[FFmpeg-devel] SW's Patchsets Overview

[Nicolas George]

Marton Balint (HE12025-04-09):
> Not showing pointer addresses also has benefits, such as easier diffability
> of output, or better human readability. It depends on actual use case which
> is "useful", so a logging flag completely makes sense to me to show or hide
> it, depending on what the user wants.

Sure, it has second order benefits: it makes a few things a little
easier. For example if you want to remove the address to diff the output
of multiple runs, you can just pipe through sed; you are already piping
to get the diff anyway.

But the problems it causes, not being able to identify the source of a
warning, are first order.

> As for the question of the default behaviour, I don't have a strong opinion,
> both can make sense, maybe I would keep the existing behaviour for the
> library, but change the default for the cli tools to no-pointer output.

Even in the CLI tool, this change is harmful as soon as there are
multiple instances of the same component, like two scale filters.

Detecting that situation to enable the flag would not be worth the
effort and would clash with the principle of least surprise.

Stop with the sunk costs, just drop this misfeature.

Regards,

-- 
  Nicolas George