[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: old GNU tarball

On Sat, Jan 25, 2003 at 06:40:20PM +0100, Marcus Brinkmann wrote:
> On Sat, Jan 25, 2003 at 12:20:43PM +0100, Joachim Nilsson wrote:
> > [..]
> Verbosity is enabled by default, because creating the device files is slow
> and people will not have any diagnostics if it fails (for example, if the
> machine hangs for whatever buggy reason).  As ideally, native-install is
> only run twice, it doesn't make sense to give it options.  People will not
> use it often enough to explore them.
> Is setting the terminal required?  I guess it helps debconf, but only if it
> has an effect on the installation itself it should be added.  People are
> required to reboot immediately after running the script anyway.

Initially my intention was not to distribute this script further. I only
made these changes for myself since I wanted a quick way to reinstall
the system. Primarily for the demo I mentioned, but also for other work
I do. So, removing the verbosity had really no function but for me while
the TERM recommendation was taken from the Install Guide.

> +# The Hurd wants libncursesw5 (wide chars), some other
> +# packages are built against libncurses5. The latter
> +# packs should perhaps be rebuilt to be able to use
> +# the wide chars supported ncurses. The same goes, I
> +# guess, for libdb2 vs. libdb3 ... /Joachim
> It's not our business to tell other packages to use wide char support, or
> the version of db.

The tarball was increasing in size. Previously the Hurd package used
regular ncurses so this was merely a note to tickle either myself or
someone else to see if the packages in the tarball depending on ncurses
could make use of ncursesw. Similar for the db2/db3 issue. Neither of
these are probably not applicable and should have gone into my own 
TODO/look-in-to-this-later file.

> +dpkg --configure libncurses5
> +dpkg --configure libncursesw5
> +dpkg --configure libpam-runtime
> +dpkg --configure libpam0g
> Why is it necessary to configure ncurses5, ncursesw5, libpam-runtime and
> libpam0g before the others?  The purpose of configuring base files and db
> before everything else is to help dpkg in the package ordering, so only one
> --pending --configure run is necessary.  Under normal circumstances, dpkg
> should be able to configure all packages in one go, including ncurses and
> libpam, without being told in which order.  The Hurd package does not require
> ncursesw5 be configured before the Hurd package itself.  Did you just add
> the lines blindly, or did you do this to fix concrete errors?  Your comment
> does not tell.  If you didn't add those to fix specific errors, please try
> again without such gratuiutous changes.  Always keep modifications to a
> minimum.

I can look into that again. At first I might say "residue from when I
first constructed the tarball", but the truth is that I don't remember,
so I'll look over it again. When I've figured it out I'll add a note.

> About mentioning nano: Doesn't nano add itself as the "editor" alias?  It
> makes more sense to mention "editor" rather than "nano" specifically.


> Your message changes belong into the documentation, not into the script. 

What documentation do you mean, the Install Guide?

> They are also sometimes slightly incorrect.  For example, it is not
> necessary to use the nano editor, any editor would work (or firing
> radioactive beams at the hard disk to diddle the bits).  This is quite
> different from, for example, /etc/passwd, for which vipw should be used.

True, but for the tarball installation I think only nano is available.
Sure I could have put 'editor' instead, but I've personally never used
that alias for anything and a first time Hurd user (say a RedHat user)
wouldn't understand I think. Of course this is only my own speculation
about user behaviour, so it's not worth much in this context of
"Debianistas". :-)

I'll revert the changes and drop another note here when I'm done.


Joachim Nilsson  :: <joachim AT vmlinux DOT org>
+46-(0)21-123348 :: <http://joachim.vmlinux.org>

Reply to: