On Sun, Aug 22, 1999 at 04:15:52AM -0700, M Carling wrote:
>=20
> I've been thinking about the huge ISDN patch for a few weeks now.=20
>=20
> I agree completely with Linus that huge patches should not be accepted at
> feature freeze time. As Linus has pointed out, they cannot be properly
> reviewed in the short time before the kernel is declared "stable". On the
> other hand, the work that the ISDN developers have done should make it
> into the tree. Asking them to submit small patches has been futile in the
> past and I don't see any reason to believe this will change.=20
Have you ever conducted a software project? Then you should try to stay in
sync with other developers, otherwise things go bad. Unfortunately (and for
reasons that have already been discussed) this has happened to the ISDN
project.
Now this has to be solved.
Somehow the current ISDN CVS code has to be sync'ed with the kernel.
Do you really think it would be best to split the diffs up into a lot of
small pieces? CVS could easily do this for you, but you won't be happy with
it ...=20
And you don't expect the ISDN developers to throw away their one year's work
and start from scratch, do you?
Your request is very valid, though, once the initial large patch has been
merged. You should not allow to loose sync again, then.
> I believe the best time to roll in the huge ISDN patch would be 2.5.1.
> This allows the most time for peer review before it makes it into a
> "stable" kernel. Users who need it sooner can continue to apply the patch
> themselves, as they do now.
I don't think so.
The patch has gotten some review by the developers. I agree that this is not
ideal, as it should also have gotten more review here.
But more importantly, it has gotten massive testing.
A major distribution, which has a large market share in Europe, where ISDN
is pretty common (much more common than in the US), has put the CVS ISDN
code into their distro since some time. This company also offers 60 days of
free installation support to all its customers. It would definitely not hav=
e=20
included it, if it had lead to more support requests.
To state it more clearly: The CVS ISDN code works much better then the one
included into the 2.2.x (x<11) kernels.
> Adding a huge patch now to 2.3 risks delaying 2.4. It also jeopardizes the
> stability of 2.4. Either of these suffice, IMO, to justify postponing a =
=20
> huge patch updating feature that is used by relatively few people (most of
> whom probably compile their own kernels) until 2.5.1.
I don't see any risk for a 2.4 delay. Those that don't use ISDN are not
affected at all. Those that do, will have a 90% possibility to get things
improved and a 10% possibility to have things broken. This does not differ
from other patches. And the changes required to help thos that got hings
broken are likely to happen with small patches in the 3(?) months left to
2.4 release.=20
BTW: Some small bugfixes since the large 2.2.11 merge were already
integrated into 2.2.12-pre. I really think, it will work from now on.
Waiting for 2.5.1 won't make it any better, BTW. The patch will be even
larger then.
Please note that the customers of the said company are not told to compile
their own kernel. They should not need to in order to not run into all sorts
of problems.
Regards,
--=20
Dipl.Phys. Kurt Garloff <kurt@garloff.de> [Wuppertal, FRG]
Plasma physics, high perf. computing [Linux-ix86,-axp, DUX]
PGP key: see mailheader / key servers [Linux SCSI driver: DC390]
--ZGiS0Q5IWpPtfppv
Content-Type: application/pgp-signature
-----BEGIN PGP SIGNATURE-----
Version: 2.6.3i
iQCVAwUBN8L/qhaQN/7O/JIVAQHs0QP+OJYtWwNG8E1WG8e9Qc2XOL8s1TjDq5BR
/MLfgq9JTnF5el5OHuV7AREL5cbXx3TdpmGpn9IOSeAw76oJpIwCRbKKmtPeujEC
E67GA5hGPjPtLMvZIt8YpqGUdfQH11MYx1vxKIM0rQQirfiG3tvHYmT8wj1KTB6d
QtuMGip4Ibo=
=94tG
-----END PGP SIGNATURE-----
--ZGiS0Q5IWpPtfppv--
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/