BuildingKernel

Bruce Dubbs bdubbs at swbell.net
Mon May 17 12:42:13 PDT 2004


Jeroen Coumans wrote:

> Bruce Dubbs said the following on 17-05-2004 18:11:
>
>>
>> This discussion should start with the the need for instructions on 
>> building the kernel at the start of Chapter 5.
>
>
> That's true, I didn't think of that in my proposal. Hmm, there goes my 
> elaborate scheme to unite the minimalist & functionalist approaches!
>
> (reads the host requirements page)
>
> Ok, those are some pretty steep requirements. Are they really 
> necessary for LFS-6.0? I see the following possible solutions:
>
> 1. maintain 2.6, NPTL, udev host requirements. Provide instructions on 
> how to upgrade the host. This is unacceptable at the moment. 

Not being familiar with the issues, could you please explain why its 
unacceptable.

> 2. provide alternative instructions for 2.6 hosts. Best of two worlds, 
> although could lead to user confusion & fragmentation. Also puts 
> additional burden on editors of having to maintain two sets of 
> instructions. 

I believe this would be unacceptable to the LFS community in general.  
It just goes too far away from the philosopy that has been around quite 
a while to have one basic path.

> 3. delay udev inclusion for LFS-6.0 and only include it's use is more 
> widespread.
>
> I heavily favor the third option. We never had such high host 
> requirements and udev is not widespread or stable enough for inclusion 
> in testing IMHO.

The question here is whether the Linux developer community expects udev 
to be widely adopted or consider it to be experimental.  Our approach in 
this area should be to follow the intentions of the developers.

  -- Bruce





More information about the lfs-dev mailing list