[FFmpeg-trac] #3895(undetermined:new): Multiple issues around the new function of -max_error_rate

FFmpeg trac at avcodec.org
Tue Aug 26 07:07:08 CEST 2014

#3895: Multiple issues around the new function of -max_error_rate
             Reporter:  JayBlanc     |                    Owner:
                 Type:  defect       |                   Status:  new
             Priority:  normal       |                Component:
              Version:  git-master   |  undetermined
             Keywords:               |               Resolution:
             Blocking:               |               Blocked By:
Analyzed by developer:  0            |  Reproduced by developer:  0

Comment (by JayBlanc):

 Unfortunately, the sample media I have that shows this are all multi-gig
 sized and licensed, and for obvious reasons I can't just send slices of
 the parts that trigger it. Use a sample that has recoverable errors, and
 enough that it triggers -max_error_rate. Prior to the creation of
 -max_error_rate, the program would only return an error value on fatal

 However, this is not a regression. It was a deliberate design change, but
 one that seems to have unintentionally changed default behaviour in a bad
 way. Now default behaviour is to return an error value after a certain
 number of 'errors' has occurred, and it is difficult to identify what
 errors are being counted and to what threshold they are being measured.
 This breaks compatibility with pre-existing scripts and embedded use of
 the CLI tool.

 Again, the requested work-around change is to make the default behaviour
 match the previous behaviour, and default -max_error_rate to 1. Which
 would restore compatibility for scripts and embedded uses of the CLI which
 do not expect -max_error_rate generated return values, and would only
 produce such return values when requested to do so.

Ticket URL: <https://trac.ffmpeg.org/ticket/3895#comment:3>
FFmpeg <https://ffmpeg.org>
FFmpeg issue tracker

More information about the FFmpeg-trac mailing list