[Parl-devel] Global vs. local community building

Fri Feb 28 13:50:41 UTC 2014

Thanks for sharing, Franklin,

Comments below quoted chunks...:

Quoting Franklin Weng (2014-02-28 07:34:01)
> To be honest I'm here not to discuss about debian-parl.  But here is a 
> very interesting topic, about global and local community needs.  Also, 
> in Jonas' mail above and the discussion between us, we were talking 
> about the same question:  When we have a target and would like to use 
> Debian as our solution, should we strictly follow Debian way and fit 
> anything into Debian, or can we give Debian some "addon", or even 
> "dirty hack" to solve our problem?

As distro designers We are bound not only by needs but also by 
abilities.  Our target can be defined as "getting the product done" but 
more wisely should be "maintaining the product".

As I understand it, EzGO! and DebianParl is similar in that our 
development teams are *very* small.  Lack of resources is a reason to be 
modest about amount of customizations - depending on the product...

If product is a show-case how Linux is nice looking and in principle 
able do a lot of stuff (like EzGO! was originally intended, I believe), 
then maintainance is less important: A demo system does little harm if 
it stops working over time - developers can just make a new hack from 
scratch based on some other newer demo system of others (e.g. a Kubuntu 
Live-CD), and users can just throw away the old CD and use the new 
(there is no user data involved).

If product however is production systems - deployments for teachers and 
students of schools, og parliamentary workers, then it is serious if a 
bug in either included code or own hacks blow up in the face of users: 
We should be no more bold in our designs than to be able to sustain it: 
Maintain the usability of our product for our users.

A Debian Pure Blend is a true subset of Debian, which implies that it 
inherits comon features of Debian.

Debian is maintained by an active and high quality developer community.  
The parts of your Debian Blend which you did *not* hack yourself (i.e. 
all of it for a Debian Pure Blend) is maintained by that community, also 
for users of your system.

Debian can be upgraded from one version to the next.  So can your 
system.

> I'm actually more interested in another debian project: debian-edu, or 
> skolelinux.  I don't know if all parliaments in all countries would 
> have the same needs or not.  I don't think so.  Anyway, for a project 
> like skolelinux, there will be definitely the common part and the 
> special part for education in different countries.  Reviewing the 
> history of debian edu, I think I can understand why it is not pure 
> enough.

No two persons use computers exactly the same way.

No two parliamentary workers - or school teachers or students - either.

But people have commonalities in usage patterns, and therefore benefit 
from systems pre-fitted a certain way.

The issue of purity is orthogonal to that: That is an issue of balance 
by the designers of a blending effort between reaching their goal 
compared to getting functionality implemented in Debian generally rather 
than locally as a hack.

> Last year in a purchasing project ezgo "defeated" Microsoft and became 
> the only operation system pre-installed in 10,000 computers.

Congratulations!

That was just the beginning, however: How sustainable is that concrete 
deployment?  Who maintains it?  How maintainable is the design of the 
system deployed?  Is there an active development and user communities?  
So those communities have good leaders (i.e. skilled in *guiding and 
inspiring a community more than other more technical skills)?

Do the communities mix with related communities - e.g. gather at 
conferences, og exchange ideas and experiences?  Do the communities have 
a strong identity (not only corporate, but adopted by participants)?

What learning has come from that project?  Is that learning shared with 
other communities?  Perhaps even internationally (i.e. translated to 
english - or french, spanish or russian)?

> ezgo "dirty hacked" Fedora(2007) / ubuntu(2008-2011) / Kubuntu 
> (2012-2013).  We added many, many good applications / public resources 
> for educations into it.  There are also some special applications like 
> Anki, PhET, Sankore (electric whiteboard), wiimode, ... etc.  We used 
> DRBL for deployment. We wrote some special applications like restoring 
> the student environment after each class. 

Yes, I know a bit about EzGO!  About your design concept about the left 
wing as symbol for copyleft software and the need for two wings to fly.

Well, as other mammals we have a urge to make children, even if that 
locks us down for some twenty years from our freedom to explore life on 
our own.  Your EzGO! is similarly locked down by all the inventions 
you've added to it - all the "dirty hacks".  Debian blending mindset is 
that only when hacks are reshaped as Debian packages and those packages 
adopted into Debian and those packages have aged to become stable (yet 
still actively maintained for future stabilization as well) are you 
truly free again.

So the more you excitedly share with me that you have invented, the more 
you are telling me that you are burdenin yourself with - and that your 
users rely on *you* (not Ubuntu or Debian or Fedora) to fix if things 
blow up in their faces.

That's why I (in my previous mails discretely to you) brag about how 
*tiny* DebianParl is. ;-)

> There are still some other important problems we need to solve, mainly 
> for people instead of system itself, however that's why I started to 
> study debian-edu and skolelinux.  Skolelinux has a good performance in 
> the computer classroom in Taiwan, however there are still something we 
> need to customize to fit the environment here, like IP netmask range, 
> font problem, ... etc.

I dare challenge you and say that you have still a lot to un-customize.

You may think that in order to streamline rollout of deployments, you 
need to hack.  But that is not necessarily true, and hacking is an enemy 
of long-term sustainability.

Hacking is fun.  I do it all the time.  But hacking and large-scale 
deployments don't go well together.

Hack to *explore* and *learn* about a system, but do the opposite - 
purify - when designing systems for larger scale deployment.  That's my 
advice.

> Therefore, go back to the real topic.  We'd like to develop a total
> solution for schools in Taiwan, and we're studying skolelinux now.
> However, in Jonas' opinion, we should follow the pure Debian way so that
> more debian developers can involve and contribute to our needs.  I 
> love this thought, just that I wonder if pure blends could satisfy all 
> our needs or not.

Given enough patience Debian Pure Blends can satisfy you.  But if driven 
by short-term deadlines just continue to hack away as you've done very 
well till now.

> Well, actually I seem not to give any constructive comments to this
> problem.  I'd like to hear thoughts and comments from all of you.

You gave great provocations for reflections.  Thanks for that!

Beware that this list is very young, and is most likely read only by few 
people.  I appreciate your posting here (I suggested that specifically 
for your challenging remark about parliamentary needs not universally 
equal). You may want to consider discussing Debian Pure Blends more 
generally at its own associated mailinglist:

  https://lists.debian.org/debian-blends/

 - Jonas

-- 
 * Jonas Smedegaard - idealist & Internet-arkitekt
 * Tlf.: +45 40843136  Website: http://dr.jones.dk/

 [x] quote me freely  [ ] ask before reusing  [ ] keep private
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 966 bytes
Desc: signature
URL: <http://lists.alioth.debian.org/pipermail/parl-devel/attachments/20140228/91b4e311/attachment.sig>