1 |
Hi Jakub, |
2 |
|
3 |
Thanks for your input. Please remember that alsa-driver isn't going away |
4 |
any time soon. |
5 |
|
6 |
Jakub Moc wrote: |
7 |
> - The in-kernel drivers seriously are not an equivalent alternative, let |
8 |
> alone the preferred one, for stuff like hda-intel or any similar drivers |
9 |
> that are under permanent heavy development, at least for now. |
10 |
|
11 |
If hda-intel (or any other driver) from the kernel sources does not work |
12 |
on your system then you should file a bug. Yes, there are drivers under |
13 |
heavily development, this also applies to many other kernel subsystems |
14 |
too. We live with it. It's not as bad as it sounds. |
15 |
|
16 |
If you have a system which does not work with the in-kernel drivers, |
17 |
please file a bug. Provided you cooperate, we will get it fixed in the |
18 |
kernel. |
19 |
|
20 |
> - This is not a duplicated maintenance effort, it's simply needed to |
21 |
> have external alsa-drivers ebuilds, and it's needed to have them |
22 |
> supported as ALSA upstream won't accept bugs about in-kernel drivers. |
23 |
|
24 |
That's not true. I have supported in-kernel ALSA drivers for a long time |
25 |
and have never seen this be the case. |
26 |
|
27 |
> - The two are basically different branches, and it's *not* about whether |
28 |
> the code is newer or older in one or the other at all. |
29 |
|
30 |
That depends on your definition of branch. From my definition of the |
31 |
word, I disagree. |
32 |
|
33 |
Using a more generic term: they are the same CODEBASE. |
34 |
|
35 |
Changes from the alsa-driver upstream repository *are* |
36 |
synced into the kernel very regularly - the large majority of the source |
37 |
base is identical. |
38 |
|
39 |
Taking the first 10 .c files under sound/pci, here are the diffs of the |
40 |
latest alsa development release vs the latest kernel development release: |
41 |
|
42 |
ad1889.c: no differences |
43 |
als300.c: no differences |
44 |
als4000.c: no differences |
45 |
atiixp.c: no differences |
46 |
atiixp_modem.c: no differences |
47 |
azt3328.c: no differences |
48 |
bt87x.c: 2 insertions(+), 1 deletion(-) |
49 |
cmipci.c: no differences |
50 |
cs4281.c: no differences |
51 |
ens1370.c: 2 insertions(+), 6 deletions(-) |
52 |
|
53 |
The bt87x differences are in both directions: the kernel adds support |
54 |
for one hardware device over alsa-driver, alsa-driver adds support for |
55 |
one device over the kernel. |
56 |
|
57 |
The ens1370 difference is just a simple cleanup, no functional differences. |
58 |
|
59 |
Now some drivers I know to be more popular: |
60 |
|
61 |
via82cxx.c: no differences |
62 |
intel8x0.c: 3 insertions(+), 7 deletions(-) |
63 |
(just a tweak to the way the device is suspended) |
64 |
emu10k1/*: 1 insertion(+), 3 deletions(-) |
65 |
(just a cleanup, no functional changes) |
66 |
|
67 |
The heavily developed monster one: |
68 |
|
69 |
hda/*: 15 insertions(+), 88 deletions(-) |
70 |
|
71 |
Interestingly in this case, the in-kernel driver is a touch newer than |
72 |
the hda-intel one. It includes support for a few more hardware devices. |
73 |
Again these are only very small differences though. |
74 |
|
75 |
Both of these sources ARE the same codebase. |
76 |
|
77 |
Daniel |
78 |
-- |
79 |
gentoo-dev@g.o mailing list |