[Gnash-dev] Re: Gnash + OLPC

[John Gilmore]

Hi folks!

> Actually, almost all OLPC developer members/volunteers choose to steer
> people towards using Gnash instead of Flash. As much as possible, the
> OLPC team wants Gnash to be targeted instead of Adobe Flash, and this
> has been very consistent in all the discussions I've had with active
> volunteers and team members so far.

The OLPC staff and volunteers have a strong preference for free
software.  The support gang has a preference, but less strong.  The
only reason that *any* of them recommend the proprietary player is
because it works in many cases when Gnash doesn't.  They're balancing
their free software preference with their preference for functional
software.  This is expectable when the free product is playing catch-up.

Gnash team, be glad that we have a distro that's shipped half a
million computers with gnash!  Adobe now lets distributors sign a
simple license to distribute the "costless" proprietary player -- end
users no longer have to individually download it from Adobe.  If OLPC
wanted to ship the Adobe player, I'm sure Adobe would be more than
happy to cut the biggest audience for Gnash off at the knees.

OLPC support has not gotten high priority from the gnash team, and I
wonder why not.  Whose bugs and feature requests should get higher
priority?  We have a few funders, and their desires deserve respect.
Our volunteers work on whatever suits them; they go away if we try to
push them into working on what doesn't suit them.  But beyond those
constraints, OLPC users are our biggest user base, and I think they
deserve better support from the team.  And OLPC is a strategic wedge
for us; if we can keep Adobe out of the distro indefinitely, by
evolving gnash to better and better meet the needs of the OLPC
community, then we have a firm base from which to expand into the
broader market.  This is the classic way that ANY project or business
gets traction in the world: capture a niche, live in it, then build
out from there.

>                                     It's just that with the
> line that was crossed when OLPC & MS partnered up, I guess the
> agreement that only FOSS be used with the XO is not there anymore.

The whole MS-OLPC thing has been exaggerated by the media.  None of
OLPC's development work goes into MS-based solutions.  All of their
work is on their existing Linux-based OS.  This is extremely unlikely
to change.  A couple of folks at the top (Nicholas and some less
visible) have bent to the pragmatism of the countries who are paying
the bills by buying the laptops.  Those countries don't want to be
stuck with unsupported computers if the OLPC organization fails -- or
sticks with its foolish quest to slowly and painfully reinvent the
concept of a GUI, badly.  Thus they want the *option* to run Windows.
And if kids using Windows on the XO do better than kids using Sugar,
most sane countries would probably educate their kids with Windows.
Free software is not their battle to fight, it's ours.

> of the pain in trying to get sound in Gnash to work on XO's latest
> signed test OS build (767) (which will be the basis for the stable
> build 8.20 which will come shipped in the upcoming XO units that will
> ship in the new Give 1 Get 1 program that's going to run this
> November).

Right.  I reported such a sound problem to the gnash team last week,
but nobody jumped on it.

I believe the most basic problem is that most flash sounds are encoded
in MP3, and OLPC can't ship the patented MP3 decoder.  It is possible
to encode sounds in WAV or ADPCM in Flash files, which are free
formats, and I suspect that free gstreamer supports both (has anyone
tested??!).  Perhaps the free Flash development tools need patches
that enable authoring with WAV and ADPCM sounds.  But I also found
that in one case, gnash on the XO produced loud white noise, which is
a more serious bug.

> Also, there's the issue of AS3/AVM2. Having done tests with a
> benchmarking app we made to test AS2 vs AS3 performance (both
> published for Flash 9, using nearly identical algorithms), the AVM2
> (AS3) version definitely trumps the AVM1 (AS2)'s performance. And this
> is extremely relevant because we need to squeeze every bit of juice
> out of the XO's limited memory & CPU.

I suspect that this is a red herring.  The best games use simple
computations to make powerful visible effects.  We have not by any
means exhausted the ability to make incredibly useful and engaging
stuff while doing only small amounts of computation.  Look at the core
of SimCity, for example; it's tiny and fast.  It ran on the C64 and
on the original Macintosh, in realtime, with 1/50th the CPU power
the OLPC has!

> As such, it would be really fantastic if some of you guys could also
> sit in at OLPC's developer mailing list at
> http://lists.laptop.org/listinfo/devel (address@hidden)

I monitor OLPC's devel list, and forward anything gnash-related to the
gnash-dev list.

OLPC is shipping the very latest public release of gnash, in a
customized RPM built by the gnash project leader (Rob).  OLPC's
release schedules slip a lot, so it's hard to time subsystem release
schedules like gnash's, so that they'll hit an OLPC release schedule.
OLPC is trying to synchronize with Fedora's release schedule, which is
similar to Ubuntu's schedule, which should make things somewhat
easier.  But if you find yourself complaining that nobody's running
the cvs/git/bzr/whatever trunk version of gnash, it's time to bend your
efforts toward cutting a release.  At the moment, the snapshots don't
even build on most platforms.

        John