adding a provides statement to custom kernel package - ill effects?

graysky · 2011-02-19 15:59:20

Can anyone think of any ill effects of adding the following to kernel26-ck's PKGBUILD:

provides=("kernel26-ck-headers=${pkgver}" "kernel26-headers=${pkgver}")

Why? In my case, I don't run with the kernel26 at all, but I do use virtualbox which depends on the kernel26-headers package. Thanks for the advice.

.:B:. · 2011-02-19 16:33:16

If you want a provides= line, you'll need a conflicts= line as well I guess.

graysky · 2011-02-19 16:35:08

@B - that's kind of the point. Is it kosher to two packages provide the same thing? In other word, kernel26-headers and kernel26-ck-headers both providing "kernel26-headers"? Can this cause problems?

ngoonee · 2011-02-19 17:34:35

I'd suggest (in that case) that both packages must provide files in the same locations as well. This wouldn't actually affect anything except any pre-compiled kernel modules (which you shouldn't be using if you're removing kernel26 anyway).

And yes, two packages can provide the same thing. Check out nvidia-utils-beta vs nvidia-utils, both provide basically the same files.

graysky · 2011-02-19 17:53:57

@ng - thanks for the reply. The other case is someone having both kernel26 and kernel26-ck on their system.

JokerBoy · 2011-02-19 18:05:05

graysky wrote:

Can anyone think of any ill effects of adding the following to kernel26-ck's PKGBUILD:
provides=("kernel26-ck-headers=${pkgver}" "kernel26-headers=${pkgver}")

provides=("kernel26-headers=${pkgver}") will be enough to do the trick.

the effects? well, an example will be: if you want to install virtualbox there's no need to install kernel26-headers from [core] 'cause kernel26-ck-headers provides the modules needed for re/compiling the vbox modules too.

tomk · 2011-02-19 18:47:11

fyi virtualbox does not depend on kernel26-headers - it makedepends on it. I use the stock virtualbox package with my custom kernel without any additional provides=, conflicts=, etc.

JokerBoy · 2011-02-19 18:54:30

package_virtualbox() {
    pkgdesc="Powerful x86 virtualization for enterprise as well as home use"
    depends=('libxml2' 'libxcursor' 'libidl2' 'libxinerama' 'sdl' 'libxmu' 'curl' 'libvncserver' 'libpng' 'kernel26-headers' 'gcc' 'make')
...
}

one of us must be wrong.

.:B:. · 2011-02-19 19:16:51

ngoonee wrote:

I'd suggest (in that case) that both packages must provide files in the same locations as well.

That's not possible, in this case .

Just check the contents of any kernel26*-headers package. Even if you'd start hacking around to move things, that would still make no sense at all. Most configuration/make files pick up the running kernel unless you specify otherwise, and that automatism will break.

graysky · 2011-02-19 20:37:51

So consensus is that having the following in my PKGBUILD is fine...?

provides=("kernel26-ck-headers=${pkgver}" "kernel26-headers=${pkgver}")

1) Folks without the ARCH kernel can benefit from packages that depends on the headers
2) This won't hurt folks with both kernels

Lemme know if I'm wrong about this.

tomk · 2011-02-19 23:13:30

Apologies for the above, had an Irish moment.

However the fact remains that I don't need to make any of the changes suggested above to my custom kernel's PKGBUILD to facilitate virtualbox, so I'm not sure why you do.

.:B:. · 2011-02-20 15:37:09

Actually, I think you should just fix the virtualbox (make) dependencies, and not use a provides= line. Your headers package will mislead other packages that might need the official headers, so whlie what ngoonee said is technically not desirable or possible, he still makes a good point.

Arch Linux

#1 2011-02-19 15:59:20

adding a provides statement to custom kernel package - ill effects?

#2 2011-02-19 16:33:16

Re: adding a provides statement to custom kernel package - ill effects?

#3 2011-02-19 16:35:08

Re: adding a provides statement to custom kernel package - ill effects?

#4 2011-02-19 17:34:35

Re: adding a provides statement to custom kernel package - ill effects?

#5 2011-02-19 17:53:57

Re: adding a provides statement to custom kernel package - ill effects?

#6 2011-02-19 18:05:05

Re: adding a provides statement to custom kernel package - ill effects?

#7 2011-02-19 18:47:11

Re: adding a provides statement to custom kernel package - ill effects?

#8 2011-02-19 18:54:30

Re: adding a provides statement to custom kernel package - ill effects?

#9 2011-02-19 19:16:51

Re: adding a provides statement to custom kernel package - ill effects?

#10 2011-02-19 20:37:51

Re: adding a provides statement to custom kernel package - ill effects?

#11 2011-02-19 23:13:30

Re: adding a provides statement to custom kernel package - ill effects?

#12 2011-02-20 15:37:09

Re: adding a provides statement to custom kernel package - ill effects?

Board footer