Re: diversion/conffile
On Wed, Feb 20, 2002 at 04:02:40PM -0600, Manoj Srivastava wrote:
> >> >> You can not divert conffiles. I've tried. And I am a dpkg devloper.
> >>
> Josip> Sure you can, we've been doing it for years now :) It doesn't
> Josip> actually work 100% but hey...
>
> I was somewhat surprised by this light hearted flaunting of a
> policy violation. I can see that policy is not exactly seen to be
> writ in stone, but I would have thought that policy violations by
> design would be at least discussed here, and perhaps noted as an
> exception to the rule because there is not other reasonable
> solution.
I picked it up from the previous xinetd maintainer. It was working, there
didn't seem to be a better way, nobody complained, so it stayed. Nobody ever
pointed at the problem, that is the only reason why it was not discussed.
I'm no longer maintaining xinetd, BTW.
> Josip> Oh and if this request is related to
>
> >> preparing to file a serios bug against linuxlogo unless conviced otherwise
>
> Josip> then the explanation, IIRC, is that xinetd can't conflict with
> Josip> netkit-inetd because netbase depends on it, and xinetd depends
> Josip> on netbase.
>
> Well, unless a means is found to have the packages cooperate,
> this is a serious bug against xinetd, isn't it?
>
> Perhaps the solution would be to have update-inetd somehow
> accommodate both xinted and netkit-inetd, and then have a virtual inetd
> package provided by both, and have netbase depend of this virtual
> inetd package (perhaps not called inetd to allow us not to deal with
> versioned virtual depends).
Yes, that would be The Right Way(tm) to fix this, but update-inetd doesn't
support writing xinetd.conf. This has been filed as a bug, and someone
(Peter Makholm, I think?) worked on it. I don't think it ever got completely
fixed, unfortunately.
> This is what the dpkg maintainer says about what xinetd does
> right now:
>
> Adam> Whenever the diverted conffile is modified upstream, dpkg
> Adam> chokes during it's installation. Verified by trying it, and by
> Adam> walking dpkg's code.
> Adam> That part of dpkg's code is very hairy.
Yes, and this has been filed as a bug, like I said before. Adam said
"chokes", but it doesn't actually cause any major damage, AFAICT.
So I don't actually see what practical reason is there to declare this a
serious bug on xinetd and thus prevent it from releasing.
--
2. That which causes joy or happiness.
Reply to: