[vlc-devel] [patch] add range for --verbose

[jnqnfe at gmail.com]

On Mon, 2018-03-19 at 18:27 +0200, Rémi Denis-Courmont wrote:
> Le perjantaina 16. maaliskuuta 2018, 21.22.49 EET jnqnfe at gmail.com a
> écrit :
> > On Sat, 2018-03-10 at 21:18 +0200, Rémi Denis-Courmont wrote:
> > > Le vendredi 9 mars 2018, 20:18:14 EET jnqnfe at gmail.com a écrit :
> > > > On Sat, 2018-03-03 at 16:56 +0200, Rémi Denis-Courmont wrote:
> > > > > Le perjantaina 2. maaliskuuta 2018, 18.56.23 EET jnqnfe at gmail
> > > > > .com
> > > > > 
> > > > >  a
> > > > > 
> > > > > écrit :
> > > > > > v2.
> > > > > > 
> > > > > > - Extended range (to arbritrary larger value of 8), leaving
> > > > > > a
> > > > > > little
> > > > > > extra room for extra plugin verbosity.
> > > > > > - Adjusted descriptions.
> > > > > 
> > > > > On second thought, an integer choice list might make more
> > > > > sense
> > > > > here.
> > > > 
> > > > Agreed, a choice list would work well here.
> > > > 
> > > > I have attached a revised patch which implements this. I
> > > > allowed
> > > > for
> > > > three 'debug extended' levels for plugins to make use of.
> > > > 
> > > > I also attached an additional possible patch for adding such
> > > > three
> > > > 'debug extended' levels to the actual message type enum.
> > > 
> > > Second patch does not apply for me :-(
> > 
> > Oh, sorry, I'd built them on my 'patches' collection branch, which
> > includes a patch to tidy up help output headings which hasn't been
> > merged (can't remember if I've even sent it in). I've moved all the
> > patches relating to the verbosity option to a clean branch, and
> > attached them here.
> 
> This does not seem right. We need a level for --quiet.

Sorry for the late response, things have been a bit hecktik.

I had looked into this a little a few weeks ago following your
response, but I could not see the problem you're suggesting is present
for --quiet.

I have found only two places where --quiet comes in to play, and both
look fine to me:
1) firstly in the console logger plugin, which actually owns this
option, where in the activation function it causes the module to fail
to load and thus supresses all log message output from it, as designed.
If this design is wrong/undersirable/to-be-improved, then that's a
separate matter.
2) in the vlc_init_avutil function of avcommon.h, which cheekily
follows this external module-specific option, where it causes
av_log_set_level to be called with AV_LOG_QUIET, which again seems to
be as designed, and unaffected by my changes.

or did you mean --<SOUT_CFG_PREFIX>quiet which comes in to play in the
"Open" activation function of x264.c, where it results in an internal
log level being set to X264_LOG_NONE, so again looking to me like
nothing is wrong.

If I am missing something, then please by all means explain it to me
and I'll create an additional patch :) Otherwise, if there's actually
no problem, can we merge this work?

Regards,
Lyndon