Re: language of choice for k-p packaging ...

[Sven Luther <sven.luther@wanadoo.fr>]

On Sun, Nov 13, 2005 at 07:11:40PM +0100, Bastian Blank wrote:
> On Sun, Nov 13, 2005 at 06:18:36PM +0100, Sven Luther wrote:
> > Ah, i was going to not parse anything and just copy it as is into INITRD_CMD.
> > Mmm, wait, i forgot something, this is not the usecase i want, what i want is :
> > ramdisk: mkinitrd.yaird mkinitramfs
> > the list of actual binaries being called, not the dependencies. No versions in
> > sight for this, and the full value need to be overridable globaly, per arch,
> > per subarch or per flavour.
> 
> This needs a reverse-lookup to generate the dependencies.

The dependencies can be set by hand, i believe, and don't really need to be
set by arch, the important thing is the list of actual tools used.

> The following includes any information to build both dependencies and
> the list of executables without hardcoding any piece of information.
> | [base]
> | image-ramdisk-generators: initramfs-tools
> | 
> | [relations]
> | initramfs-tools: bla
> | 
> | [image-ramdisk-generators]
> | initramfs-tools: mkinitramfs

Yeah, well. might be overkill for what we need, do we really need all this
complexity ? 

> I'm not sure about the linux-initramfs-tool part of the dependency.

That is a virtual package, so future ramdisk tool can satisft the dependency,
without needing to recompile the kernel.

Friendly,

Sven Luther