Le ven 01/10/2004 à 01:40, Bastien Nocera a écrit :
> Hello Matias,
>
> On Fri, 2004-10-01 at 01:07 +0200, Matias Feliciano wrote:
> > Le jeu 30/09/2004 à 09:33, Bastien Nocera a écrit :
> > > Hello Matias, Jeff,
> > >
> > > On Thu, 2004-09-30 at 05:50 +0200, Matias Feliciano wrote:
> > > > I don't know if it's a issue since Fedora (s-c-sc) doesn't support it
> > > > anymore.
> > >
> > > Support for multiple soundcards was never in the code,
> >
> > Are you sure ?
> > Don't understand.
>
> That I added this support in the run to FC3, and that it never really
> worked for OSS.
>
> > > and that it even
> > > unmutes the first card is a bit of magic, as there isn't any code to do
> > > that in there.
> > >
> >
> > What is that :
> > amixer_path = "/usr/bin/amixer"
> > amixer_pid = os.fork()
> > amixer_args = [amixer_path, "-q", "sset", "Master", "75%", "on"]
>
> Good point, missed that. I attached a patch for it unmuting the card for
> which we're trying to play a sound.
>
> > > > > so at least those 2 issues should be filed in 2 seperate reports.
> > > > >
> > > > > Also if you have enhanced the codebase
> > > >
> > > > No user visible enhancement (except bugfix).
>
> Bug fixing isn't ripping out most parts of the code. system-config-
> soundcard expects a good kudzu.
Since *you* tell me to not load modules, "my" new
system-config-soundcard don't use kudzu.
> By using only a "live" configuration,
> that means that we won't be able to configure cards for which a driver
> isn't loaded.
As i understand, it's a job for kudzu/initscripts. Remember that you
tell me to not load module. Is system-config-soundcard touching /proc ?
No.
So I don't understand your point.
btw, kudzu (and your system-config-soundcard) don't load the sound
drivers when detecting hardware. kudzu load alsa and no more.
If currently "Play test" work it's only because there is a bug in alsa.
that let /dev/snd/controlC0 even if there is no driver.
"your" system-config-soundcard does not load modules of the other card.
And for the first card, you are lucky.
This why in my previous version I "modprobe" all drivers.
>
> <snip>
> > > > And this is my following question for fedora-test-list :
> > > > Does system-config-soundcard have to continue to load all
> > > > modules whereas initscript/kmodule should do this job ?
> > >
> > > system-config-soundcard doesn't load the modules anymore,
> >
> > Fixed :
> > http://feliciano.matias.free.fr/system-config-soundcard/
> >
> > > they're auto-
> > > loaded when the drivers are needed.
> >
> > With udev ? No.
>
> They should be.
It's not autoloading module like we can do with a static /dev .
Rigth now, a module can be loaded even if we don't need it (read
rc.init).
btw, udev never load modules.
> File a bug with udev. It isn't the job of the config-
> tool to load drivers on demand.
>
> > > Please, follow Jeff's advices if you want those bugs to be fixed soon.
> >
> > Seems too later for FC3.
> > To be honest, i don't want to split my patch.
> > This is too much work and ugly.
>
> + # Ugly
>
> And I'm not too interested in merging a patch with "Ugly" or "Hack" in
> the comments.
The problem is the comment or the code ?
Sorry but your code is ugly and broken in many place even if you don't
have the honesty to state that in the comment.
cat system-config-soundcard-1.2.10/src/*py | wc -l
439
cat system-config-soundcard-1.2.10.orig/src/*py | wc -l
688
which one is ugly ?
>
> Furthermore:
> - self.apply()
> + self.mainWindow.destroy()
>
> Your code won't work in the context of firstboot.
Read the code.
apply() had been removed since it do nothing ("ugly" comment missing).
apply() is only call from closeClicked().
mainWindow.destroy() is call from apply() (move to closeClicked()) and
from testButtonClicked() (always there).
>
> > soundcardBackend.py is the more modified file. Check this :
> > wc -l system-config-soundcard-1.2.10.orig/src/soundcardBackend.py
> > system-config-soundcard-1.2.10/src/soundcardBackend.py
> > 298 system-config-soundcard-1.2.10.orig/src/soundcardBackend.py
> > 138 system-config-soundcard-1.2.10/src/soundcardBackend.py
> >
> > diff -uN system-config-soundcard-1.2.10.orig/src/soundcardBackend.py
> > system-config-soundcard-1.2.10/src/soundcardBackend.py | diffstat
> > soundcardBackend.py | 300
> > ++++++++++++----------------------------------------
> > 1 files changed, 70 insertions(+), 230 deletions(-)
> >
> > How can I build independant patch for each bug ?
> >
> > btw, is it too hard to verify a 140 lines file ?
>
> At 8 o'clock in the morning when I'm sipping my tea, yes.
> $ gnomevfs-cat http://feliciano.matias.free.fr/system-config-
> soundcard/system-config-soundcard-1.2.10-mat.patch | wc -l
> 645
>
> I'm sorry, but I still don't see what your patch is trying to achieve.
http://www.redhat.com/archives/fedora-test-list/2004-September/msg01634.html
This url is in the previous message. Read and you see how *your*
system-config-soundcard is broken.
> You're ripping out 2.4 kernel support (this code I might still need for
> RHEL3),
I request to do this here :
http://www.redhat.com/archives/fedora-devel-list/2004-September/msg01197.html
No reply.
I state that i removed all old code here :
http://www.redhat.com/archives/fedora-test-list/2004-September/msg01634.html
> removing kudzu usage and adding problems for people for whom the
> drivers aren't loaded
What you want ?
First, i should not load module.
Next i should.
My previous version that I drop a few minutes ago...(RRRrrr), load
module with :
- kudzu : like you
- "modrobe -c"
My previous version do exactly what initscripts (rc.sysinit) does !
Is your system-config-soundcard broken or initscripts ?
> (relying on the OS configuration, rather than the
> system's one).
Why not ?
With udev, all modules should be already loaded. If not, it's a bug.
>
> I'm interested in patches that would solve actual bugs, rather than
> rewrites and changes of directions made without asking the maintainer.
First, the code is GPL.
Right ?
Second :
http://www.redhat.com/archives/fedora-devel-list/2004-September/msg01197.html
And I have post many messages in fedora-test-list stating that sound in
Fedora is broken.
>
> Sorry, but I will not commit your patch.
It seems better.
>
> Cheers
>
> ---
> Bastien Nocera <hadess@xxxxxxxxxx>
>
signature.asc
Description: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e=2E?=
--
Fedora-config-list mailing list
Fedora-config-list@xxxxxxxxxx
http://www.redhat.com/mailman/listinfo/fedora-config-list
|
