liking. Fix some whitespace & missing axes inconsistencies.
-
Update Xbox One joystick mapping now that I found a Bluetooth one. Remove L2/R2 trigger mapping as it conflicts if you try to use the GTK mapping in SDL and it wasn't necessary anyway.
-
Updated joymaps with new auto-saved info from r46046. Split ps3 one into 2 mappings, original Sony driver and DsHidMini based because they are completely different. Add comments about drivers used.
-
Add several joystick mappings for PS3 controller, Xbox One controller and Logitech Extreme 3D Pro USB joystick, created in Windows GTK3 Vice (trunk).
-
The "io" command (which is what you should be using btw) also does this if you do it enough times. btw this only seems to happen to VIA #2 at $1c00 not VIA #1 at $1800...
-
Partial fix committed in r46023, by removing the pointless success on save-as messages from the 3 cases where that was locking up vice. All error cases still to be resolved - listed in my 2026-03-04 comment above ( https://sourceforge.net/p/vice-emu/bugs/2205/#a611 ).
-
Partial fix for bug 2205, remove pointless "everything worked" messages from the three Save-As widgets that were locking up the GTK3-Vice GUI on Windows. A successful Save-As now no longer locks up Vice. Error case(s) still to be resolved, see ticket.
-
OK I did more testing trying to find a Save-As that didn't break and found the following: The following 3 are all broken the same way as described in this bug: Preferences > Settings > Host > Hotkeys > Save as Preferences > Settings > Input Devices > Keyboard > Save current keymap to file... Preferences > Settings > Input Devices > Joystick mappings > Mappings > Save As BUT.. If I force an error by e.g. trying to overwrite a read-only file, all the following are also broken: Preferences > Save settings...
-
OK I did more testing trying to find a Save-As that didn't break and found the following: The following 3 are all broken the same way as described in this bug: Preferences > Settings > Host > Hotkeys > Save as Preferences > Settings > Input Devices > Keyboard > Save current keymap to file... Preferences > Settings > Input Devices > Joystick mappings > Mappings > Save As BUT.. If I force an error by e.g. trying to overwrite a read-only file, all the following are also broken: Preferences > Save settings...
-
WFM fix to scale values appropriately for POT inputs which are expecting uint8_t
-
Confirmed with 46005 on win10. Weird thing is, the settings dialogue still functions, e.g. if you move the axis/button on the controller, the screen updates, highlighting whatever you just moved/pressed. Mouse also changes shape correctly as you move over e.g. text field. Emulated machine is still working fine in the background. But no keyboard or mouse input works, in the settings window, menus or into the emulated machine.
-
Make the 3rd button on a joystick actually visible on the status bar when you press it.
-
Update SDL MSYS2 instructions, replacing pacboy references with pacman. Also remove manual xa assembler steps as it's now a standard MSYS2 package.
-
Update MSYS2 instructions to only use pacman and remove pacboy references. Minor formatting improvements.
-
Monitor 'io' command was showing CIA2/dd00 twice and 'io dd00' was spitting out the CIA2 io dump twice.
-
Make the C128's MMU IO dump in the monitor ("io d500") significantly more readable.
-
Monitor fill, move commands affect both VDC and c128 RAM simultaneously
-
Confirmed fixed, thank you!
-
Monitor fill, move commands affect both VDC and c128 RAM simultaneously
-
Add MMU chip markings from Strobe's flat PAL c128.
-
Update with results from Strobe's flat pal 128 for tests 16, 18-21
-
Update NEWS with various stuff I poked at for 3.6, including one I forgot to add to the 3.5 notes.
-
Just tried one of the suggested examples, -moncommands with simply "m 1000" in a text file and trunk (r41107) hangs immediately using 100% of a core and has to be killed from task manager. A command that "waits" for the machine to be ready like the below works fine: autostart "myfile.prg" So maybe -moncommands should not be executed until the same conditions are met that autostart waits for? (which then makes it hard to debug things happening before then I suppose..)
-
Issue 1 is fixed in r40618 with some improvement to the VDC busy state emulation, cycles per line now varies slightly between runs as expected. Issue 3 would be resolved implementing R34/35 (horizontal blanking) support in the renderer. Last time I tried to do that the raster code got in the way by drawing the border when we didn't want it to, and raster is an undocumented pile of spaghetti code.
-
This was implemented in R40364 and the vdcdump testprog passes but unfortunately only with 64K VDC setup (which is the default). 16K VDC is still not quite right and needs further investigation e.g. it partially fails with the vdcdump testprog (only says OK on one test, it should be OK for all 4). I suspect the vdc.vdc_address_mask (which should be 0xffff or 0x3fff) is not always set correctly and/or is being ignored somewhere.
-
Fix whitespace>tab for fussy makefiles
-
Add missing x128 SDL symbolic keymap for US keyboard. Also fix typo.
-
Add emulated keyboard layout to c128 keymaps, fix some missing/outdated comments
-
GTK US symbolic keymap: fix '@', '*', ':' and '+' in x128 and also fix up-arrow in 128/64/dtv/scpu64/vic
-
Improve accuracy of VDC busy status by making ugly hacks uglier. This is the way.
-
Right Shift+Z differs from Left Shift+Z in GTK3
-
Think it's fixed in r40541, please test.
-
Fix bug 1519 shift-z behaviour for US symbolic, probably introduce other bugs.
-
SDL2 is fine. Keymaps look OK for both Z and "?". GTK3 only issue it seems.
-
Right Shift+Z differs from Left Shift+Z in GTK3
-
Add additional comments and missing documentation to x128 keymaps
-
Update testbench for vdcdump to use debug cart
-
Add debug cart support to vdcdump test and set border colour appropriately for pass/fail.
-
WFM with trunk on win10x64 (20h2). We'll need a more info on your setup to replicate.
-
Update SDL msys2 docs, fix some version issues and add notes on building SDL instead of SDL2
-
Finish fixing bug #919 - VDC memory mirroring. Now all VDC memory accesses go via translation functions. The vdcdump testprog mentioned in the ticket now passes, saying "OK" instead of "EMU".
-
Some work towards fixing VDC memory mirroring bug #919 - now use vdc_ram_read() for all drawing rather than directly accessing the VDC ram buffer. Also let that function handle the 16k/64k masking.
-
msys2 building docs - make sure pacboy is installed before trying to use it
-
Improve CRTC IO dump - dump internal registers, add more interlace info and align output a bit better
-
Implement VDC "interlace sync" mode now that our renderers can support interlace output. Thanks DQH!
-
The default bank is CPU :) Yes and no :) When you start vice the bank is set to cpu (not default), but if you literally set the bank to default with bank default it sets it to ram instead. IMO when you start vice the bank should be set to default which should be cpu as that is what most people would expect. At the moment it is not set to default, but it is set to cpu. Confusing as..
-
to elaborate a little more, the "m" command shows what the current bank is set to and the default bank is ram. So if you do "bank io" the "m" command will then show you the contents of the IO chips at dd00 instead. Note that the CPU $00/$01 port isn't considered IO for reasons someone else can explain :)
-
For now set VIC-II Render filter from "CRT emulation" to "Unfiltered" and then use Snip or a similar screen capture tool to grab the vice window. If you want closer to 1:1 rendering also disable double size and shrink the window to its minimum with Alt-R (or your mouse).
-
This is the case for all the emulators - the screenshots don't have an aspect ratio, they are the raw graphics buffer before it goes through any renderer (like CRT) or any scaling, so as a result always have square pixels, they are not intended to be WYSIWYG. It might be possible to set the DPI/PPI in the image (currently unset) so that it scales correctly when shown in a suitably modern image viewer. That way they could display with the correct aspect ratio but still work for the testbench for example...
-
Docs on configure options don't match actual configure behaviour
-
Update with required prerequisites for the binary releases/nightlies, making sure we install both 32 & 64 bit ones so we can build either, and mention 7z as a bindist option.
-
"make bindist7zip" as we have been releasing 7z binaries for a few years now rather than zip. Also fix a few typos.
-
Fix MMU io d500 monitor dump page pointers, which are 17 bits
-
Fix several issues with IO d500 (MMU) monitor dump on x128 - configuration and preconfiguration register output was wrong with backwards logic, it effectively always showed 8502 as the CPU and the RAM Configuration Register (RCR) had the wrong abbreviation.
-
update VDC news a bit
-
This was mostly fixed in r39272 Looking at windows icon specs, 128x128 is not required. Todos: 1. 64x64 (which we don't have) is required for classic mode. 2. The small sizes still need some manual pixeling to be legible.
-
x128 VDC Crash
-
fixed in r39298, please retest.
-
Segfault less, fixes bug #1104
-
no just attach it to this ticket (you should see 'Add attachments' just below where you add a comment)
-
C128 VDC emulation glitches
-
Plasma part issue fixed in r39293, should make it into 3.5.
-
Implement VDC internal double buffer, fixes plasma part glitches in RFOVDC as per bug #581
-
update VDC changes and other minor additions/spelling fixes
-
Fix size issues with x128 icon - make 'C128' text larger in SVG, regenerate PNGs in correct size and add missing 24x24 icon.
-
Fix interlace logic to handle both odd & even char row heights properly and remove no longer needed hacks for vdcmodemania mono interlace parts which now 'just work'
-
quick hack to fix vdcmodemania mono modes
-
Handle attribute offset seen in vdcmodemania fli part which now works
-
Fix long standing bug with handling wrap around of VDC ram
-
On windows the x128 icon is a different size to the others, but it's like that at every different icon view size in windows explorer, see attachment. So even though there are some missing sizes for the 128 icon, I don't think that's the only issue. I suspect there is some DPI setting difference between the SVG used to generate the x128 icon vs the others?
-
Fix compiler warnings when REG_DEBUG is defined
-
Better VDC FLI handling, 3 byte attribute offset doesn't always apply
-
Improve the prerequisites for msys2 builds, add note about building without pdf docs
-
wtf, can't replicate this! must be some combination of how it's launched or something. I will narrow it down and update...
-
x128 doesn't autostart from c64 mode, only loads
-
Fix VDC interlace modes with odd row heights, soci VDC test #5 displays again
-
Missing and/or corrupted icons; x128 msys2 gtk3 builds give Pango warning couldn't load font "EurostileLTStd-BoldEx2 Not-Rotated 175"...
-
don't get your resize logic stuck in a loop resizing the screen every frame.. (which I still don't think should make the screen flash in CRT renderer, but at least that glitch caught this one)
-
Mostly fixed for VDC in r38683. Aspect ratio is still wrong for VDC and VIC-II is still chopped off at the bottom for video (but not for still images e.g. PNG) For others that want to look at the VIC-II side, through trial & error I determined that the media size seems to be mostly determined from raster->geometry->first_displayed-line / last_displayed_line x64/x64sc recorded video is chopped off at the bottom. It's not obvious where the 272 vs 288 height difference is coming from.
-
Mostly fix bug #1307 - the screen size for the VDC media recording. The aspect ratio is still wrong and VIC-II still needs some work.
-
Video recording of VDC is far too big. VIC-II is too small
-
Video recording of VDC is far too big. VIC-II is too small
-
x128: wrong border size in VDC window if VIC-II model is NTSC
-
Should be fixed in r38642, please retest.
-
Resize the VDC canvas and the vsync height according to video mode, fixes bug #1198. Add glue for general auto-resizing in future.
-
Don't assume unused bits are zero, & include reg 5 in frame size / FPS calculations
-
Change the order of the mem_ioreg_list so "io" command in monitor spits out the IO space in address order instead of any which way but loose.
-
Show accurate FPS now that mon_out() handles float with r38531
-
Add to the VDC's IO monitor command full frame size with fps calculation plus other minor tweaks
-
Fix the height of the VDC screen shot. probably break other stuff.
-
Improvements to the monitor char/sprite display - use # over * for more contrast, show the memory value(s) alongside binary view and fix bug showing multiple sprites
-
Changing x128 machine model in GTK3 UI doesn't always work as expected
-
seems fixed, thanks dqh :)
-
Changing x128 machine model in GTK3 UI doesn't always work as expected
-
Corrected aspect ratio for VDC output.
-
Add another invalid display setting to display tests. Update some comments.
-
Size of VDC RAM is shown in register (which does not happen on real hw)
-
Applied in revision 37363. Thanks for the patch!
-
Apply (simplified) patch #211 by Marco Baye. VDC reg #28 bit #4 sets the addressing mode, it does not actually reflect the amount of ram installed.
-
No ruckeln on stretch vertically
1 >
