On 4/15/21 1:46 PM, Rémi Bernon wrote:
Signed-off-by: Rémi Bernon rbernon@codeweavers.com
programs/explorer/desktop.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/programs/explorer/desktop.c b/programs/explorer/desktop.c index 94d178880a8..4bf295ee325 100644 --- a/programs/explorer/desktop.c +++ b/programs/explorer/desktop.c @@ -39,7 +39,7 @@ WINE_DEFAULT_DEBUG_CHANNEL(explorer); #define DESKTOP_CLASS_ATOM ((LPCWSTR)MAKEINTATOM(32769)) #define DESKTOP_ALL_ACCESS 0x01ff
-static const WCHAR default_driver[] = {'m','a','c',',','x','1','1',0}; +static const WCHAR default_driver[] = L"mac,x11,null";
static BOOL using_root;
@@ -837,6 +837,7 @@ static HMODULE load_graphics_driver( const WCHAR *driver, const GUID *guid ) { TRACE( "display %s using null driver\n", debugstr_guid(guid) ); null_driver = TRUE;
wcscpy( libname, name ); break; }
This last patch makes it possible to use the null driver by starting the prefix with "DISPLAY= wine ...". It changes the behavior a bit, as it doesn't fallback to nodrv_CreateWindow anymore in that case.
However, nodrv_CreateWindow fallback is still used if the prefix (ie: explorer.exe) was started with an X11 display, and some other process is later started without a display.
I'm not sure in any case if that matters much or if it would be better to add the nodrv_CreateWindow warning to nulldrv?