[FFmpeg-soc] [soc]: r2552 - mlp/mlpdec.c

[Diego Biurrun]

On Mon, Jun 23, 2008 at 11:33:19PM +0100, Ramiro Polla wrote:
> Diego Biurrun wrote:
> > On Sun, Jun 22, 2008 at 07:15:10PM +0100, Ramiro Polla wrote:
> >> ramiro wrote:
> >>> Log:
> >>> Copy IIR precision to FIR if only IIR is used, since the filtering code
> >>> always uses the FIR precision.
> >>>
> >>> --- mlp/mlpdec.c	(original)
> >>> +++ mlp/mlpdec.c	Sun Jun 22 20:02:56 2008
> >>> @@ -685,6 +685,11 @@ static int read_decoding_params(MLPDecod
> >>>              }
> >>> +            /* Both filters must have the same precision, so the filtering
> >>> +             * code always use the FIR precision. If only IIR is used, we copy
> >>                                ^^^
> >>
> >> Oops, typo. Since I'll have to fix this anyways, does anyone have a 
> >> better idea for the comment? I don't like the text the way it is now. 
> >> It's supposed to mean something like "To simplify the filtering code, 
> >> since both precisions are the same, only the FIR precision is used. If 
> >> only IIR filtering is used, copy its values to the FIR filter so it can 
> >> be used by the filtering code."
> > 
> > First off: If you want to have a comment on the text, don't cut it off.
> > I had to dig out the source file for this review.  That's cumbersome and
> > I cannot promise you that I will bother the next time around.
> 
> Sorry... But isn't the original svn log+commit good enough? It's just 
> one e-mail up. No need to go to the source.

I cannot tell from this email that the complete comment will be in the
previous email.  Besides, the previous email is in a different folder.
It's cumbersome.  All you needed to do is delete one line less from your
email.

> > BTW, all of this begs the question why they have two different precision
> > values to begin with ...
> 
> I'm not sure what you mean by this. Do you mean why the context has 
> place for reading the 2 precisions, why they may differ in the 
> bitstream, or why the bitstream has both values coded in, or... ?

There seem to be two precision values in the code, but apparently only
one is used.  I wonder why there are two in the first place.  No, I
haven't looked at the code, so this comment might welll be off the
mark...

Diego