Call to contribution for the '2020 FLOSS roadmap'

[Posted May 8, 2009 by ris]

The 2020 FLOSS roadmap is a collaborative roadmap and with predictions on how the free / libre / open source software ecosystem could evolve over the next ten years. The first version was published last December, but the call for contributions for version 2 is currently open.

Call to contribution for the '2020 FLOSS roadmap'

Posted May 9, 2009 12:03 UTC (Sat) by sebas (guest, #51660) [Link] (4 responses)

Has anybody heard of this before?

My first reaction is that it might be potentially very useful to provide some cross-project communication, so we know what's going on in other Free Software projects and gather new ideas for things to drive further across the whole Free Software stack.

Within KDE, we're currently creating a high-level overview of what's going on across the whole project. This includes answers to questions like:

- Which big things have we achieved over the last years (roughly since the release of 4.0)
- What are the Next Big Things we want to achieve
- What are the major pain points we need to tackle to achieve these goals

Since nobody is an island, I'd be encouraging anybody to do the same and communicate it, so bits and pieces can be picked up by others as well as pain point often cross the borders of single projects.

Back to this FLOSS Roadmap, could this be the kind of tool to share this information?

This is also a nice thing for the press to put Free Software more in the spotlight, and for 3rd party vendors' product planning and R&D.

Call to contribution for the '2020 FLOSS roadmap'

Posted May 10, 2009 0:45 UTC (Sun) by zenaan (guest, #3778) [Link] (2 responses)

TinFoilHat-worthy if you ask me.

"Cap-Digital - Business Cluster for digital content" - site down, in French.

SYSTEM@TIC PARIS-REGION - www.systematic-paris-region.org/en/index.html
"A strategy combining key technologies and growing target markets".

Perhaps a VC outfit, presented in a community way, I'd say. May be babelfish can give you a better overview.

There's a smell about this. Something doesn't _feel_ like it's fully disclosed.

Call to contribution for the '2020 FLOSS roadmap'

Posted May 10, 2009 1:32 UTC (Sun) by jordanb (guest, #45668) [Link]

It's an industry group for tech companies in the Paris metro region (Ile-de-France). I figured the phrase "ICT Cluster" was a mistranslation of some sort but googling for it I find more uses of the term from elsewhere in Europe. For instance:

http://www.lcictcluster.org/
http://www.wmictcluster.org/

It's clearly not a venture firm given that it includes companies like Alcatel and Groupe Bulle, not to mention the French Republic and Il-de-France governments.

Call to contribution for the '2020 FLOSS roadmap'

Posted May 10, 2009 8:12 UTC (Sun) by fermigier (guest, #12330) [Link]

System@tic and Cap Digital are business clusters (http://en.wikipedia.org/wiki/Business_cluster), financed by the French State and the Paris Region local governments.

No hidden agenda. Everything is public (including the contract between the clusters and the state).

More information in French: http://fr.wikipedia.org/wiki/System@tic and http://fr.wikipedia.org/wiki/P%C3%B4le_de_comp%C3%A9titiv...(France)

Call to contribution for the '2020 FLOSS roadmap'

Posted May 13, 2009 7:53 UTC (Wed) by jplaisne (guest, #3827) [Link]

Thanks a lot for your understanding.

Please consider this Roadmap as a new tool for FLOSS advocacy from FLOSS communities towards decision makers, journalists, VCs, governmental orgs etc. We designed it in order to have some kind of global predictability which will reinforce trust in FLOSS worldwide.

Of course we may be wrong on some points but every year until 2020, we will reevaluate our positions and check progress, failures or new trends. This is the reason why we are avidly looking for comments and enrichments to our first release.

This document is written as any good software i.e. in an open and collaborative manner. This means that this is open to all contributors and if you have a specific Theme to develop (e.g. FLOSS and Human Interface for the Future), just propose and we will offer you some dedicated space for you and your contributors to work on.

Kind regards,
JPL.

Call to contribution for the '2020 FLOSS roadmap'

Posted May 10, 2009 20:51 UTC (Sun) by shapr (subscriber, #9077) [Link] (9 responses)

I'd like to see some sort of automatic framework for taking advantage of multiple processors.

Right now, only a few of the gnu utilities can use multiple CPUs. Newegg offers four core desktops already, it's not too long till we see six and eight core desktops.

Call to contribution for the '2020 FLOSS roadmap'

Posted May 11, 2009 3:45 UTC (Mon) by jordanb (guest, #45668) [Link] (8 responses)

Surely the 'automatic framework' is the Kernel's scheduler? If you have four CPUs and four active processes, they will each likely get their own CPU already.

For parallelism in individual programs, I don't think there can be anything 'automatic.' Being parallel is an architectural issue that will affect the basic structure of the program.

Also, for most GNU utilities I don't think CPU parallelism won't make any difference. 'ls' isn't CPU bound, it's IO bound. The only one that pops into my head that is CPU bound and used quite a bit is GCC, and it supports parallelism for many jobs with the proper configuration.

Firefox isn't even remotely parallel of course, even though threaded tabs would be an incredibly good idea. But that's just the very tip of the iceberg when discussing architectural problems in Firefox.

Call to contribution for the '2020 FLOSS roadmap'

Posted May 11, 2009 6:01 UTC (Mon) by nix (subscriber, #2304) [Link] (3 responses)

GCC doesn't really support parallelism yet (other than 'running multiple copies' which works great as long as you don't have a decent make tool that can run GCC in -combine mode over all of them at once).

But it will.

GCC

Posted May 11, 2009 11:39 UTC (Mon) by alex (subscriber, #1355) [Link] (2 responses)

Well running multiple copies is a fairly simple and easy way to use those CPU cores while letting the OS get on with its thing of doing that efficiently. As soon as you add multiple processes that are directly dependent on each other your open up a whole can of worms.

I think the original commenter was after a frame-work to make the parallelisation of code "easy". Intel have realised that now the Ghz race is over they need to keep multiple-cores maxed out if people aren't going to complain about being short changed on CPU power. They have come up with TBB (http://en.wikipedia.org/wiki/Intel_Threading_Building_Blocks) as an attempt to smooth over these issues. I am yet to be convinced this is a magic bullet though.

Scanning the GCC paper I noticed it's called WHOPR, seems very close to the War Games master computer :-)

GCC

Posted May 11, 2009 18:37 UTC (Mon) by nix (subscriber, #2304) [Link]

WHOPR's job isn't autovectorization: it's allowing the compilation process
to be parallelized when you run GCC over all (or many) of the translation
units in a program with --combine (and -fwhole-program for non-.so's), to
allow inter-translation-unit optimizations. (--enable-final in KDE does
something vaguely similar.)

In this situation a *single instance* of GCC has to be able to do things
in parallel: running with make -j BLAH is not sufficient.

GCC

Posted May 11, 2009 18:50 UTC (Mon) by oak (guest, #2786) [Link]

> I think the original commenter was after a frame-work to make the
parallelisation of code "easy".

Something like the Eclipse Parallel Tools Platform:
http://www.eclipse.org/ptp/
?

There was a nice presentation about it (and its OpenMP/MPI support) in
2008 Tracing Summit:
http://ltt.polymtl.ca/tracingwiki/images/d/d5/EclipsePara...

Call to contribution for the '2020 FLOSS roadmap'

Posted May 11, 2009 18:48 UTC (Mon) by oak (guest, #2786) [Link] (3 responses)

> Firefox isn't even remotely parallel of course, even though threaded
tabs would be an incredibly good idea.

Google Chrome runs tabs as separate processes. That way you can also
better monitor their memory usage.

There's even a comic explaining it:
http://www.google.com/googlebooks/chrome/index.html

(Done by Scott McCloud who's pretty well known...)

Call to contribution for the '2020 FLOSS roadmap'

Posted May 12, 2009 2:32 UTC (Tue) by dlang (guest, #313) [Link] (2 responses)

and I'm still waiting for a release that can run on any of my linux systems.

I'm not going to go out and buy a copy of windows just to run their experimental browser.

Call to contribution for the '2020 FLOSS roadmap'

Posted May 12, 2009 11:35 UTC (Tue) by xanni (subscriber, #361) [Link]

CrossOver Chromium

You're welcome.

Call to contribution for the '2020 FLOSS roadmap'

Posted May 12, 2009 12:21 UTC (Tue) by wookey (guest, #5501) [Link]

Yes, that comic made it all sound rather good. I was particularly sold by being able to tear off tabs and put them in different windows - I wish I could do that every day (maybe I can actually? - I currently use epiphany).

But then I see that there is no Chrome for Debian/Linux. Oh well, that's no bally use then. I knew there was a reason why I hadn't taken any notice of it to date :-)