Re: stupid dependencies on update-inetd
- To: debian-devel@lists.debian.org
- Subject: Re: stupid dependencies on update-inetd
- From: "David Lopez Zajara (Er_Maqui)" <er_maqui@darkbolt.net>
- Date: Thu, 09 Aug 2007 08:44:09 +0200
- Message-id: <[🔎] 46BAB7B9.40009@darkbolt.net>
- In-reply-to: <20070730102628.GA15581@ikki>
- References: <20070728225703.GB18126@bongo.bofh.it> <20070729114639.GA4303@ikki> <20070729115751.GA30501@bongo.bofh.it> <20070729153838.GA23254@ikki> <20070729165840.GB5218@bongo.bofh.it> <20070730102628.GA15581@ikki>
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Klaus Ethgen wrote:
> Hi Marco,
>
> Am So den 29. Jul 2007 um 18:58 schrieb Marco d'Itri:
>>>> If you want a system without an inetd then do not it install one and do
>>>> not install packages depending on it. It's really that easy.
>>> Sorry but I think you didn't understand what I tryed to explain.
>> I do, it's you who have no idea about the discussions of the past five
>> years.
>
> Might be. There are so much flame war on that list. Also my English is
> not that good and my spare time not that much that I can read any mail
> in the list; sorry.
>
> However:
>
>>> be installed too. Xinetd do disable such a inetd in postinstall script.
>> Yes, the xinetd package needs to be integrated with the rest of Debian.
>
> I agree.
>
>>> However. What I did explain is that xinetd do not need to have a
>>> update-inetd as the configuration is done with single files for single
>>> packages.
>> No, it needs an update-inetd program which can create configuration
>> files in the appropriate format.
>
> Hmmm, Wrong in my opinion. If xinetd would have its own update-inetd and
> software is installed in xinetd and $ADMIN decides to switch back to
> traditional inetd the configuration is inconsistent. Also the way
> around.
False, because if the <new> xinetd uses another config packages, the
system have a config for fallback to inetd-superserver but doesnt have a
working configuration.
Packages need to provide their own update-inetd, and in the case of a
change of inet-superserver, this package manages the calls to touch the
appropiate configuration. Also, the packages "needs" an config-conversor
for change inetd.conf to their files in their postinstall, and another
to change back to inet.d on remove.
> It might be a better way to have a lintian warning if a package has a
> update-inetd call and no xinetd config or vis versa. Note that xinetd do
> not need the existence of a update-inetd tool ever. And other, newer
> inetd might be the same. Maybe there can be a dh_ tool for creating all
> needed inetd configurations.
>
> I think that all the stuff should allow to switch from whatever inetd to
> another without loosing configuration (coming from the package). And
> there is only two ways how you can do that:
> 1. Creating all needed configurations at build time
> 2. Having a single update-inetd which create ALL configurarions at
> install time. But this make update-inetd package as dependency for
> installation and NOT inet-superserver (the later can be suggestion as
> well.)
>
> Gruß
> Klaus
- --
er_maqui@darkbolt.net || http://maqui.darkbolt.net
Linux registered user number: #363219
PGP key avaliable at KeyServ. KeyID: 0x4233E9F2
- ------------------------------------------------------------------
Los hombres somos esclavos de la historia
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFGure5fFjA4EIz6fIRAhayAKC0TyiokxXRvm7LzA2tx4+FwY8PnQCg7abV
R4OFOSXXgmfIxfqXgbVd3KY=
=fLX8
-----END PGP SIGNATURE-----
Reply to: