[lfs-dev] dbus, systemd, polkit, and consolekit

Ken Moffat zarniwhoop at ntlworld.com
Sat May 10 20:37:39 PDT 2014


On Sat, May 10, 2014 at 08:07:18PM -0500, Bruce Dubbs wrote:
> Ken Moffat wrote:
> >
> >  I see you've done that - if I sound grumpy it's because I'm still
> >debugging the changes I made to my own scripts to have the three
> >alternative builds [ found an interesting error in my own chapter 7
> >scripts - left a place for setting the systemd init (and did not set
> >sysvinit like the book did), with a comment, but forgot to code it :
> >init fell back to /bin/sh ].
> 
> Before I started making changes, I created a hybrid branch.  I just
> committed it, so that work isn't lost.  I do appreciate your help in the
> experiment, but I just didn't think it would work out.  I really tried to
> combine sysd and sysv, but in the end, they required a divorce.
> 
 I thought it was an interesting experiment, although intrinsically
not to my taste.  In terms of the books, I still think branches
would be easier to follow, as would trying to limit our commits to
single changes [ or groups of packages which ought to be done
together ] but I appreciate that maintaining branches in svn is
unpleasant, despite what you have said in the past.

> >Now plugging on, for what may well be
> >my last ever build of systemd.  Dunno, I suppose I might try it in
> >the distant future if I ever want to build gnome or kde, but that
> >does not look likely at the moment.  Anyway -
> 
> I don't think KDE requires systemd.  I hope they don't but into that.
> 
 I recall something on lwn.net from last year, which I think said
that the kdm maintainer intended to make it require systemd.
Perhaps I misunderstood.

> >  Systemd brought in gperf (for the keymaps - I guess that eudev
> >could use it, if anyone has a need for these), acl, attr, libcap,
> >xml-parser.  Can we look forward to any of these additions returning
> >to BLFS ?
> 
> I was thinking that we should leave those in LFS.  We have never advertised
> it as a minimal system.  Having them in LFS simplifies the requirements
> sections in BLFS by quite a bit.   We can revisit that later if there is a
> huge hue and cry lamenting that.
> 

 For me, acl only ever used to get built if I was building kde to
test it.  For my server, whenever I rebuild it, I know which of
these I can do without ;)  For my desktops, the others get pulled in
(recall that I use nfs) so it's no big deal for me.

 However, we have tended to keep old text claiming how small LFS can
be.  Against that, some/many people have thought they did not want
autotools [ hah! let them try to build an application which needs
its config file to be patched ].

> I will redo dbus in BLFS however.
> 
>   -- Bruce

 That one definitely needs to be in BLFS if we are not using
systemd.

ĸen
-- 
das eine Mal als Tragödie, dieses Mal als Farce


More information about the lfs-dev mailing list