Hi Jakub,
Thanks for your input. Please remember that alsa-driver isn't going away
any time soon.
Jakub Moc wrote:
- The in-kernel drivers seriously are not an equivalent alternative, let
alone the preferred one, for stuff like hda-intel or any similar drivers
that are under permanent heavy development, at least for now.
If hda-intel (or any other driver) from the kernel sources does not work
on your system then you should file a bug. Yes, there are drivers under
heavily development, this also applies to many other kernel subsystems
too. We live with it. It's not as bad as it sounds.
If you have a system which does not work with the in-kernel drivers,
please file a bug. Provided you cooperate, we will get it fixed in the
kernel.
- This is not a duplicated maintenance effort, it's simply needed to
have external alsa-drivers ebuilds, and it's needed to have them
supported as ALSA upstream won't accept bugs about in-kernel drivers.
That's not true. I have supported in-kernel ALSA drivers for a long time
and have never seen this be the case.
- The two are basically different branches, and it's *not* about whether
the code is newer or older in one or the other at all.
That depends on your definition of branch. From my definition of the
word, I disagree.
Using a more generic term: they are the same CODEBASE.
Changes from the alsa-driver upstream repository *are*
synced into the kernel very regularly - the large majority of the source
base is identical.
Taking the first 10 .c files under sound/pci, here are the diffs of the
latest alsa development release vs the latest kernel development release:
ad1889.c: no differences
als300.c: no differences
als4000.c: no differences
atiixp.c: no differences
atiixp_modem.c: no differences
azt3328.c: no differences
bt87x.c: 2 insertions(+), 1 deletion(-)
cmipci.c: no differences
cs4281.c: no differences
ens1370.c: 2 insertions(+), 6 deletions(-)
The bt87x differences are in both directions: the kernel adds support
for one hardware device over alsa-driver, alsa-driver adds support for
one device over the kernel.
The ens1370 difference is just a simple cleanup, no functional differences.
Now some drivers I know to be more popular:
via82cxx.c: no differences
intel8x0.c: 3 insertions(+), 7 deletions(-)
(just a tweak to the way the device is suspended)
emu10k1/*: 1 insertion(+), 3 deletions(-)
(just a cleanup, no functional changes)
The heavily developed monster one:
hda/*: 15 insertions(+), 88 deletions(-)
Interestingly in this case, the in-kernel driver is a touch newer than
the hda-intel one. It includes support for a few more hardware devices.
Again these are only very small differences though.
Both of these sources ARE the same codebase.
Daniel
--
gentoo-dev@gentoo.org mailing list
