Whoop, there it is!

Brian Elliott Finley
Mobile:  630.631.6621

-----Original Message-----
From: "Bernard Li" <[EMAIL PROTECTED]>
Date: Wed, 7 Dec 2005 10:15:02 
To:<[EMAIL PROTECTED]>, <sisuite-devel@lists.sourceforge.net>
Cc:"Brian Elliott Finley" <[EMAIL PROTECTED]>
Subject: RE: [Sisuite-devel] Re: boot-standard requires

Hey Erich:

You can actually build it on the headnode but on the_image_.  In that
case you can potentially have a different arch image than that of the
headnode.

I think we should keep the initrd_template as noarch.

Cheers,

Bernard

> -----Original Message-----
> From: Erich Focht [mailto:[EMAIL PROTECTED] 
> Sent: Wednesday, December 07, 2005 10:07
> To: sisuite-devel@lists.sourceforge.net
> Cc: Brian Elliott Finley; Bernard Li
> Subject: Re: [Sisuite-devel] Re: boot-standard requires
> 
> On Friday 02 December 2005 17:43, Brian Elliott Finley wrote:
> > Benard and I are contemplating the packaging of SI 3.5+.  
> We're trying
> > to make a decision about how the initrd_template files should be 
> > packaged.  Currently they're part of $boot_package.  
> > 
> > Here are some options:
> > 
> >     1) Create a new package, systemimager-initrd_template, and have
> >        systemimager-client depend on it.  only the arch for 
> the client in
> >        question would be necessary, and this would end up 
> being an arch
> >        dependent package:
> > 
> >         systemimager-initrd_template-$version.i386.rpm (for example)
> >         systemimager-initrd_template-$version.ia64.rpm (for example)
> 
> Could I build an initrd on another node than the client? For 
> example on the
> master? In that case I might prefer to have the package as 
> noarch, just in
> case the master is of different architecture than the client.
> 
> systemimager-initrd_template-x86_64-$version.noarch.rpm
> makes sense, too, one needs to put the %Requires: lines inside %ifarch
> blocks. With multiarch setups one would have to install the additional
> packages manually.
> 
> But if you_allways_ want to build initrd on the client, an 
> arch dependent rpm
> is fine, of course.
> 
> Regards,
> Erich
> 
> 

Reply via email to