. simplify the interface FrameBase to make it easier to implement it for different cases (remove HDC and make parameters more explicit)
. remove functions which are only called on a Win32Frame (in which case a cast is guaranteed to succeed)
. otherwise there is the risk that every FrameBase implementation wants to add its own variants.
. FrameBase::FrameRefreshStatus() simplify implementation: pass all flags explicitly
* Video / FrameBase: move arch specific code to FrameBase.
* Video::Initialize & SetFrameBuffer.
Ensure initialization and SetBuffer can only happen in the right order.
* Video: move virtual functions to FrameBase.
With these changes all the virtual functions are in FrameBase and Video gets closer to be (only) the Apple ][ Video device.
* Move a few more functions from Video to FrameBase (snapshot related)
Now, the inclusion is one way with Video *not* including FrameBase.
* FrameBase::VideoRefreshScreen move Video related code to Video.
And only leave management to FrameBase.
Here are a few changes to make 64-bit build possible.
The changes are mainly to use windows api in the correct way that works for both 32-bit and 64-bit builds.
* WinVideo: move some generic functions back to Video.
Introduce a new virtual function (VideoPresentScreen) to draw the video buffer to video memory.
* Move Win32Frame::VideoRedrawScreen() to FrameBase as it is generic.
* Fix GCC build.
1) do not fwd declare enums
2) include only base class header file IPropertySheet.
* DSInit(): make sure sound devices are cleaned before the 2nd call.
And use std::string to avoid need of memory management.
. class hierarchy: WinVideo IS_A Video (ie. WinVideo is a subclass of Video)
. GetVideo() singleton instance of WinVideo in AppleWin.cpp, exposed via Interface.h
. NB, the Microsoft SDK 7.1A doesn't contain dxguid.lib & ddraw.lib, so they are linked against these two dx libs that are already part of the AppleWin repo.
* Interface.h: ensure that functions in the interface are not exported by other header files.
This is generally harmless, except for the presence of default arguments, in which case the version with default arguments must come first.
To avoid the issue, these functions are only ever exported in the Interface.h header file.
This is not complete as header files from Windows/.. are still included in:
Keyboard.cpp
SerialComms.cpp
Joystick.cpp
But probably these are arch specific and will have to be completely reimplemented elsewhere.
. Do not include alarm.h as it is not needed.
The code in alarm.h is not complete: all the extern functions are missing and different linkers might complain about this.
. Do not redefine some windows.h types in VICE/types.h as it becomes harder to keep them consistent.
Biggest problem is DWORD_IS_LONG as the best way to define it in linux 64 bit is as an int. Prefer to use StdAfx.h.
1. Newer VC runtime stat(), when using 32-bit file size, returns error
if file size is >2G. For file existence check, using 64-bit file
size is more accurate as files bigger than 2G would be reported as
non-existing otherwise.
2. <sys/stat.h> is required for VS2013 or before.
. Split AppleWin -> Core with the functionality really needed by the emulator.
. Split AppleWin -> Utilities for generic code not called by other emulator components.
. Split AppleWin -> CmdLine for the command line option parsing.
Split MouseInterface to isolate DirectInput code.
Avoid overflow in platforms where RAND_MAX is a big integer constant.
Fix include files for gcc (where enums cannot be forward declared).
Fixed debugger main "data" window behaviour:
. When activating the main data display ("data" command) the cursor keys wouldn't work until the minidump ("md1") was also enabled. NB. The cursor keys should work in the main data window, independently of whether the minidump is active.
Whenever harddisks/disks are inserted (or removed) and *if path has changed* then:
. Then the internal save-state's path & filename will be updated to reflect the new defaults.
. LoadConfiguration(): Read the save-state pathname from Registry before harddisks/disks.
Also:
. CiderPress: only save pathname on OK.
. Refactored CPropertySheetHelper::BrowseToFile().
. Extended support for -d1,-d2,-h1, etc such that if the param is "", then it will eject/unplug the disk/harddisk.
And fix two general issues:
- confusing behavior when entering the builtin (mini) assembler mode,
- and also fixes the "cl" (clear flag) command - which was documented, but did not work at all (due to an "interesting" command/syntax conflict...).
Switched Mockingboard/6522 Timer interrupts & Mousecard's VBlank interrupt to use synchronous events.
This is a linked-list of ordered timer-based event, where only the head of the list needs updating after every opcode.
Video mode: "Color (RGB Card/Monitor)"
. RGB HGR rendering (no half-dot-shift, no vertical blend)
. RGB DHGR rendering
- Color 140-mode is a real 140x192 mode with no color fringe (Video7 patent, shows that way on real hardware)
- Mixed mode: correct transition for Color 140-mode <-> B&W (validated on a real IIc adapter.)
. Complete Féline support (eg. 2 distinct greys)
. Fixed Video7 SL7 inverse text
. Fixed DHIRES mixed mode detection
. NB. Prince of Persia will switch to B&W (560-mode) when going from HGR back to DHGR (same on real hardware).
Video mode: "Color (Composite Idealized)"
. Previously this was "Color (RGB Monitor)" (and before that, the AppleWin 1.25 "Color (Standard)")
. Now this mode does not support the extra RGB cards' video modes
Currently only Apple Color Card and Video7 SL7 are supported.
The RGB card can be defined through the command-line:
-rgb-card-type: apple (default), sl7, eve, feline
The default text color can be modified when SL7 is selected. On the actual card this was defined by dip switches.
-rgb-card-foreground <color>
can be 6 (blue), 9 (amber), 12 (green) and white (15)
Details:
. Added RGB FG/BG Text + mixed GR mode
. Added Color Text80
. Added Duochrome HGR
. RGB card defaults to Apple Color Card + B&W text
Removed 65d02.h and reconstructed this code using a combination of existing 6502.h/65c02.h and extra C Pre-Processor macros to include (or omit) the extra heatmap functionality.
We still end up with a normal 65c02 instance and also a debug 65c02 instance, but both will derive from the same 65c02.h file (+ same for the 6502 normal/debug instances).
Also:
. Added cpu_heatmap.inl for the built-in debugger's read/write operations.
. Support CpuRead/Write() from Z80 to hook the heatmap r/w.
. It's an Apple II plus clone.
. The 48K bank-switched Rom is controlled with AN0 and AN1.
. The character generation video Rom has two full character sets: English and German (F10 to switch).
NB. At the prompt, "TEST" runs the ROM self-test.
* Add missing Riff.h include and fix const correctness of arguments.
* Cosmetic improvements to Perf breakdown in LogPerfTimings().
* Logging - removed unnecessary cast.
. FT demo MAD2.DSK - original version now correctly crashes at start of full-screen scroller
. FT demo MADEF.DSK - graphics glitch on vertical edge where it transitions from GR to HGR mode
Removed very slow BitBlt() GDI calls in PrintGlyph() - now 18x faster in Release config!
. Replaced GDI call FillRect() with new FillBackground()
. Added missing bitmap declarations in GetDebuggerMemDC()
Fixed AdditionalIncludeDirectories for Debugger source files
. During init: only need to call IoHandlerCardsOut() when loading save-state
. MemInitialize() -> MemReset() -> ResetPaging() which always inits memmode
. Due to INTCXROM still being set and it not being excluded when in II+ mode.
. Fix is to reinit memmode on restart and only check INTCXROM if IIe or above.
. support relative paths (#663)
. updated the current directory with the path for each loaded image (#663) & saving state (#691)
. added a new switch -current-dir <path>
. Improved DEVICE_SELECT' I/O support (including Echo+).
. SSI263_Read() only for Phasor in Phasor mode.
. save-state: v6 (Phasor unit): phasor mode extended.
SSI263 (MB/Phasor) bug fix for when RESET/Power-cycle during phoneme playback.
Logging:
. improved for SSI263
. in CPU.cpp, added 'IRQ' / 'ISR-end'
. The auto-switch is done on each reset, or on inserting a disk at the start-up screen (MODE_LOGO).
. The Window's title include '(S6-13)' if DiskII card has 13-sector f/w.
. The debugger's 'disk info' cmd will show FW13 or FW16 depending on f/w.
This change adds the `-conf` switch which, if specified, will use an
INI-file instead of the Windows Registry.
This addresses AppleWin/AppleWin#709 and AppleWin/AppleWin#446.
Also:
- Allow creation of a blank (WOZ2) image
- multi-zip support extended to scan for the first valid image (useful for most woz-a-day zips which have at least 2 entries and were previously failing)
. Display MessageBox with all unsupported commands that were passed in.
. MessageBox yes/no response will either continue or exit AppleWin.
Changed -left-control-alt-buttons to -left-alt-control-buttons (for consistency with -right-alt-control-buttons) (#743)
. The MessageBox will now confirm whether you typed it correctly or not.
Fixed DHIRESON/OFF typo in comment!
. added GetRef(), which can throw a std::runtime_error
. NB. dynamic_cast for a reference will throw a std::bad_cast error on failure
Added try/catch around main VM restart loop
. split the loop into ProcessCmdLine(), GetAppleWinVersion(), OneTimeInitialization() RepeatInitialization() & Shutdown()
. catch std::runtime_error & std::exception
Support 2nd Disk][ in slot-5, via command line:
- -s5 diskii
- -s5d1 \<imagefile\>
- -s5d2 \<imagefile\>
NB. there's currently no Configuration UI support, except the Drive icons' tooltips show what's in slot-5 & slot-6 (for drive-n). So there's no way to eject the disks or insert new disks. The use-case I'm supporting it Wasteland which just has the 4 disks in the 4 drives.
Improved card management:
- Added `class Card` (in Card.h) which all other cards (that exist as classes) derive from (eg. LC,SSC,Mouse,Disk2).
- Added `class CardManager` (in CardManager.cpp\h) which now manages the 8 slots (and aux slot).
- Added `class Disk2CardManager` (in Disk2CardManager.cpp\h) which provides methods for operations that act on all Disk2 instances at the same time.
- Currently limited to just 1x SSC and 1x Mouse card (why would you need more?). This simplifies things, meaning there's no need to have dedicated SSCManager / MouseCardManager objects.
- Currently the 2nd Disk2 card can only be put into slot-5. This limitation is just due to the complexity of the Configuration UI. Having a more general drop-down per slot UI would remove this limitation.
. 6502 interrupt delays 1 opcode when interrupt occurs on last cycle of opcode (#724).
. Only 1-cycle delay for VF_TEXT & VF_MIXED mode changes (#656).
NB. Mad Effect 1 still has a bit of flicker on Space Invader (left edge)
ImageInfo is not a POD and cannot simply be initialised with ZeroMemory()
. the std::string constructor must be called.
. ImageInfo: simplify code using new / delete vs VirtualAlloc
Also fixed mismatching new / delete [] reported by valgrind
. Changed to 6522.TIMER underflowing at 0x0000 -> 0xFFFF (#652)
. Changed MB_Update() to be based on cycle delta (was TIMER1 interval)
. this improves support for small 6522.T1C interval
. removed MB_GetFramePeriod()
. removed overly-complex dual-timer support
. Replaced MB_EndOfVideoFrame() with MB_PeriodicUpdate()
. moved vert,horz onto same row
. made hex & apple modes the default display config
. added 2nd line showing cycles
. moved mini-mem views over by 1 char
. set m_headWindow = 0, after >50 bitcell gap
. bias fake bits to be ~30% chance of a 1 bit
Disk logging:
. added timestamps to I/O accesses
. DumpTrackWOZ() now outputs sync bits between nibbles
This change does two things:
1. Updates the registry APIs to reduce the likelihood of uninitialized
variables.
The code wasn't always checking the return value of registry load operations.
In some cases, this led to uninitialized memory being used, and crashes could
result. For example, LoadConfiguration in Applewin.cpp was using an
uninitialized value for the computer type if no registry variable for the
"Apple 2 type" was set.
New registry reading methods and macros have also been introduced, allowing
default value fallbacks for the cases where a registry variable is not found.
This makes registry access simpler and safer when a default value is known in
advance.
The registry code's style has also been updated to conform with the rest of
the code base (tabs instead of spaces, naming conventions, etc.)
2. Introduces string safety improvements.
A number of code paths have been modified to use safe-string functions instead
of their unsafe counterparts (e.g., strcpy, sprintf). In the process, some
strings were converted from "char" to "TCHAR". This was done mostly for
consistency with the rest of the code-base.
Supports:
- all "woz test images" v1.3 (WOZ1, WOZ2) are working, except 3.5"
- additionally: Frogger (spiradisc), Choplifter (not Enhanced //e!), Lode Runner, Marble Madness, Skyfox.
- woz images can be .gz or .zip compressed (ie. same as other supported images)
- save-state
Limitations:
- read-only, so WOZ images are forced to be write-protected
. as a result, games that need r/w images won't work (Stickybear Town Builder, Wizardry)
- 5.25" only (not 3.5")
- Added Configuration GUI to include checkbox for "50Hz"
- Implicitly use PAL or NTSC base 6502 clocks depending on video refresh rate
- Added new -50hz and -60hz command line switches
- Updated save-state for video refresh rate
1.28.8.0: Updated version & history.txt
Updates for DHGR MIX (#633):
. Support new switch -rgb-card-invert-bit7 to invert bit7 for Dragon Wars.
- Reverted DHGR MIX mode and AN2 off to invert bit7
. Support DHGR MIX mode and AN2 off to invert bit7 (undocumented)
. Improve the video-mode precondition to check for 80COL access occurring before $C05F
. Relax the video-mode precondition to just checking VF_MIXED
. In DHGR B&W mode, then HGR screen is also B&W
. For '50% scan lines', don't blend in NTSC B&W mode (as this was inconsistent with the RGB colour rendering), and DHGR MIX mode would look odd!
For the RGB hires look-up table:
- extended to include the previous video byte's high bit
- so it's now: {previous high bit + prev 2 video bits + next 2 video bits} & current byte
For all the RGB look-up tables:
- reduced from 512 to 256 lines (only 256 were being used, so it was just wasting space)
Refactored CopyMixedSource():
- fixed the Rainbow demo (#627)
- sped up in Debug config
Bumped version to 1.28.3.1.
Support the old AppleWin 1.25 vertical blending for hires:
- extended Config dialog to include 'Vertical Blend' checkbox
- Persist 'Video Styles' to Registry
- new cmd line options to select this style & also select 'RGB (Color Monitor)'
- code refactor to support enum VideoStyle_e (and replaced g_uHalfScanLines with a bit in g_eVideoStyles)
Bumped version to 1.28.2.0.
Supported modes selected via toggling AN3 and clocking in 80COL:
- 140 color mode (Apple calls this mode 2, Video7 calls this mode 0).
- mixed mode (Apple calls this mode 3, Video7 calls this mode 2).
- 560 mono mode (Apple calls this mode 1, Video7 calls this mode 3).
Save-state is also persists the extra state.
And there's a few corrections to APPLE2E.SYM for the 80STORE and 80COL I/O addresses.
Renamed this video mode to: "Color (RGB Monitor)"
Also renamed "Color (Monitor)" to "Color (NTSC Monitor)".
As for the colours: I've changed them from the original 1.25 colours. Instead I runtime-generate the colours from the NTSC code. See NTSC.cpp's GenerateBaseColors(). This shifts the same 4-bit pattern in, combining with NTSC color phase, until the colour stabilises. Then I average the next 4 RGB values to get the final colour. The reason for this is that we now have consistent colours between NTSC and this simplified rendering mode.
NB. The 2 greys (in GR,DGR,DHGR) are now the same RGB value.
Added new cmd-line switch: -videorom <file> to replace the video ROM for the Enhanced //e.
- Support video ROM sizes of 4K, 8K and 16K (top 8K only).
- NB. The rocker switch is set to European video ROM.
F10 (for //e or Enhanced //e models) emulates the PAL //e's rocker switch (under the keyboard) to toggle between European or US video ROM.
Other:
- Fixed debugger's view of the AltCharSet soft-switch (it was showing the opposite state).
* Refactor Language Card:
. MemSetPaging() now not used for $C080-C08F accesses.
- Instead LC registers its own I/O handler like all other slot 1-7 cards.
. Saturn uses base LC's 16K for bank0
* Move all 'modechanging = 0' to UpdatePaging()
Added support for #590:
* Changed Apple II original to default to no LC
* Added new cmd-line switch: -s0 <languagecard|lc>
* Changed so when LC is in slot-0, it now use Apple II+'s F8 (auto-start) ROM
Support for #408:
* added slot-0 LC and Saturn 128K for Apple II and II+
* added save-state support
- for slot-0 LC/Saturn, save the LC state in the slot-0 card, not the //e MMU; and switch to a new Apple2 unit ver2
* added g_Slot0, g_SlotAux
* added new LanguageCard.cpp/h to project and moved Saturn code into these new files.
* updated VS2013,2015,2017 projects
Support partial reads for sequential accesses of 6 or less cycles.
- And 31 or less cycles when in "Disk Access Speed" = "Authentic Speed"
Save-state support for partial read cycle (Disk][ unit v3)
Added boot-time logging for Phasor1.dsk (ProDOS 1.1.1)
. For the 14M video modes (DHGR,DGR,80COL), start rendering 1x 14M pixel early to account for these video modes being shifted right by 1 pixel
. Revert the display width back to 560 by disabling EXTEND_14M_VIDEO_BY_1_PIXEL
. reverted default so that ALT+TAB is not hooked (#556)
. reverted default so that ALT GR's fake LCONTROL is not hooked (#558)
. added new switches: -hook-alt-tab and -hook-altgr-control to support hooking these key combo's (#556)
. When Alt Gr is pressed, then manually post WM_CHAR message on receiving a WM_KEYDOWN (manually translate and account for shift/control/caps-lock)
. Hook filter: suppress Alt Gr's (ie. RMENU's) fake LCONTROL messages
Also:
. Hook filter: allow Ctrl+Shift+Esc (for Task Manager)
. Keyboard: refactor only use accessor functions to get the Alt/Control/Shift state
. Created a dedicated thread with message loop (#570)
HookFilter dll: Only call GetKeyState() if keycode is ESC (instead of every time)
DebuggerCursorUpdate(): reduce sleep from 10ms to 1ms
. When in MODE_LOGO, don't pass WM_CHAR to Apple II keyboard (now consistent with WM_KEYDOWN)
. Small refactor for Apple II keyboard's WM_KEYDOWN handler
Move code for log init & done to Log.cpp
Changed HGR rendering, such that that last video byte (at hpos=64) clears g_nLastColumnPixelNTSC so that the end-of-line off-screen pixel is zero (see #555).
. Fix for GetKeyState() not detecting the special keys as being up (so AKD was erroneously reporting the key still down).
. eg. Whilst pressing TAB, press LEFT ALT, then release TAB.
. Also suppress ALT+SPACE
. PostMessage to AppleWin message-pump for WM_KEYDOWN, WM_KEYUP for these special key combos
. Add special any-key-down (AKD) handling for these special key combos
. Start video rendering from screen position -1 (ie. 1 DHGR pixel offscreen)
. Allows 559th DHGR vertical column to be visible
. But now an HGR HCOLOR=2 vertical line at column 0 is only 1 DHGR pixel wide
Also:
. Fix TEXT80 to be aligned with DHGR
. Fix FLASH in TEXT80 (inverse aux bits needed to be masked with 0x7f)
. DCD state is now a mirror of DSR unless -dcd switch is used. (Fixes#553)
. Removed -dsr,-dtr switches, so -modem is now an alias for -dcd. (Also for #553)
. Wait for events (RLSD,DSR,CTS) in CommThread and call GetCommModemStatus() in CommThread.
. SSC: For TCP mode, support DSR, DCD & CTS status bits (#308).
. SSC: Now DTR must also be set to enable interrupts (in addition to the respective Tx/Rx interrupt bit).
. SSC: When reading the Status register, throttle calls to GetCommModemStatus() to a maximum of once every 8ms.
. Support TX IRQ for both TCP and COM modes (fixes#522)
. Support CTS/RTS (fixes#311)
. Fixed reading DIPSW registers
. Fixed TCP mode which was only generating an IRQ for first byte received
. Added UpdateCommandReg() to consolidate all updates to SSC command reg
. Updated reference URLs
So added a 2nd "Pure CPU MHz" for full-speed (ie. no video update), eg:
---------------------------
Benchmarks
---------------------------
Pure Video FPS: 2860 hires, 2868 text
Pure CPU MHz: 10.3 (video update)
Pure CPU MHz: 241.3 (full-speed)
EXPECTED AVERAGE VIDEO GAME
PERFORMANCE: 476 FPS
(ref: #424, #540)
Changed struct Disk_t:
. Added new struct Drive_t with "has-a" relationship to Disk_t
. Split properties of drive into Drive_t and properties of disk into Disk_t
* Fix the following errors reported by VS2017 Code Analysis.
applewin\source\applewin.cpp(761): warning C6385: Reading invalid data from '"Disk Image"': the readable size is '11' bytes, but '21' bytes may be read.
applewin\source\debugger\debug.cpp(6624): warning C6386: Buffer overrun while writing to 'sName': the writable size is '31' bytes, but '32' bytes might be written.
applewin\source\debugger\debugger_display.cpp(3715): warning C6053: The prior call to 'strncpy' might not zero-terminate string 'sText'.
applewin\source\log.cpp(42): warning C6053: The prior call to '_vsnprintf' might not zero-terminate string 'output'.
applewin\source\debugger\debug.cpp(2759): warning C6011: Dereferencing NULL pointer 'pFont'.
applewin\source\debugger\debugger_symbols.cpp(243): warning C6053: The prior call to '_tcsncpy' might not zero-terminate string 'pText'.
applewin\source\diskimagehelper.cpp(1132): warning C6053: The prior call to '_tcsncpy' might not zero-terminate string 'pszExt'.
applewin\source\diskimagehelper.cpp(1141): warning C6053: The prior call to '_tcsncpy' might not zero-terminate string 'szFilename'.
applewin\source\parallelprinter.cpp(242): warning C6053: The prior call to '_tcsncpy' might not zero-terminate string 'g_szPrintFilename'.
The one about RegSetValue(), according to Microsoft
https://msdn.microsoft.com/en-us/library/windows/desktop/ms724922(v=vs.85).aspx
the last argument cbData is ignored, so I set it to 0, as in some cases was anyway wrong (see "DiskImage").
Signed-off-by: Andrea Odetti <mariofutire@gmail.com>
* Use sizeof() rather than hardcoded value.
Fix one more case on non terminated string.
Signed-off-by: Andrea Odetti <mariofutire@gmail.com>
Changes to ease code compilation in GCC.
Disk_t and HDD contain a std::string and for this reason they need a proper constructor, ZeroMemory is not guaranteed to work.
. .dsk / .nib images
. ProDOS format / DOS 3.3 init
. authentic / enhanced disk access speed
For zero-length files, resize to the complete file size when first opened (#506)
. Support both .dsk and .nib
Created a new class FormatTrack to encapsulate the new track formatting logic
Improved precision of 'authentic' drive mode's spin emulation (#125)
Save-state: (save-state DiskII unit v2)
. support Format Track state
. save DiskLastCycle
DenibblizeTrack(): added some debug asserts and comments
Updated for VS2008/VS2013/2015/2017 projs & remove dependency on ddraw.lib for VS2013/2015
Updated disk logging:
. Moved all LOG_DISK_xx macros to new DiskLog.h (since shared by Disk.cpp and DiskFormatTrack.cpp)
. For write nibble: option to log cycle count & sync byte count
. For written track: option to log gap1/2/3 and track size
. For disk latch r/w: option to log when D5AA96 detected
Other:
. Debugger: Fix CD cmd to support absolute paths (#505)
* Support precise AY8910 sound-buffer filling when polling 6522 IFR.Timer1 (ACR=Free Running mode)
- same as if using 6522 IER.Timer1 interrupt
* Support 6522's Timer2 and some code refactoring
* Bump version: 1.26.3.4
. STA $CFFF will set memdirty[0xCF] flag
. The loop at the end of UpdatePaging() would copy a page of zeros over pCxRomInternal[0xF00] (ie. $CF00-CFFF)
