[OE-core] [PATCH 1/2] remove allarch from recipes which rdepends on TUNE_PKGARCH packages

richard.purdie at linuxfoundation.org richard.purdie at linuxfoundation.org
Tue Aug 7 10:44:38 UTC 2018


On Tue, 2018-08-07 at 17:37 +0800, Kang Kai wrote:
> On 2018年08月06日 22:51, Mark Hatle wrote:
> > On 8/3/18 4:28 PM, richard.purdie at linuxfoundation.org wrote:
> > > On Fri, 2018-08-03 at 09:49 +0800, kai.kang at windriver.com wrote:
> > > > From: Kai Kang <kai.kang at windriver.com>
> > > > 
> > > > The following allarch packages rdepends on TUNE_PKGARCH
> > > > packages. So
> > > > remove allarch from these recipes.
> > > > 
> > > > ca-certificates -> openssl
> > > 
> > > Should we reverse this, i.e. have openssl RRECOMMENDS ca-
> > > certificates?
> > 
> > I believe this is what other distros do.  I think the dependency on
> > ca-certficiates was there because there is a script included that
> > uses OpenSSL
> > to generate the certificates.  (Theoretically you can use this
> > script post
> > install to update the certs.)
> > 
> > /usr/sbin/update-ca-certificates
> > 
> > I suspect the others had similar dependencies.. some included
> > script needed
> > something from another package to run.
> > 
> > Not sure if this changes the suggestion though...
> > 
> > --Mark
> > 
> > > > docbook-xsl-stylesheets -> perl
> > > > font-alias -> font-util
> > > > initramfs-module-setup-live -> udev-extraconf
> > > > initramfs-framework -> ${VIRTUAL-RUNTIME_base-utils} (busybox
> > > > by
> > > > default)
> > > > ppp-dialin -> ppp
> > > 
> > > ppp RRECOMMENDS ppp-dialin?
> 
> ppp-dialin installs script ppp-dialin which calls /usr/sbin/pppd.
> 
> 
> > > 
> > > > resolvconf -> bash
> > > > weston-init -> weston kbd
> > > 
> > > weston RRECOMMENDS weston-init?
> 
> weston-init install init file which calls weston or weston-launch
> from 
> package weston and openvt from package kbd.

Agreed.

My worry is that this move effectively makes allarch usesless since any
script using an interpreter will now need to be not allarch since most
interpreters are not allarch.

I'm not sure this is a good move...

Cheers,

Richard




More information about the Openembedded-core mailing list