> > >I'm definitely interested in comparing with whatever you generate as well.
> well, here we are:
> mostly, I set apt up like you would to do a network install like in
> the docs (but to my hurd partition on the same box, instead) -- with
> apt downloading the files from in linux, then dpkg installing them
> from in the hurd. the difference is that I don't have any network
> connection in the hurd at all, so I'm forced to reboot between linux
> and the hurd on the same machine.
> I wanted to handle the package installation ordering automatically,
> though, so I wrote a script to build a log of what apt tries to do
> with dpkg in linux, then reconstruct that sequence of dpkg calls once
> I rebooted into the hurd. it worked once on a dist-upgrade, but no
> guarantees, because I only barely know what I'm doing :)
Nice. I hadn't thought of doing it that way.
Our two methods are very similar, except that I rely on apt to do all
the work, and you record what apt would have done and drop down to the
dpkg level after the package fetch.
apt still hangs occasionally in the Hurd, so your way may be able to
avoid that annoyance.
The amount of travel from machine to machine is still the same. I'm just
lucky to have a spare machine to run the Hurd on. I do my network
fetches for it using this method to lower the occurences of apt hanging
on Hurd, since the big network download occurs on a linux machine.
Thanks for posting your stuff.