[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Apt-pinning confusion



On Tue, 27 Mar 2012 14:01:12 +0000, Ramon Hofer wrote:

> On Tue, 27 Mar 2012 15:23:25 +0300, Andrei POPESCU wrote:
> 
>> On Ma, 27 mar 12, 12:07:08, Ramon Hofer wrote:
>>> 
>>> Thanks for the explanation!
>>> So why didn't they "just" update the version that won't receive any
>>> updates?
>> 
>> The new version changed ABI[1], which means all modules compiled
>> against bpo.1 need to be recompiled for bpo.2.
>> 
>> [1] http://en.wikipedia.org/wiki/Application_binary_interface
> 
> So the ABI is about the same as a module? Like the one I had to compile
> (jme.ko [1]) to get the network up?

Mmm, not quite so although it shares the same essence. 

In brief, to my understanding, kernel ABI helps developers to keep track 
for module changes that need to be recompiled and thus avoiding to 
recompile them all. When that happens, it is exposed by incrementing the 
last number of the package (in this example, "bpo.1" → "bpo.2").

> [1] http://lists.debian.org/debian-user/2012/02/msg02240.html
> 
> Still I don't understand why that kernel update couldn't trigger the
> recompilation of the new modules.
> Maybe there's a reason why they are held separately?

They are treated as two different packages because they are indeed two 
different packages providing different modules.

What you need to keep the kernel package updated to the latest available 
version in the backports is the kernel metapackage ("linux-image-686-
pae"), as this will trigger the most recent version.

Greetings,

-- 
Camaleón


Reply to: