[Discussions] Quick & dirty report from the OWF 2010 think tank on Open Forges
Olivier Berger
olivier.berger at it-sudparis.eu
Thu Oct 21 13:58:58 CEST 2010
(exported from my org-mode notes)
Report from the OWF 2010 think tank on Open Forges
==================================================
Author: Olivier Berger et al.
Here are notes taken by me and colleagues from COCLICO during the think
tank on Open Forges session held at the Open World Forum 2010.
The most interesting part was probably the panel discussion on forge
lock-in, IMHO.
Maybe these notes will be of interest to some.
Other people have reported at :
- Giulio_Iannazzo in [OWF 2010]
- Luis Canas : [Back from the Open Forges Summit 2010]
[OWF 2010]: http://blog.codendi.com/?p=1001
[Back from the Open Forges Summit 2010]:
http://sanacl.wordpress.com/2010/10/04/open-forges-summit-2010/
Table of Contents
=================
1 Morning session - Trends in forges architecture - use and evolutions
of state-of-the-art forges
1.1 11.00-12.30
1.1.1 Qualipso factory: a new generation forge - Christophe
Bouthier - INRIA
1.1.2 Essentia ESP forge: Gopi Ganapathy
1.1.3 CloudBees, a forge in the Cloud: François Dechery
1.1.4 OW2 forge migration to NovaForge - Emmanuel Rias,
NovaForge R&D leader, Bull
1.2 12.30-14.00 Cocktail-Lunch and networking
1.3 Afternoon session - Forges Interoperability and project data
freedom: data lock-in issue and interoperability between forges
1.3.1 14.00-15.15 Panel
1.4 15:15-16h00 Experiences in interoperability: cases in
interoperability and project migrations
1 Morning session - Trends in forges architecture - use and evolutions
of state-of-the-art forges
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1.1 11.00-12.30
================
1.1.1 Qualipso factory: a new generation forge - Christophe Bouthier -
INRIA
-----------------------------------------------------------------------------
- Difficulty to build an ecosystem around a forge
- Goal: Make it easy to integrate new services in the new generation
forge
- Implement with Component model + framework of core services
- for the backend
- for the frontend (using GWT), looking like desktop app (looks like
Eclipse, IMHO)
- Java technology (Eclipse, Maven)
- Customizable forge because of a composition model of services
- Paths using to navigate the tree of composed services -> XACML (?) for
security
- Quite impressing UI demo/screencast IMHO (creating a SVN repository in
the context of a particular bug)
- slides : [http://www.slideshare.net/funckychris/qualipso-factory]
- screencast : [http://vimeo.com/15527478]
- Content negociation in the forge to serve RDF in addition of pages
(useful for annotation support)
- SPARQL query on the metadata : preview of resources inside the SPARQL
results (similar to what's proposed by OSLC)
- Q: who will be the early users ?
A: eating their own dog food at end of october
- Q: not asked: which APIs supported ?
1.1.2 Essentia ESP forge: Gopi Ganapathy
-----------------------------------------
- [http://www.essentia-corp.com/esp.html]
- Hosts communities like Jasperforge, OpenBravo
- links to CollabNet in the past ?
- hosted OpenOffice, Java, Net, ...
- Targetted to "commercial open source communities", with hosting
provisioned by companies (editors)
- Jaspersoft : 320 K users per day !
- Optimized for communities developping extensions to existing products
- 2 classes of features supporting : Development + Social (private
communities, private blogs for users...)
- Integrating a lot, but... closed source platform
- Q: Cost of provisioning a forge ?
A: 1000 US$ (on the cloud)
- Essentia Unify -> integration, interop ?
1.1.3 CloudBees, a forge in the Cloud: François Dechery
--------------------------------------------------------
- [http://www.cloudbees.com/dev.cb]
- HaaS : Hudson as a Service : provisioning of machines for the builds
(get rid of bottlenecks of the build machine capacity)
- Links to JBoss middleware engineering and Sun Glassfish
- Java PaaS (Platform as a Service) : dev at cloud (vs. run at cloud)
- Targeted to Java community mainly
- Runs over Amazon, Rackspace, Terremark, others
- Not provisioning another full forge necessarily : integrating with
existing forges
- Private maven repositories
- Towards :
- more tests services (Selenium, Cucumber)
- QA, Sonar
- Cloud is interesting for builds : whereas other means of collaboration
may not require lot of processing power
- Q: lock-in ?
A: independance from underlying cloud providers
1.1.4 OW2 forge migration to NovaForge - Emmanuel Rias, NovaForge R&D
leader, Bull
-----------------------------------------------------------------------------------
- 2 central NovaForge deployments internally at Bull
- OW2 forge initially based on GForge
- Lack of roadmap for GForge
- Limitations of features (Java projects)
- Data migration tools developped, using Talend
- users, projects, roles
- bugs
- gforge -> mantis
- deliverables
- documents
- Talend connectors used for the migration will be contributed on
Novaforge soon.
- Migration difficulties == Human difficulties mainly
1.2 12.30-14.00 Cocktail-Lunch and networking
==============================================
1.3 Afternoon session - Forges Interoperability and project data
freedom: data lock-in issue and interoperability between forges
=================================================================================================================================
1.3.1 14.00-15.15 Panel
------------------------
Panelists:
- Moderator: Jeffrey Bates, SourceForge
- BerliOS community: Lutz Henckel, BerliOS Community leader, Fraunhofer
FOKUS
- FusionForge community: Franck Villaume, CapGemini
- Geeknet Representative : Scott Collison
- Codendi user: Laurent Charles, ST MicroElectronics
- Microsoft Codeplex: Jonathan Wanagel
Ideas discussed :
- Privacy regulations
- Social responsability of forges owners to backup the code
- Importance of the number of self hosted forges
- Market shares: 25% SourceForge, 18% Google Code => majority is self
hosted forges
- Ability to change from forge.
- Importance of ability to work on different forges and having a common
dashboard for all my projects.
- Need for sharing the workflow in addition to sharing the
data/documents/artifacts (i.e. the history and metadata)
- Is there a problem of locking ? Does it already happens ? Yes, with a
project called Mojo something
What is really the problem ?
1 - how to get all the data
2 - what to do with all those data, how to get back to a service with
the same service level
One big point is not only to get the data, but how to put it back in
another forge
- Impact on users, in addition to devs. when moving a project : great to
have control on URLs to not lose users
URL is very important because it's a part of the identity/image of the
project/community
- Heard : "Ohloh is a failure"
rationale was: "we had all those data about open source but no one
f*cking care",
and about the fact that the creator was selling it for something
like 100 times less than the price he gets when he sold its previous
company.
- Need for history of the project
- Interop issues when devs. belonging to different projects at once
- Mashups
- Interoperability among the forges : various aspects
- one is at the forge level (have a big dashboard of your different
projects in different forges)
- one is at the service level (use the best tool for the situation,
mix-and-match services in different forges)
- Also, interoperability is, to a certain level, a guarantee of freedom
for the developer, just like licenses.
- Interoperability of tools (dynamic interop): the forge should be able
to talk to other tools of the company.
- Peace of mind : backup (holiness of data)
- How to import and use the data
- Losing some data when moving (download history, for instance)
- Dynamic interoperability
- Who's caring ? : important but never a priority
- Research studies requires transparency
- Ownership of data
- Download data (stats) from SF.net soon freely available
- Lack of standards (who, how, when ? : igniting momentum ?)
- Open source tools for traceability of the workflow
"to get back your data, you NEED to have the software and its CODE to
be sure to have it all".
1.4 15:15-16h00 Experiences in interoperability: cases in
interoperability and project migrations
==================================================================================================
+ Coclico project - Olivier Berger, Telecom SudParis
"COCLICO project’s efforts towards better forges interoperability" :
- OSLC Core interesting to create a standard dump format
- Improve ForgePlucker and extend it to make it support this format
(JSON which uses RDF instead of forgeplucker's own format)
- Many other interesting / strange / irrealistic (pick your own)
ideas ;)
- Slides :
[http://www-public.it-sudparis.eu/~berger_o/weblog/2010/10/02/my-presentation-about-coclico-and-forges-interoperability-at-owf-2010/]
- long textual version :
[http://www-public.it-sudparis.eu/~berger_o/weblog/2010/10/04/coclico-projects-efforts-towards-better-forges-interoperability-long/]
+ OSLC - Steve Speicher - IBM Rational
[http://slidesha.re/aU8lDG]
- Interesting presentation on dynamic interoperability
- Everyone a bit tired, so not too much feedback
Best regards,
--
Olivier BERGER <olivier.berger at it-sudparis.eu>
http://www-public.it-sudparis.eu/~berger_o/ - OpenPGP-Id: 2048R/5819D7E8
Ingénieur Recherche - Dept INF
Institut TELECOM, SudParis (http://www.it-sudparis.eu/), Evry (France)
More information about the Discussions
mailing list