Re: [Gnash-commit] gnash ChangeLog server/sprite_instance.cpp serv...

[zou lunkai]

explanation of current implementation:

bool
sprite_instance::get_frame_number(const as_value& frame_spec, size_t&
frameno) const
{
      //GNASH_REPORT_FUNCTION;

      as_environment* env = const_cast<as_environment*>(&m_as_environment);

       //becuase the "toString()" is expected to be invoked...
      as_value str(frame_spec.to_std_string(env));

      //alway convert the std_string to a number first. If this
conversion succeeded,
      //then we try to fetch the number-th frame. We do this even the
result number
      //is not a valid frame index.
      double num =  str.to_number(env);

      if ( isnan(num) || isinf(num))
      {
           return m_def->get_labeled_frame(frame_spec.to_string(env),
&frameno);
      }

      // TODO: are we sure we shouldn't check for frames labeled with negative
      //numbers ?
      // I think so, we are not going to look for a number-labeled frame.

      if ( num < 1 ) return false;

      // all frame numbers >= 0 are valid, but a valid frame number may still
      // reference a non-exist frame(eg. frameno > total_frames).
      frameno = num - 1;

      return true;
}


the drawback of the current implementation is that even if
get_frame_number() returns "true", we still need to check the
"frameno" here and there, as it might be still not in the valid
range.(eg, frameno > total_frames).  And it's also not good to clamp
the "frameno" to [0, total_frames-1] inside this function...