[FFmpeg-cvslog] r19901 - trunk/tests/vsynth.regression.ref

Reimar Döffinger Reimar.Doeffinger
Thu Sep 17 23:44:06 CEST 2009


On Thu, Sep 17, 2009 at 02:30:15PM -0700, Baptiste Coudurier wrote:
> On 09/17/2009 01:59 PM, Reimar D?ffinger wrote:
> >On Thu, Sep 17, 2009 at 01:06:23PM -0700, Baptiste Coudurier wrote:
> >>Huh, they are reliable, they just don't memset 0 aeras that are
> >>useless (reserved), and in the real world this cause absolutely no
> >>problems.
> >
> >In other words they are leaking random date, which e.g. in case of encoding
> >from a password-protected http server might include passwords or all kinds
> >of other sensitive data in case of more complex applications than FFmpeg.
> >
> >>They just don't produce bitexact output if memory is not initialized.
> >>That's the burden of many fixed frame size codecs.
> >
> >Well, IMHO this is the burden of crappy code.
> 
> I recognize you well here.

I guess we both know our opinions strongly diverge here.
I find code that behaves like this on purpose simple inexcusable even it
didn't mess with regression tests.



More information about the ffmpeg-cvslog mailing list