view libmpcodecs/vfcap.h @ 9177:01a713dcaf23

disable free() in string and string_list parsers. yes, it's a hack (and a little memleak), but i can explain :) [note it's just a few kB memleak, but it's the price of stability without full code review/audit - there are hunderds of possible double free()] the old config parser didn't free() strings/stringlists, but didn't even allocate them by default. the new one always free(), and it causes memcorruption/sig11 at cases like this: char* dvd_device="/dev/dvd"; {"dvd-device", &dvd_device, CONF_TYPE_STRING, 0, 0, 0, NULL}, since string constansts (allocated in .TEXT segment) cannot be free()'d

// VFCAP_* values: they are flags, returned by query_format():

// set, if the given colorspace is supported (with or without conversion)
#define VFCAP_CSP_SUPPORTED 0x1
// set, if the given colorspace is supported _without_ conversion
#define VFCAP_CSP_SUPPORTED_BY_HW 0x2
// set if the driver/filter can draw OSD
#define VFCAP_OSD 0x4
// set if the driver/filter can handle compressed SPU stream
#define VFCAP_SPU 0x8
// scaling up/down by hardware, or software:
#define VFCAP_HWSCALE_UP 0x10
#define VFCAP_HWSCALE_DOWN 0x20
#define VFCAP_SWSCALE 0x40
// driver/filter can do vertical flip (upside-down)
#define VFCAP_FLIP 0x80

// driver/hardware handles timing (blocking)
#define VFCAP_TIMER 0x100
// driver _always_ flip image upside-down (for ve_vfw)
#define VFCAP_FLIPPED 0x200
// vf filter: accepts stride (put_image)
// vo driver: has draw_slice() support for the given csp
#define VFCAP_ACCEPT_STRIDE 0x400
// filter does postprocessing (so you shouldn't scale/filter image before it)
#define VFCAP_POSTPROC 0x800