Features Download
From: Andreas Tille <andreas <at> an3as.eu>
Subject: Re: blends linked from the main debian page
Newsgroups: gmane.linux.debian.devel.project
Date: Monday 11th April 2011 21:33:17 UTC (over 6 years ago)

On Mon, Apr 11, 2011 at 04:50:14PM +0200, Stefano Zacchiroli wrote:
> > 3.  In Debian we have a number of great sub-projects which are not
> > exposed even within Debian itself.  I have once suggested to float all
> > the blends up to the main Debian page to make it at once look appealing
> > and specific to a broad range of specialists.  But I failed to
> > the proposal, so it hasn't worked out yet.
> I like a lot the idea of providing links to blends from the Debian
> homepage [1], thanks for proposing (and revamping) the idea!  However,
> from the above paragraph I do not understand what you needed to
> generalize and why it is blocking the implementation of the proposal.
> Can you be more specific about that?

As you can probably assume I'm in favour of making Blends more popular
and thus I in principle support the idea to make them visible on the
main Debian page.

However, my concern is of a bit different nature than yours (web page
design).  The problem we have is that while three Blends are working
nicely and fully adopted the idea (namely Debian Edu, Debian Med and
Debian Science) and people are actively working with the existing tools
as well as having understood the idea we have currently projects which I
would call "potential" Blends.  There are for instance Debian Jr
(actually the first Blend) which is orphaned - my regular pings at
relevant places were not answered by anybody and currently the only sign
of Debian Jr is SPAM on its mailing list.  (Well, OK, I updated the
metapackages if needed - but that's no project work.)

We also have Debian Lex which is more or less a stub.  It would not be
really helpful if you would direct lawyers onto this project - except if
it will be made really clear that this is a place for developers but not
for users.

We have Debian Accessibility which is using the tasks pages but do not
want to use the metapackages and do not really identify itself as a
Blend.  As long as the main drivers of a project do not adopt the idea
(see below) I'm hesitating to advertise them as Blends.

Than we have those projects which are rather packaging projects which I
tried to convince adopting the ideas of a Blend.  These are Debian GIS
(just convinced them to do conversation on a list named debian-gis
rather than pkg-grass), Debian Multimedia (some people there consider
the idea of Blends cool but there is not much effort done to really
maintain a complete Multimedia system but rather a set of single
packages),  DebiChem (where one main supporter is happy about the Blends
approach but has time constraints), Pkg-Games which did not even started
working on Blends stuff, but IMHO would have great potential becoming
a Blend (and could even copy / adopt Debian Jr).

So the question is: What Blend is actually worth beeing mentioned on the
main page?

>From my point of view the point of Blends is that we technicans
understand that it is not enough to simply add technically perfect
packages one after the other to a flat package pool.  We need some
substructure and this substructure needs to be organised and actively
maintained.  If we approach this we can reach much more in the following
real live cases:

  1. Contacting upstream authors (perhaps about fixing their licence):
     a) Single maintainer:
        Hi, I'm Joe Random Debianpackager and would like you to ...
     b) Blends maintainer:
        Hi, I'm writing you on behalf of  name.  The Blend has
        the purpose to establish your software into a complete framework
        with programs covering the field .  If you think
        your program fits well into this Blend amongst ... could you
        please ...

  2. User asks you: If I want to do the following task, what software
     should I install?
     a) No substructur via Blend:
        I would suggest to install packages a, b, and c.  Furthermore
        it might be useful to install d or e as you want.  Also f is
        interesting. ...
     b) Blend:
        Just install metapackage x and your system is ready prepared.

  3. Some journalist asks you as an outsider of a specific field about
     solutions in this field.
     a) No substructure:
        Well, I have heard that there should be some packages amongst
        those 29.000 Debian packages which might be useful.
     b) Existing Blend:
        I'm no expert in this field but I know there is a team inside
        Debian which actively cares exactly for these issues.  Please
        contact this team (via Mailinglist) to get more detailed

There are a lot more examples why such a substructure is helpful.  IMHO
it is one way to advertise Debian.  I would go that far that Debian can
grow dendritic into different fields and spreads over different fields
into the masses (and thus is "the last final step for total world
domination" - to quote Enrico).  We could even declare Debian as an
application store for applications in the fields covered by Blends.

You know that I'm driving Debian Med since 9 years.  Honestly, I think
medicine is a quite unimportant field compared what's there in Free
Software for Multimedia, Games, GIS and others.  Medicine has even an
additional problem:  The users tend to be quite conservative regarding
their software and usually have some money left to not care much about
license fees.  So my point here is:  If it is possible to build a Blend
inside this niche field - how much more success could other topics have
if people would be more engaged in running a Blend?

Blends could be a way to make sure that the following dialogue could
be quite frequent:

   User 1: What Linux distribution should I use if I'm engaged
           in workfield X.
   User 2: Use Debian.

I think for X=Medicine and X=Education we are almost there.  So why are
we not in other fields where this could happen?

You might even ask:  Debian has so many packages - do we really need so
many quite specific packages with low popcon.  The usual answer is: If
the packages are properly maintained and bug free - why not.  For proper
maintenance we are "hiring" developers out of this specific field.  We
have (at least! - I should make a poll about this) five Debian
developers which are *only* in the project *because* Debian Med exists
(some more are in NM queue).  These people would not have joined Debian
otherwise but they became engaged because we are caring for their topic
and they see the advantage.  This includes proper training of packaging
and having a warm welcome for engaged people.

Yes, this all is work.  It is different work than all this technical
packaging fun.  You need people who like to do this work.  If you do not
like to do it in your own Blend try to catch some other people who
understood the idea and bring in their time to manage a Blend.

Uhm, that's a long mail - just to say:  I'm fine with pushing Blends
to the main page if we can be sure that we will not list stubs there
but really living projects who are obliged to the Blends idea - not
only the pure technical part.

Kind regards


CD: 3ms