[Mageia-discuss] Suggestions

[Frank Griffin]

Michael Scherer wrote:
>
> To the defense of the drakx developpers, I do not think that choosing in
> the installer is really a so good idea :
>
> - during installation, you do not have web access. Thus, you will have a
> hardtime to really find information on what does a software. If you use
> rpmdrake, you can ask to friend, ask on forum, ask on a search engine.
>   

This is really a more general issue of the availability of detailed help
during the install.  To focus on package descriptions, which really
*are* of interest only to more advanced users (very few newbies know
enough about Linux to care about minimalist installs), completely misses
the point that there is a lot of other information about what's going on
in the install that *would* be of interest to newbies.

The issue, as always, is competition for space or bandwidth between help
and program content.  If you access it through the network, people
without network access won't get it.  If you put it on the media, it
redices the space available for programs.

This is why I think that such help, package descriptions, etc., should
be separate from the rpms.  In the past (and maybe still, as I haven't
done a from-media install for a while), the install asked the user if he
had additional media to use.  A slight expansion of this could ask how
many CDs/DVDs the user has available and whether the network will be
available (or should be activated) in order to access additional
packages and help content.

For the install media, we should go back to the arrangement we had in
the multi-CD days.  Cooker required something like 9 CDs for everything,
but the essentials were placed on the first CD, and content was arranged
on the others by type.  The "standard " install used 2 or 3 CDs, and the
install basically tailored itself to the number of CDs available.

In the same spirit, we could have a set of package-related ISOs, and one
or more documentation ISOs.  If a non-network user wants extended help
and package descriptions in translated format, he obtains these ISOs. 
If not, he doesn't.  At the start of the install, the user gets a prompt
with checkboxes for each of the possible ISOs, and can indicate which
are available.  For any that aren't, the install doesn't even try to use
what's on them.  If the install detects enough available unused disk
space, then the first use of any ISO can copy some or all of the ISO to
hard disk for the duration of the install.  Any prompt for an ISO has a
way for the user to say he really doesn't have that one, in which case
it is not prompted for again.  All this should minimize the amount of
disk-swapping.

That answers the objections of those who don't want to have to download
many ISOs to do an install, and also addresses the needs of non-network
users (e.g. small schools) who want a full-featured set of install media
that can be reused repeatedly for friendly installs without network
access.  It also minimizes disk-swapping, unless the system is really
tight on space, in which case the install is at least still possible,
albeit with some disk swapping (assuming the user wants to use multiple
ISOs).

As always, network users could opt to download dynamically anything they
didn't have ISO media for, with the same provision for caching, if space
allowed.