Re: Preparation of Debian GNU/Linux 3.0r2 (II)

[Henning Makholm <henning@makholm.net>]

Scripsit Osamu Aoki <osamu@debian.org>

> As long as the removal request clearly recognizes that this request/innuendo
> by the Hitachi is found to be baseless, I think we should remove these
> old fonts as useless dataset.

> I think previous message by Henning Makholm also think this way.
>   Message-ID: <[🔎] 87wu9vldmo.fsf@kreon.lan.henning.makholm.net>

I was talking (or at least, thinking) about unstable/testing. I think
removing from *stable* is a much more serious matter. It ought only to
happen if a plausible case can be made that keeping the package/file
in the distribution can actively harm users and/or mirror operators.

As a concrete example, suppose that somebody has built a mission
critical system around woody 3.0r1. Part of this system is scripts
that have the names of the offending fonts hardcoded. Now the user
wants to prepare a new machine for running his system. He boots a
minimal woody system from cd-rom and proceeds to do a network install
of the rest of woody 3.0r3. Oops - the fonts his existing scripts look
for are missing. That is exactly what the tag "stable" implies should
*not* happen, or at least should only happen when we genuinely have no
choice. One can expect to need to tweak one's scripts when moving from
one stable release to another, but it should not happen with point
releases.

-- 
Henning Makholm                         "This imposes the restriction on any
                                  procedure statement that the kind and type
                             of each actual parameter be compatible with the
                       kind and type of the corresponding formal parameter."