[p2p-research] Fwd: launch of code depository for authentic free network services

Sun Apr 12 09:45:32 CEST 2009

Dear Sepp,

the following is an announcement on the launch of a free network service
depository, by the autonomous group, which seems an important development
for open standards, but too technical for me to report on.

if you ever have time, I would appreciate if you could at least announce it,
say something about their principles for inclusion and mention the 3
services that seem to pass the test, listed here just below,

Michel

[1] Fossil SCM
http://www.fossil-scm.org

[2] Bcfg2
http://bcfg2.org

[3] dotCloud
http://bitbucket.org/dotcloud/dotcloud/wiki/Home

---------- Forwarded message ----------
From: Daniel Clark <dclark at pobox.com>
Date: Apr 12, 2009 10:26 AM
Subject: [Autonomo.us] = code.autonomo.us: take 2 proposal version 1.0 =
To: Discuss at lists.autonomo.us

= code.autonomo.us: take 2 proposal version 1.0 =

== History of take 1 ==

The first version of code.autonomo.us was started very quickly at the
Libreplanet 2009 conference, in the spirit of getting hacking done; it
created some useful discussion that has lead to this proposal for a new
concept for a second version of the service, one that I hope is more in
line with the ideals of the autonomo.us project.

The first version was based on the idea of using Trac plus a project
that provided some amount of automation around the creation of new Trac
projects. The plan was to work towards exposing all of the used Trac
code, configuration and data (except for authentication tokens), so
people could in theory replicate the entire system and also host on
computers they controlled if they decided to do so; however the fact
that Trac was not designed to be a free network service would have made
it cumbersome, and other issues such as non-support for any version
control systems other than SVN in the stable version of Cluemapper, the
Trac project software chosen after evaluating 3 alternatives (none of
which supported anything but SVN either), made it unattractive for many
people. Also, it came to light on the mailing list that a closely
aligned project, the Open Knowledge Foundation, was already running such
a service (that was not found in the initial evaluation period) called
kforge or knowledgeforge, so the added value of having autonomo.us host
such a service was diminished.

== Take 2 proposal: overview ==

The proposal is for take 2 to be based on software that actually
conforms to our principles on a technical level (although sadly in no
cases using our chosen license to avoid being screwed, AGPLv3 - if there
is software I haven't found that is AGPLv3 I would love to hear about
it). This has 3 elements:

1. Require that each project have ultimate control over its services, by
requiring that each project control its own DNS domain. We will request
but not require that each project have "hosted by code.autonomo.us"
(with a link to http://code.autonomo.us) displayed on each web page
somewhere, or at least on the front page. Logos for this would be welcomed.

2. Only use software that allows users to easily maintain control over
their project's data, configuration, and code; in practice without legal
shenanigans this would seem to mean that they can work with their
project the same way locally, using only resources on a machine they
own, as when on the network, and are only dependent on the network as an
easy-to-use display mechanism for other people (and are not dependent on
the continued existence of code.autonomo.us for that purpose, as they
can switch their DNS to use another server at any time.)

3. Expose the entire configuration of code.autonomo.us, so that users
are also not dependent on code.autonomo.us for its aggregation features
(as they could easily replicate them themselves). This information will
be included in the code.autonomo.us Fossil SCM [1] instance, which will
be the only project that is actually hosted by the code.autonomo.us
service using an autonomo.us domain. Plan is to do this with Bcfg2 [2]
and maybe dotCloud [3].

[1] Fossil SCM
http://www.fossil-scm.org

[2] Bcfg2
http://bcfg2.org

[3] dotCloud
http://bitbucket.org/dotcloud/dotcloud/wiki/Home

== Take 2 proposal: technical details ==

=== code.autonomo.us meta-project ===

code.autonomo.us website becomes a Fossil SCM instance
* Main page is a DOAP [4,5] aggregator; very simple, just a script that
runs from cron, looks for the latest page (say
code.autonomo.us/admin/projects-list.wiki) that is signed by one of a
defined set of GnuPG keys, and then processes the DOAP URLs in that
document, and spits out an HTML page that gets checked back in to fossil.
* Page will indicate type of hosting
* Will need to use PGP-signed DOAP [6]
* PGP-Singed DOAP page will control other things hosting does as well
(eg make backup copies of other DVCS repos perhaps)
* All scripts and config (eg Bcfg2 repo) will also be in this Fossil SCM
repo. With this one file, you will be able to completely recreate the
entire service (except for some auth info).
* Small number of already externally linked to pages can have redirect
at the apache level (look through logs and/or use google to find these)

[4] DOAP: Description Of A Project
http://trac.usefulinc.com/doap

[5] DOAP A Matic
http://crschmidt.net/semweb/doapamatic/

[6] PGP Signing FOAF Files (should be applicable to DOAP as well)
http://usefulinc.com/foaf/signingFoafFiles

=== Hosted projects ===

With both choices you are required to control your own domain name.
Explain why this is important for autonomy (I think someone on the list
wrote a good summary of this at one point; just need to find it). We can
point to some registers; gandi.net, namecheap.com, etc (and also
recommend against some, like GoDaddy).

==== Recommended choice ====

All-in-one with Fossil SCM; this will include replication across at
least 3 geographically distributed servers (need to contact FSF France
and Joyent to accept their offers; I think this will get us the US
(FSF), Canada (Joyent), and Europe (FSF France) to start off with).

For communications I almost think we should set up usenet as the main
service with a gateway to mail; usenet's architecture seems closer to
the FNS concept than mailing lists.

Where possible things that are not integrated into Fossil SCM (such as
usenet / mailing list archives and configuration) will be inserted into
the hosted project's Fossil SCM repo with simple scripts run from cron.
The idea is that everything having to do with a project will be on each
developer's personal box, since it will be there as a side-effect of the
process of checking in and out code. The project should have no
dependence on code.autonomo.us's continued existence at any given moment
to retain all of its code, documentations, and bug database
information (perhaps behind a few hours, which shouldn't be crippling to
any project).

We may want to do some kind of load balancing / geographical redirection
service to make code.autonomo.us automatically redirect to or
transparently use us.code.autonomo.us, fr.code.autonomo.us,
ca.code.autonomo.us.

Reason that Fossil SCM would be recommended over a la carte service is
that a big part of what makes web services so attractive is ease of use;
Fossil SCM is very easy to use locally (one executable, one data file,
build-in web server, works on GNU/Linux, Mac OS X, and Windows), whereas
using a la carte service options locally would require significantly
more local software setup.

==== A la carte service ====

Mix-n-match:
* DVCS: Your DVCS of choice (Bazaar / bzr, darcs, git, Mercurial / Hg,
monotone, svn / svk) hosted elsewhere (list the common sites -
bitbucket.org for Hg, launchpad.net for Bzr, etc.) with transparent
proxy from the domain of your choice, hosted by code.autonomo.us (so you
can tell people to check out from for example http://hg.yourproject.org
instead of http://yourproject.bitbucket.org; the transparent proxy would
be handled by code.autonomo.us, however since you control
yourproject.org you chould change that to a server you control at any
point).
* Wiki: ikiwiki [7], Ajatus CRM [8] or Fossil SCM
* Bug/Ticket tracking: Best Practical SD [9] or Fossil SCM
* Search across resources: TBD

[7] ikiwiki
http://ikiwiki.info/ ikiwiki

[8] Ajatus CRM
http://www.ajatus.info

[9] SD
http://syncwith.us/sd/using

==== Totally external recommendations ====

Mention that if you really want to use Trac, look at KnowledgeForge [10]
(and also eventually config.fsf.org if you want a do-it-yourself
example, as well as whatever bkuhn sets up for SFLC).

Mention the more freedom-respecting traditional services (eg the FSF's
Savannah, FSF France's service I'm forgetting the name of at the moment;
please make suggestions...)

[10] KForge (project of The Open Knowledge Foundation)
http://www.knowledgeforge.net/

--
Daniel JB Clark   | Sys Admin, Free Software Foundation
pobox.com/~dclark | http://www.fsf.org/about/staff#danny

_______________________________________________
Discuss mailing list
Discuss at lists.autonomo.us
http://lists.autonomo.us/mailman/listinfo/discuss

-- 
Working at http://en.wikipedia.org/wiki/Dhurakij_Pundit_University -
http://www.dpu.ac.th/dpuic/info/Research.html -
http://www.asianforesightinstitute.org/index.php/eng/The-AFI

Volunteering at the P2P Foundation:
http://p2pfoundation.net  - http://blog.p2pfoundation.net -
http://p2pfoundation.ning.com

Monitor updates at http://del.icio.us/mbauwens

The work of the P2P Foundation is supported by SHIFTN,
http://www.shiftn.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listcultures.org/pipermail/p2presearch_listcultures.org/attachments/20090412/e2575694/attachment-0001.html>