Re: ffmpeg-related crashes, was: Re: [xine-devel] beta9 at the end of this week?

[Stephen T. <st...@sb...>]

On Thu, 2003-03-20 at 13:25, Heiko Schaefer wrote:

> maybe it would make sense just to leave the asserts in ffmpeg's code as
> they originally were ?
>=20
> Stephen: why did you change them ?!

An attempt at being a bit more informative than "assert(x!=3D0)".

> if there was a real reason to change them, something like a XINE_WARN or
> any sort of debugging output mechanism would probably be a better
> replacement.

How about this? Its the XINE_ASSERT code minus the abort(). I do not
know if on a warning it would be informative to give a backtrace.

/**
 * Provide assert like feature with better description of failure
 * Thanks to Mark Thomas
 */
#ifndef NDEBUG
# if __GNUC__
#  define XINE_WARNING(exp, desc, args...)                          \
  do {                                                              \
    if (!(exp)) {                                                   \
      printf("%s:%s:%d: assertion `%s' failed. " desc "\n\n",       \
             __FILE__, __XINE_FUNCTION__, __LINE__, #exp, ##args);  \
      xine_print_trace();                                           \
    }                                                               \
  } while(0)
# else /* not GNU C, assume we have a C99 compiler */
#  define XINE_WARNING(exp...)                                      \
  do {                                                              \
    if (!(exp)) {                                                   \
      printf("%s:%s:%d: assertion `%s' failed. ",                   \
             __FILE__, __XINE_FUNCTION__, __LINE__, #exp);          \
      printf(__VA_ARGS__);                                          \
      printf("\n\n");                                               \
      xine_print_trace();                                           \
    }                                                               \
  } while(0)
# endif
#else
# define XINE_WARNING(...) /**/

Stephen

--=20
Stephen Torri <st...@sb...>