[Angstrom-distro-users] Lack of modules for current kernel

[Paul Sokolovsky]

Hello Aaron,

Friday, November 2, 2007, 11:10:53 PM, you wrote:

> Am I going to need to build the kernels I need by
> myself or are the modules for the different kernels in
> the pipeline?

  Well, let's first start with pronouncing that all this is not some
kind of magic mess, but pure technology based on rational logic. So,
everything which you see has a reason behind it, and everything you
request should have a reason too, and you should make it explicit,
or it's hard to give you any helpful answer.

> When the mentors (I am assuming they are
> the only ones who build the images) build the images
> why do they they not include all the modules?

  Well, the answer would be obvious - because *all* modules are *big*
size, and most people won't need 80% of the modules, so why waste
precious space?

> I have
> searched all over the feed browser and I can not find
> a g_ether kernel module for the 2.6.21-hh17 kernel.

  That's why I was talking about the reasons - why do you need that
module? Because there're multiple ways for you yourself to figure out
the situation with g_ether and yet you dump bunch of messages to the
list over a week. So, let's trace the reasoning:

1. Once you didn't find g_ether package in hx4700 feed, you would
immediately think that it is built into kernel, check the defconfig,
and see that it is true.
2. Alternatively, you would need g_ether not for its own purpose, but
to connect the PDA to a desktop machine. Then, you would try to do this
right away, and see that it works out of the box.

> Also as I understand it there is a 2.6.23 kernel for
> the zaurus machines? why are there so many different
> kernels modules for the different machines?

  This is called "Embedded Linux feudalism".

> I have
> read the FAQ and searched the wiki but there is very
> little information on the ipaq hx4700. Is this machine
> the 
> Angstrom red headed step child?

  Every machine is the same, and differences only emphasize the
similarities. So, there should not be "much" information about hx4700,
instead all machines should be made work the same, and this consistent
behavior should be documented. This is not easy aim to reach, given the
mentioned inherent feudalism of the area, but it proceeds in that
direction. As a user, you just should trust that it's all being done for
good, and don't look for problems where there're none of them, like in
g_ether case above.

[]

-- 
Best regards,
 Paul                            mailto:pmiscml at gmail.com