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

Re: Please test debootstrap



On Mon, Aug 20, 2007 at 11:17:49PM +0200, Luca Capello wrote:
> With the following patch at least we removed the error:
> --8<---------------cut here---------------start------------->8---
> --- sid.ORG	2007-08-20 22:39:39.000000000 +0200
> +++ sid	2007-08-20 22:41:54.000000000 +0200
> @@ -59,7 +59,7 @@
>      if doing_variant fakechroot; then 
>          setup_devices_fakechroot
>      else
> -        setup_devices
> +        test $ARCH != "hurd-i386" && setup_devices
>      fi
>  
>      x_feign_install () {
> @@ -85,6 +85,8 @@
>  }
>  
>  second_stage_install () {
> +    test $ARCH = "hurd-i386" && setup_devices
> +
>      x_core_install () {
>  	smallyes '' | in_target dpkg --force-depends --install $(debfor "$@")
>      }
> 
> Diff finished.  Mon Aug 20 22:52:18 2007
> --8<---------------cut here---------------end--------------->8---

OK, good.  I'll try to talk to Colin Watson about this, because he's
both been doing some Hurd porting in the past and is now a debootstrap
co-maintainer via d-i.
 
> But the segmentation fault is still there (so no second-stage) and I
> cannot try on qemu because of the already reported error :-(

Do you mean the ext2fs.static error you mentioned earlier, or some other
issue?
 
> > No idea whether this would work, but that is what we should aim to
> > get working at some point.
> 
> I don't know the general plan, but I was thinking about using the
> GNU/Linux d-i_[i386|amd64] to install hurd-i386 (thanks to
> debootstrap).  Or are we talking about an hurd-i386 d-i CD?

Well, whatever comes first.  Certainly the long-term goal is to have a
native d-i for us.

If somebody can hack together a cross-d-i which uses the current
GNU/Linux d-i with minimal changes to cross-debootstrap hurd-i386, that
would be nice.

However, I don't think d-i is fit for using debootstrap's split
second_stage thing - it probably assumes a full debootstrap run at
various points.  On the other hand, this could be a valid use case - one
could put a harddisk into a i386 and install e.g. a mips or arm Debian
GNU/Linux on it to put it into some semi-embedded box without CD drive
etc. and finish from there.  So this might be worth bringing up with d-i
people.


Michael



Reply to: