Approximately three or four times now, the issue of pointers being
in an inconsistent state been an issue in the video backend renderers
with regards to tripping up other developers.
Global (ugh) resources are put into a unique_ptr and will always have a
well-defined state of being - null or not null
It's so that the string in ControllerConfigDiag will match the string
in GameCubeConfigPane. Right now, it unnecessarily appears twice in
the list of strings to translate.
Commit 33487ab5f2 introduced a regression
where items would vanish from the toolbar. This adds a call to Realize()
after the reinsertions of the play/pause button as required per
documentation.
Thanks to Simonwayneee for noticing this!
Callers can now check whether reads fail, either by checking the return
value or by setting the buffer to a known bad value and seeing if it stays
untouched. I've added error checks to FileSystemGCWii and Boot_BS2Emu,
but not to Boot since it doesn't check any of its other reads either.
Includes cstring in EXI_DeviceMic.cpp to fix the undeclared function
errors for memset and memcpy when building with portaudio enabled and
pch disabled. Also adds the std:: prefix to those function calls
because there is no guarantee that they are put in the global namespace
when using cstring.
Thanks to David Brooke for noticing this!
This was relying on behaviour that GLExtensions was adding fake extensions to the supported list with ES.
This no longer happens so it needed to be changed.
This removes some nonsense in the extension loader where under an ES context we would still pull all function pointers and just continue onward if we
fail to pull one.
Now function pointers are only pulled if the version of GL or ES actually supports that function.
This fixes changing the play/pause button's label depending on the
emulation state. Before, wxToolBarToolBase's SetLabel() function was
used. This function, however, is not implemented in wxGTK which leads to
the label not changing on linux when the button is clicked. Although the preferred
method (according to the wxWidgets documentation) to change the properties
of a tool is to use the toolbar's setters, there is no such setter for
the label. Therefore, this implements a workaround where the
button is deleted and readded afterwards with the updated properties.
Thanks to linkmauve for noticing this!
This was due to specifying negative source coordinates for the texture copy, which must lie within the bounds of the source and destination textures.
The behavior now is to clamp the copy region to [0 <= size <= backbuffer size], resulting in a copy region that can be smaller than the backbuffer, but never larger.
Since ResolveSubresource cannot be used with depth textures (and throws an error with the debug layer enabled), use a shader which selects the minimum depth value from all samples.
Changes the sampler by XFBEncoder to use a linear filter, rather than point, to match GL behavior.
Rather than rely on the developer to do the right thing,
just make the default behavior safely deallocate resources.
If shared semantics are ever needed in the future, the
constructor that takes a unique_ptr for shared_ptr can
be used.
"-TR" Wiimotes don't accept output reports via the Control Channel. HidP_SetOutputReports will send the data via the Control Channel, whereas WriteFile will send it via the Interrupt Channel. Therefore using WriteFile enables "-TR" Wiimotes on Windows. There are some issues to be aware of. First the Toshiba Bluetooth Stack needs the output report buffer to have the size of the largest output report supported by the device. This requirement is also enforced by the Windows 7 default stack. However the Toshiba Stack, will only send the actual report bytes to the device, whereas on Windows 7 the full resized buffer is sent, resulting in an error on the Wiimote. This issue renders WriteFile unusable on Windows 7 with the default stack. On Windows 8/8.1/10 this requirement is somehow not implemented and it is possible to send smaller buffers via WriteFile to the device, enabling "-TR" Wiimotes.
Check is done by checking the driver provider property of the HID Class Driver. As the Toshiba Bluetooth Stack provides its own.
The initial enumerated device node is a empty hid interface node. Therefore first one node is moved up and then the provider property check is done.
The data passed in isn't modified in these functions
Also normalizes variables with prefixed underscores in the modified
functions (and normalizes outliers to our current coding style), as
single-underscore followed by any lowercased/uppercased character is
reserved for use in the global namespace (it's a common misconception this
is assumed to only be the case for underscores followed by a capital
character, but this is only the case in C, not C++).
It only marks a string for translation. It doesn't actually do anything
at runtime, so the message will always be displayed in English. Even if
we would've had a way to make the translation work, we shouldn't
translate this, because OSD doesn't support non-ASCII characters.
Caused by the recent merge 1c95cd5.
This should fix this panic message I saw when playing Super Mario Strikers:
Failed to compile pixel shader [...]: error C7011: implicit cast from "int" to "float"
The spec says that vendors can set the max texture size to be 65KB and we want 1MB.
Check the maximum supported and drop to the max if it is less than 1MB
This lets us sort by the underlying integers while only displaying the
icons. Currently, in both DolphinQt and DolphinQt2, we display both the
icon and the integer, but cut off the column width to not show the
integer. We also currently sort by the size's formatted string, not by
the size itself, which leads to "1 MB" sorting to less than "2 KB". This
commit fixes these issues.
In the future, we can use the filter methods here to allow for
searching for games.
m_need_prepare needs to be set before the Device thread is started.
Otherwise the thread blocks on IORead and the LEDs and Rumble is executed
after the user presses a button on the Wiimote.
Also the Prepare-Call on Refresh doesn't need to reset the Index, because it is
set once on the initial Connect/Prepare. Therefore the index assignment
was refactored.
fileplatform is moved so it's in the same place as the other platform
icons, and nobanner is moved just because it fits better in Resources.
Both of them were identical in all of Dolphin's themes.
Jules Blok