[FFmpeg-devel] [PATCH] avfilter/af_astats: count number of NaNs/Infs/denormals for floating-point audio too

[Alexander Strasser]

On 2019-04-25 18:49 +0200, Paul B Mahol wrote:
> On 4/23/19, Alexander Strasser <eclipse7 at gmx.net> wrote:
[...]
> >
> > On 2019-04-22 11:51 +0200, Paul B Mahol wrote:
> >> Signed-off-by: Paul B Mahol <onemda at gmail.com>
> >> ---
> >>  doc/filters.texi        |  6 +++
> >>  libavfilter/af_astats.c | 86 ++++++++++++++++++++++++++++++++++++++---
> >>  2 files changed, 86 insertions(+), 6 deletions(-)
> >
> > I was a bit surprised when I first saw the number of lines it
> > takes to add this feature. OTOH this is not a problem of this
> > patch, because it is mostly caused by the structure of the
> > code that was in place before.
> >
> > Changing the structure doesn't seem worth it yet. If
> > there will be an addition that is applicable to all the
> > individual stats, it should IMHO be reconsidered.
>
> What are you proposing to change?

Sorry if I wasn't clear enough, but as I stated above I am not
proposing to change things now. No objections against your
patch!

To clarify a bit what I meant with changing structure: There seem
to be many places where things are duplicated for each individual
statistic. Maybe things could be re-organized, so that it isn't
needed to repeat those for every of the 2 dozen statistics that
can be measured.

But that's easier said than done, and even if one tried to do
it, one could end up after a lot of work with a result that's not
good enough :(

That's why I said, maybe to reconsider it next time, when something
gets changed that touches all statistics (if that will ever happen),
because one would have to touch everything anyway and would probably
be in good position (detailed mental model of the entire code of that
filter) to try out structural changes.

Pardon me, I'm having problems to express this more precisely
at the moment. I hope the message got through.


  Alexander