Re: [med-svn] r9384 - trunk/packages/fis-gtm/fis-gtm-initial/trunk/debian
On Mon, 23 Jan 2012, Andreas Tille wrote:
Just explaining my motivation - not insisting that this was correct: We
are currently talking about / working on the initial package to
bootstrap a real fid-gtm package.
Yes, but I assumed that we only wanted to have one -initial package. So it
should contain both bootstrap tar files for amd64 and i386. Depending on
the architecture of the buildd only one is really used.
As the configure-script from fis-gtm needs to be run as root (this was
explained by Bhaskar some time ago), this package can not be build like
any "normal" source package. So everything has to be done in postinst.
Up to now, this has nothing to do with multiarch but just has to be done
due to the requirements of fis-gtm.
The debconf stuff with user and group in postinst is a another nice
feature of fis-gtm. You can restrict the usage of fis-gtm to a certain
gid. This might be useful in an hospital where more strict security
requirements needs to be fullfilled. Although Bhaskar told me that this
feature is not used often, I left it in. It is defined with low priority,
so everybody who wants to use it might use it and the majority will not
be bothered by it.
Anyway, most of the questions that come up now have been discussed some
time ago when I started to work on the package. So it might be really
helpful to browse the archive.
IMHO there are no user decisions
needed in this phase. Once we are talking about some kind like
multiarch installations things will be different.
See above, this is not multiarch as you might know from libs.
However, I'd like to
keep things as simple as possible in the beginning. So I also would
recommend getting rid of the debconf stuff you injected into the initial
package (again not questioning that it might be needed later).
This package is already beyond the beginning. The debconf stuff is in, it
is kind of useful, so why shall it be removed again?
Thorsten
Reply to: