On Thu, 15 Nov 2007, Frans Pop wrote:

> On Thursday 15 November 2007, you wrote:
> > the new stack is very promising,
> > we will reconsider later if no eth1394 shows up,
> > for now that's just a minor regression.
>
> No, that is not a "minor" regression. Half the functionality of the old
> drivers is missing!

eth1394 is missing that is by far the less used firewire driver.

the switch to the juju stack allowed to close a huge nr. of bugs.
the old stack is just not supportable and never was, plus is a
security nightmare. ieee1394 in fact deserves the Kbuild variable BROKEN.
the new stack is aiming for feature parity, but with interface
incompatibility, so the various firewire userland needs to pick up
the switch. the new stack although more compact is already more stable.

> What discussion? I googled for it and after skimming through about 20 pages
> I still have not found it. What I _did_ find is several other reports of
> problems with the new stack, two of which complain about missing Ethernet
> support:
> - http://lists.debian.org/debian-kernel/2007/10/msg00414.html
> - http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=441206
> - http://www.debianhelp.org/node/9328

i consider that smallish to the *huge* usability improvement
on not having a strange firewire ethernet device to choose
for your network connectivity in d-i.
so no tears for it.

> So Google does not help. Let's check the changelog:
> linux-2.6 (2.6.22~rc5-1~experimental.1) experimental; urgency=low
> * Enable the new firewire stack labeled to be more simple and robust.
> -- Bastian Blank <waldi@debian.org> Tue, 19 Jun 2007 17:49:52 +0200
>
> Strange. No mention of the fact that the new stack is [1]:
> - experimental
> - not recommended for distributions by its upstream developers
> - has a _major_ regression in it's lack of Ethernet support
> - lacks userspace integration
> - has had only limited testing

if i'm wrong, we can still enable both and start blacklisting
the bad oldie one in m-i-t early enough.

RH did rewrite the stack and it's main upstream developer ships
it already enabled in 2 major releases fedora 7 and 8.
juju has already nicely improved since the 2.6.22 inclusion,
so still looks for the most promising option for the lenny release.

--
maks



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Hello,
On Tue, Nov 20, 2007 at 08:01:40PM +0100, maximilian attems wrote:
> eth1394 is missing that is by far the less used firewire driver.

But for some people the driver used early in the installation (where a
kernel rebuild is not very convenient).

> the switch to the juju stack allowed to close a huge nr. of bugs.
> the old stack is just not supportable and never was, plus is a
> security nightmare. ieee1394 in fact deserves the Kbuild variable BROKEN.

Well, it worked fine for me, except that the version from 2.4 could
not interoperate with the version from 2.6. The only minor oddity was
that sometimes the logs were flooded with certain warnings (I can dig
them out if necessary).

> i consider that smallish to the *huge* usability improvement
> on not having a strange firewire ethernet device to choose
> for your network connectivity in d-i.
> so no tears for it.

Well, I cannot really argue with you on technical merits, but then it
was one of the (long) supported options for d-i, so please *document*
it properly. I would've started out with an Etch image, if I had known
this regression.

> RH did rewrite the stack and it's main upstream developer ships
> it already enabled in 2 major releases fedora 7 and 8.
> juju has already nicely improved since the 2.6.22 inclusion,
> so still looks for the most promising option for the lenny release.

It would be great if also eth1394 would reappear in the new stack,
especially since the original developer is an @debian.org person.

Greetings

Helge

--
Dr. Helge Kreutzmann debian@helgefjell.de
Dipl.-Phys. http://www.helgefjell.de/debian.php
64bit GNU powered gpg signed mail preferred
Help keep free software "libre": http://www.ffii.de/

On Tue, Nov 20, 2007 at 08:32:44PM +0100, Helge Kreutzmann wrote:
>
> It would be great if also eth1394 would reappear in the new stack,
> especially since the original developer is an @debian.org person.
>

you seem to have a strange confidence in someone, whose stack went
so badly down the road, but yeah an upstream push for firewire
juju eth1394 is best for all.

--
maks



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

On Tue, Nov 20, 2007 at 08:01:40PM +0100, maximilian attems wrote:

> > What discussion? I googled for it and after skimming through about 20 pages
> > I still have not found it. What I _did_ find is several other reports of
> > problems with the new stack, two of which complain about missing Ethernet
> > support:
> > - http://lists.debian.org/debian-kernel/2007/10/msg00414.html
> > - http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=441206
> > - http://www.debianhelp.org/node/9328

> i consider that smallish to the *huge* usability improvement
> on not having a strange firewire ethernet device to choose
> for your network connectivity in d-i.
> so no tears for it.

The device was listed in d-i because people were using it. This thread is
here because they can no longer do so. It's rather rude of you to declare
that it's ok for the kernel team to "fix" usability issues in the installer
by removing functionality from the kernel.

I understand the technical arguments regarding the replacement of the stack,
and have no informed position on this bigger issue (though it seems that
plenty of people have been using the previous stack without complaint?), but
that doesn't make it ok that functionality has been dropped, whether or not
you happen to like that functionality.

--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
vorlon@debian.org http://www.debian.org/



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

On Tue, Nov 20, 2007 at 11:50:58AM -0800, Steve Langasek wrote:
>
> The device was listed in d-i because people were using it. This thread is
> here because they can no longer do so. It's rather rude of you to declare
> that it's ok for the kernel team to "fix" usability issues in the installer
> by removing functionality from the kernel.

as i mentioned on the mail announcing the switch eth1394 might go
missing for a while, if it doesn't reappear soon enough for lenny,
the m-i-t blacklisting of the old stack seems an easy way to get it back.

and yes i have seen quite a lot of people stumble on that dialogue.


> I understand the technical arguments regarding the replacement of the stack,
> and have no informed position on this bigger issue (though it seems that
> plenty of people have been using the previous stack without complaint?), but
> that doesn't make it ok that functionality has been dropped, whether or not
> you happen to like that functionality.

plenty of people have left *lots* of bug reports on ieee1394.
it is trivialy easy to crash any box as user with it.
it's buggyness was the reason for rh to sponsor the rewrite.
also it's sysfs device name support is quite lacking afair.

--
maks



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org