Permalink
Commits on Aug 14, 2018
Commits on Aug 12, 2018
  1. Merge pull request #1115 from ajshell1/patch-1

    m9710797 committed Aug 12, 2018
    Typo fix
Commits on Aug 11, 2018
  1. Typo fix

    ajshell1 committed Aug 11, 2018
    If you go to http://tabajara-labs.blogspot.com/2018/06/esmiuncando-o-msx-kibe-ax-230-da-al.html, it clearly says that his name is Alexandre Souza" with an "A" at the end instead of an "O".
Commits on Jul 28, 2018
  1. fix Wformat-truncation warnings in gcc-8

    m9710797 committed Jul 28, 2018
    At two places in the code we print a formatted date
      YYYY-MM-DD
    As long as the year is less than 9999 this will always fit in 10+1 characters
    (+1 for zero-terminator). But of course the compiler doesn't understand
    calendars and warns that the output might get truncated (e.g. because the year
    is -123456 or because it doesn't know a month is at most 2 digits).
    
    Theoretically the compiler is right, but in practice there's no problem. But
    the fix is easy enough: just enlarge the buffer (and has no performance impact
    at all: it's a local buffer on the stack).
    
    After this patch, a debug/devel build on linux with gcc-8 is warning-free again.
  2. Implement YMF262 mix levels

    m9710797 committed Jul 21, 2018
    That is OPL4 register 0xF8. This controls the attenuation factors for the
    left/right audio channel for the whole FM output. (Register 0xF9 was already
    implemented, mix levels for the wave part).
    
    Internally implemented via the 'setSoftwareVolume()' mechanism (but now
    extended to allow a different factor for left and right). This saves 2
    multiplications per sample compared to implementing it in the
    generateChannels() method (as was done before for the wave part).
    
    I verified that before/after this patch, with the default settings for
    registers 0xf8/0xf9 the output volume remained the same. (I'm not 100% sure
    the balance between wave and FM is the same as on the real chip).
  3. integrate OPL4 improvements made by Valley Bell

    m9710797 committed Jul 21, 2018
    Thanks a lot Valley Bell!
    
    I reviewed all changes, and they all look fine. Though I cannot be 100% sure
    there won't be any regressions. So please test if you can.
    
    One thing that did change (unintentionally according to Valley Bell) is the
    content of the am_depth[] array. Though it's not clear that this is good or bad
    (the corresponding change in the dl_tab[] array _is_ intended. We've fixed it
    as good as we can for now, but needs more investigation.
Commits on Jul 24, 2018
  1. Initial config of Yamaha CX11 Music Computer

    MBilderbeek committed Jul 24, 2018
    Thanks to Alexandre Souza!
Commits on Jul 17, 2018
  1. First update of documentation for Android build.

    MBilderbeek committed Jul 17, 2018
    TODO: list all modifications needed on commandergenius clone to make it work
    with openMSX....
  2. Fix path for sdl-config when the SDL port path is absolute

    mthuurne committed Jul 17, 2018
    I fixed the path for freetype-config as well, just in case, even though
    that will always be a relative path currently.
  3. Explicitly link in liblog, for older Android versions.

    MBilderbeek committed Jul 17, 2018
    With this, the app works again on my Android 5.0.1 phone.
  4. Build appdata.zip from bindist/install/share content

    MBilderbeek committed Jul 17, 2018
    The content of this dir is explicitly put there for distribution, so it's only
    logical to use it also for the Android distribution.
    Also removed the check whether it is still valid. Rebuilding it is a matter of
    seconds, so just always create it.
    Lastly, do not copy C-BIOS, as it's already there.
Commits on Jul 16, 2018
  1. Add workaround that makes probe link test pass on Android

    mthuurne committed Jul 16, 2018
    It will even pass when the probe should fail, but that's the best I
    can do for now; when we migrate to SDL2 this should be cleaned up.
  2. Revert "Improve probe for libraries that try to intercept main()"

    mthuurne committed Jul 16, 2018
    This reverts commit c75bb04.
    
    I tried to be clever and it was punished for it: this trick broke the
    mingw build.
  3. Override freetype-config and sdl-config harder

    mthuurne committed Jul 16, 2018
    SDL_ttf's configure script is amazingly stubborn and will prefer flags
    from the host's pkg-config over the explicitly passed config scripts,
    even when doing a cross compile.
  4. Also show the menu button on Android.

    MBilderbeek committed Jul 16, 2018
    On newer devices, there is no MENU button visible that can show the OSD main
    menu, so let's try it with this for now.
  5. Update some defaults to get Android build more usable.

    MBilderbeek committed Jul 16, 2018
    For some reason, the updated Android build needs these changes.
    The AccessSdCard makes the app request permission to access files. Without
    this, it cannot even read the Tcl scripts.
    Immersive mode seems necessary on some devices to get the standard Android menu
    bar hidden, which cause a shift of touch locations.
  6. Explicitly tell the library link command to create a library

    mthuurne committed Jul 16, 2018
    I guess the "-shared" option ended up somewhere in LINK_FLAGS before,
    but not anymore after I cleaned those up.
  7. Add support for library as main executable in 'install' Make target

    mthuurne committed Jul 16, 2018
    This is necessary for Android now that we build 'staticbindist'.
  8. Improve probe for libraries that try to intercept main()

    mthuurne committed Jul 16, 2018
    Usually when main() is intercepted, it is provided by the library.
    However, this is not the case for Android, where the entry point
    for an application is in Java.
    
    Previously I tried linking the test executable as a shared library.
    While that does remove the errors, it also defeats the purpose of
    the probe, as missing symbols are no longer errors.
  9. Do not adopt commandergenius' Android compile flags

    mthuurne committed Jul 16, 2018
    We like to have fine control over the compile flags, rather than
    getting a huge collection of flags, most of which are not actually
    required.
  10. Pass the preferred C++ compiler to the openMSX build

    mthuurne committed Jul 16, 2018
    This allows use of a standalone toolchain, which works much smoother
    with autoconf-based packages.
  11. Build all Android libraries except SDL using 3rdparty

    mthuurne committed Jul 16, 2018
    SDL 1.2 for Android is an actual port with lots of changes, but the
    other libraries can build fine from the original source code with
    a few configure overrides.
  12. Pass exact locations of FreeType and SDL config scripts to SDL_ttf

    mthuurne committed Jul 16, 2018
    This is more predictable than passing the prefix where the configure
    script can go search for the config script. Also it makes it easier
    to support scripts in non-standard locations.
  13. Treat Android as Linux in 3rdparty lib build

    mthuurne committed Jul 16, 2018
    While there are some differences with traditional Linux systems,
    most libs build fine for Android with "linux" in the host tuple.
  14. Move omission of ALSA on non-Linux systems from Make to Python

    mthuurne committed Jul 16, 2018
    The idea is to have the build configuration in Python and the execution
    in Make.
    
    Note that the platform check switched from using the value of TRIPLE_OS
    to OPENMSX_TARGET_OS. This has no effect right now, since the only
    platform for which these two are different is OpenDingux, which has
    ALSA as a system library. But it will be relevant for the updated
    Android support.
  15. Remove support for DIR_IF_EXISTS in makeutils.py

    mthuurne committed Jul 16, 2018
    This construct hasn't been in the Makefiles for quite some time.
Commits on Jul 13, 2018