[gentoo-dev] Unified tool for Gentoo specific tasks?

[Michael Kohl]

Hi!

One of the things which disturbs me about Gentoo Linux (admittedly only
slightly, but still) is the "clutter" of seperate tools I have
installed which are Gentoo specific: mirrorselect, ufed, gentoolkit
(although I only have it because of etcat), some things from the forums
like (ebb - http://forums.gentoo.org/viewtopic.php?p=275364#275364
and femerge - http://forums.gentoo.org/viewtopic.php?t=45827), epm
(which I haven't installed, but still). Sure this is not really a big
problem, but various posts in forums and mailing lists lead me to the
conclusion that quite a few users are not aware of one or the other of
this useful little helpers.

Hence an idea came to my mind, which I'd like to discuss on this list
for a while before going through the hassle of writing a full-featured
GLEP:

Daniel Robbins never made a secret of Gentoo (or at least the Portage
system) being inspired by BSD. One of the things I really like about
FreeBSD is /stand/sysinstall, where you can configure many aspects of
your system quite comfortable via a curses based interface. This lead me
to the conclusion that something similar would be nice for Gentoo Linux.

As I already hear you scream, here's what I _don't_ want to propose: no,
I'm not arguing in favor of a tool which configures every aspect of your
system. linuxconf, Webmin and the likes do this job fair enough,
reduplicating their efforts would be something I'd rather not want to
see. Also I have the feeling that most Gentoo Linux users are familiar
and comfortable with editing config files by hand.

What I was thinking about is a unified, curses-based frontend for the
various aforementioned tools. And creative like I feel today I already
came up with a code name for this proposal, "Emergency". ;-)

So, what do I "Emergency" want to be:
1. a unified frontend for _Gentoo specific_ configurations
(mirrorselect, ufed, etc.)
2. I'm voting for a curses based frontend so this tool can be used on
machines without XFree, via ssh etc. but think the tool should be
modular enough for different interfaces (e.g. gtk, qt, browser based).
The kernel config comes to mind as an example (compare make menuconfig
to make xconfig and make gconfig (gtk, development sources)), as well as
mldonkey.
3. people not interested in "Emergency" shouldn't have to use it! Maybe
a useflag can do the trick and install tools like ufed and mirrorselect
either as standalone applications or as plugins for "Emergency".
4. In regard to 2. and 3. the whole application should be designed as
modular as possible, and allow for plugins to be written in as
many languages as possible (shell scripts, Perl, Python and Ruby come to
my mind, but I'm sure there are also some people who'd love to see C or
more exotic languages thrown in the mix). So maybe "Emergency" wouldn't
really be an application on it's on, but a framework which allows the
authors of the various config tools to integrate them nicely.
5. Perl being famous for being a good "glue language" would be my
preference for the framework, but I'm not biased towards any language in
particular and open for argument on the pros/cons of alternatives.

Tools I'd love to see included: mirrorselect, ufed, epm, an ebuild
browser (as full featured as possible, making use of all the nice stuff
available in etcat), gcc-config, java-config, genkernel (also I haven't
looked at it yet, maybe it's not really a candidate for being included
in this kind of tool), a new tool for managing all the"portage options"
in/etc/make.conf(like portdir overlay, rsync retries, logging,
fetchcommand, offering the comments from make.conf as ufed does with
use.desc etc.). I'm pretty sure if we all think for a while we can come
up with quite a few more modules... 

Please note that I didn't really do much of coding during the last few
years (except for web related stuff - php, mysql etc.), but just played
around with various languages (namely Perl, Ruby and nowadays Python)
for fun. _BUT_ if there's interest in a tool like this, I would love
to put a considerable amount of time into this project, helping out
wherever I can, be it coding, improving the concept, planning or
coordinating of the development...

Sorry for this rather long message, I just wanted to make the idea as
clear as possible.

Curious about what all of you think about this,
Michael

-- 
www.cargal.org 
GnuPG-key-ID: 0x90CA09E3
Jabber-ID: citizen428 [at] cargal [dot] org
Registered Linux User #278726