[DebianGIS-dev] Bug#588812: Debian GIS metapackages available [Was: debian-gis_0.0.1_i386.changes ACCEPTED]

[Andreas Tille]

On Wed, Sep 29, 2010 at 02:44:52AM -0700, Hamish wrote:
> Sorry if I come off a bit grumpy :)

No problem.  Sometimes e-mail communication is hard. ;-)

> > The more part is that I'm having the hope that the list
> > will not only get a new name (and home)
> 
> (by 'home' I take it you mean @lists.d.o instead of @alioth.d.o server?)

Yes.  That's exactly the whole point of the arguing.  I would not make
such a fuzz about just a list renaming on alioth and the bug report I
was refering to is exactly about this.
 
> > but turns into a medium where we actively can attract users (and later
> > developers) better than before.
> 
> among those who frequent the lists of debian lists, yes, and that is not
> a bad crowd to be exposed to.

Exposal to people - that's actually the keyword ...
 
> > So what to say?  My first point is that she did not understand the
> > Do-O-cracy principle in Debian.  Not "Debian has decided" but one (or
> > more) developers just have done something (packaging Qgis).
> 
> my first impressions on that comment are
> - ...
> - I'd forgive the "debian executive has made a decision to package qgis"
> as it is quite a strange notion to outsiders who are used to corporations
> or top-down NGOs.

I can forgive this as well for sure.  My point was: We need a visible
channel to educate people about the difference in Debian and who they
can become part of the movement.  It happens 3-4 times a year on Debian
Med list that we are able to turn people with similar imagination about
Debian into members of the packaging team (currently there is also a
thread active where I try to patiently explain about proper packaging of
OpenMRS (hospital management system) to a very engaged Debian newbee who
is closely connected to upstream.  This guy is learning quite quickly
and we have contact *because* there is a visible list.

> > backport or not).  So she simply failed to get the information
> > she was seeking.
> 
> and yet she did hear of it from somewhere anyway..?

But only the rumor - not a way to apt-get install ...
 
> there is one, two actually if you include UbuntuGIS, and three if you
> include the qgis list(s), and many more if you include the local osgeo
> chapters and international osgeo lists.
> 
> I fully agree that lowering the barriers to entry are important.

That's the point.  We should not trust other parties like Ubuntu or
different upstream.  IMHO it is our obligation to run such a list (and
for sure we should cooperate with the others).
 
> >    http://blends.debian.net/liststats/authorstat_pkg-grass-general.png
> 
> grain of salt: you must put that graph in perspective with the rise
> of the OSGeo movement, which often informally communicates things which
> would otherwise find a home here. Or maybe Frankie is just so damn
> efficient and everything works that we don't get many complaints :o)
> aka determining causation is a difficult call.

Sure.  Such graphs do not really tell a whole story.  There are often
more than just one simple explanations and BTW, I'm confident that
Frankie is really efficient.  However, please mind the run-over-by-bus
effect.  And it actually does not need an accident to lose developers.
If you remember that developers become parents and need to care for
children more then for Debian a one man driven team might be in danger
to become ineffective.
 
> I congratulate you on that, by any measure of success that is a healthy
> situation to be in.
> 
> but does FOSS 4 Med have an equivalent to OSGeo already filling that niche?

I admit I do not know OSGeo enough to be able to do a real comparison.
I can confirm that there are FOSS 4 Med related lists where I was
subscribed in former times.  I learned that there is a lot of *really*
interesting *discussion* but just discussion without *implementation*
finally does not help.  The good thing in Debian is that we just have an
*implementation ready to run* the things we are discussing about and
IMHO this is a major point.
 
> > The tendency on the pkg-grass-general list is rather contrary.  I do not
> > want to say that it is just a question of the name of the list.  But I
> > know for sure from the example above (and others not mentioned here) that
> > a pure packaging list on Alioth is not a good entry point to start.
> 
> Perhaps that's where our miscommunication lies, I've still been thinking
> of it as mostly a packaging dev list, as opposed to say an osgeo-discuss at .
> (not that I have any strong feeling of what it should be, just observation
> on what it has been)

For me as a GIS outsider who just started to dive into GIS issues
(mostly via OSM but perhaps I might have other applications from my job
- think of epidemiology issues) this is really the point.  The fact that
we are continuosely CCing both pkg-grass lists (because the difference
is not really clear) and the discussion on pkg-grass-general which is
also quite packaging centric is something which did not really attracted
me and the suggestion is not only a rename (and move) but also shifting
the focus a bit more into the user direction.
 
> > That's for sure.  But did you gave users a reason to use Debian packages
> > and not Ubuntu, Fedora, OpenSuse, etc.? 
> 
> should I? If we are doing our jobs well Ubuntu gains and takes some credit
> for it, but I'm not too worried about that, they help us too. It's not a
> zero-sum game so I'm not too concerned about poaching users from other
> distros. I'd rather poach unhappy refugees from other OSs by way of better
> software.

Perhaps I was a bit short in my sentence above:  I'm not hunting for
users of other distros.  I know this will not work.  I want to make
Debian just the best distribution for a certain purpose and users will
realise if we succeed in doing so and in this way I would like to give
users a reason to switch.  It is a different think if I would say: "Hey,
use Debian!" (which I never would) or if some Debian user says to a
friend: "If you want to work on GIS I'd recommend to use Debian because
it is perfectly prepared for this purpose."
 
> > Do you think that you have given a random GIS user an explicite reason
> > why he should use Debian?
> 
> I'd avoid the words "I have", but in general I genuinely think we've done
> quite a bit of good work for the world in that regard.
> 
> IMO a solid application and flawless user experience is the best
> advertisement.  `apt-get install it-just-works` sells itself better than
> any advertising campaign.

I perfectly agree here regarding the packaging work.  We now just need
this kind of service part which can realise on the mailing list in
question which enables others to really realise this packaging work.
 
> considering live.osgeo.org (based in large part on our .debs) will go out
> to some thousands of professionals this quarter our work is certainly not
> going unused... well to be honest DebianGIS isn't getting much visible
> credit out of that right now, but I can certainly make an effort to change
> that if the group wishes.

I'm fine if others like OSgeo (or UbuntuGIS or whomever) makes use of
our work.  That's why we are working on Free Software.  In case all
these users realise that they need to help DebianGIS (the root) if they
want to make the final product (OSgeo or whatever) better, I'm fine with
it.  As I said I'm not clear about the role of OSgeo but I think I
learned in the past, that there is a lot of friction leakage between
Debian and derivatives (for whatever reason).  So the idea behind Blends
is:  Please do everything straight into Debian.  If you want to
*release* a *product* from this Debian (GIS) Blend and give it a product
name (like OSgeo) everything is fine.  But fix problems at the root,
care for inclusion of packages at the root and strengthen the root team
finally will be good also for the leaves.

Perhaps I'm beaten by some bad examples in my field.  For instance there
is BioLinux who call their product Debian based (sometimes they jump to
Ubuntu and back, but following always the same principle).  They do NOT
use *our* Debian packages but repackage everything and move it to non
policy conform packages to /usr/local.  Please don't ask me why - I
tried to contact them several times but failed.  Avoiding double work
for our friends (and I'd call derivatives friends in Open Source
movement) should be the goal and IMHO this can be done best if we join
at the root.  IMHO Debian is a perfect root and we should in our own and
also our derivatives interest propagate this idea.

> when the pools of people become shallow, often it comes down to the
> presence or absence of key personalities. FWIW I do hope some key OSM
> people decide to come our way, we can lower the barrier to entry, but
> it still up to them to come in.

Sure.  Nobody can or intends to force them to join.  It is about
lowering the entrance barrier.
 
> > > (I assume the Alioth admins are willing and able to provide the mbox
> > > file?)
> > 
> > I think so.
> 
> anyway now that I think of it there's a magic listserv URL you can use
> which sends you the entire archive as mbox, somewhere in my notes...

It will be possible somehow, yes.
 
> [zero spam]
> > But I admit this condition of yours is probably not fulfilled.
> 
> well, I was being a bit absolutist to make a point :) you are correct
> that perfection would never be satisfied, but it still feels good to
> try as we can.

Sure.
 
> > > *** the pkg-grass-devel list be renamed debiangis-commit (i.e. what it
> > > is) to stop the confusion there of humans posting to that list.
> 
> as pointed out, while current frozen packages point to this as the
> maintainer email, it can't change yet.
> understood.

Fine. 

> The current -devel list is non-human commits and automated bug report
> incoming. Personally I browse that one in a news reader and am happy to
> not have to dig through all the traffic for the occasional personal comms.

Fine - we also have such a commit list.
 
> The current -general list is the developer packaging chatter, low volume
> and easy to follow. Personally, I like this.

Perhaps my original attempt is wrong and we should not move -general but
rather freshly create debian-gis at lists.debian.org.  I'm following this
low volume list only for some monthes and might have missunderstood the
intention.  If you think that you stick to really *technical* packaging
only discussion on pkg-grass-general and pkg-grass-devel is just for
robots probably the later is the best idea.
 
> And the current general user oriented FOSS GIS chatter is handled by one
> of these many lists, which are up to something like 15,000 subscribers in
> all:
>   http://lists.osgeo.org/

Uhmmm, what a lot of lists ...
 
> Mainly I don't see the point of trying to change the focus of this list
> to compete with the established OSGeo lists for a general audience.

None of these lists seems to be Debian related and that's my point.
IMHO we need a list where GIS users (and potential developers) feel home
as well as where we can talk about strategies and DebianGIS project
development, contact to upstream etc. which is not actually packaging
centric.

Kind regards

       Andreas.

-- 
http://fam-tille.de