[linux-elitists] Linux OpenGL

D. Joe deejoe@raccoon.com
Tue Jan 17 09:47:26 PST 2006


Warren DeLano has been walking the open source walk developing
PyMOL--it has all the good things we expect from
free-as-in-freedom software: a reasonable license; a public
development process; the cultivation of a worldwide, active
user/co-developer community, supported through several kinds of
social software supplemented by face-to-face meetings.  It's
useful, being used and supported by academics and industry
alike.  He has staked his livelihood on this development model,
working on PyMOL full time, and so far seems to have made it
work.

If I read the history right, he spent at least some time in a
lab suffering the legacy of software contractually ensnared by
Borg-like proprietary restrictions.[1] Given history in the
field like this and given the license under which he continues
to release new versions of PyMOL, Warren DeLano seems like
someone who "gets it".  He is doing good work for anyone working
or studying in the associated fields of molecular biology,
structural biology, biophysics--and by extension--anyone who
benefits from their work (e.g., anyone who uses modern
pharmaceuticals, for starters).

Unfortunately, it seems his frustration trying to find a
well-integrated x86-64 hardware+free OS platform on which to
develop[2] coupled with a positive response from Apple[3] in the
face of pleas to get them better to support Stereo 3D
visualization[4] has got him singing Apple's praises (which is
fine, so far as that goes) but at the expense of talking down
Linux[4,5].  

This has a whiff about it of the old Unix Wars of yesteryear. 
The first round of it raised my eyebrows,[4] but after the most
recent round,[5] I figured it was time to say something.  [see
the forwarded message below].

Beyond the whole Unix-Wars-are-icky aspect, I'm bringing this
here, to linux-elitists, because it's an illustration of some
moment that goes to what we've been talking about earlier
regarding support of video cards that offer accelarated OpenGL
performance--it's not just all about the games.[6]

Perhaps DeLano's pitch[3] to Apple may serve as a glimpse into
the same market for system integrators out there who want a
piece of the pricing premium that Apple is now poised to take.

-- 
D. Joe


[1] tip of the iceberg mentioned e.g. in passing here 
    http://nmr.cit.nih.gov/pipermail/xplor-nih/2003-January/000035.html
[2] http://sourceforge.net/mailarchive/forum.php?thread_id=9190005&forum_id=60
[3] http://sourceforge.net/mailarchive/forum.php?thread_id=8681419&forum_id=60
[4] http://sourceforge.net/mailarchive/forum.php?thread_id=6551238&forum_id=60
[5] http://sourceforge.net/mailarchive/forum.php?thread_id=9473432&forum_id=60
[6] http://zgp.org/pipermail/linux-elitists/2005-November/011365.html
    http://zgp.org/pipermail/linux-elitists/2005-November/011364.html
    

----- Forwarded message from "D. Joe Anderson" <deejoe@iastate.edu> -----

Date: Mon, 16 Jan 2006 11:26:05 -0600
From: "D. Joe Anderson" <deejoe@iastate.edu>
To: pymol-users@lists.sourceforge.net
Subject: Re: [PyMOL] Quadro-based 3D Mac Feedback

On Sat, Jan 14, 2006 at 04:51:21PM -0800, Warren DeLano wrote:
> The Quadro FX 4500-based PowerMac G5 is Steve Jobs' timely and direct
> answer [...]
> The UNIX-based Mac has become the robust, low-cost, well-supported, and
> high-performance replacement for SGI workstations that Linux never
> delivered despite nearly a decade of opportunity and effort (including
> our own).  Why does Linux fail with visualization?
[...]
> History shows that it takes an integrated hardware & operating systems
> vendor like Apple (or formerly SGI, or Sun) to deliver and maintain
> trouble-free OpenGL under UNIX.  Apple is the world's leading UNIX
> vendor, and unlike with Linux, Mac hardware and operating systems are
> continuously integrated with cutting-edge OpenGL graphics cards. 

> Has Apple in fact given us what you asked for?  If so, then have you
> followed through with purchases?  If not, then what are you waiting for?
> Is something crucial still missing?  Please share your thoughts.  We
> guarantee that Apple and others will hear them.

Despite the OpenGL workstation niche having been a yawning chasm
for several years now, it took the concerted begging to get
Apple to respond to that niche.  While congratulations in
getting Apple to respond are still in order, this blatant
shilling contrasts jarringly with the open source context in
which PyMOL is developed, distributed, and supported: PyMOL is
attractive in part because it promises long-term, flexible
access to the tools to do our work, without users falling into a
vassal role with respect to one industry-dominant company.

Launching such a hyperbolic cheerleading campaign may belie the
stability of the support one might expect from Apple for this
technology, if one feels such campaigns are necessary to keep
that support alive.

It seems premature, at best, to extrapolate from Apple's
welcome, but very few and very recent, product releases to
describe their support as "timely", "robust", "well-supported",
"trouble-free" and "continuously-integrated".  Only with time
can one make those claims fairly.  It seems they and their
developer and user communities will have their hands full with
the upcoming hardware architecture migrations, during which
support for many different niches might lag behind or fall by
the wayside.  While one may hope they can pull it and keep it
all together, let's not get ahead of ourselves: The list of
Apple's predecessors in this niche, rather than being
complimentary to Apple should serve as a sobering reminder of
the difficulties Apple faces, and as a warning against hitching
one's scientific productivity yet again to just one star.

As for specifics of our own purchases, our purchase cycle had us
replacing hardware early last summer, before the indicated
hardware was available.  So, it doesn't seem to be a question of
"what are you waiting for" so much as asking Apple "what took
you so long?".  For this round, their support came too late,
even though they've been trading for years on their claims of
support for creative endeavors (cf publicity of the Genentech
orders of the original G4 iMac).  Our next crack at it will
probably come well after the migration to a Intel-based
architecture is complete.

As it is, we had to wait a year more than I'd have liked for
funding, which put us in the position of having to navigate the
replacement of hardware alongside a major release of the
operating system (MacOS 10.4).  That brought with it additional
support problems, notably a lag (or complete drop in) support
for MacOS from other third-party developers, which issues we're
still working through.  

Whatever the problems behind OpenGL support for Linux,
entreaties to abandon it will not help convince Linux
integrators or graphics vendors to support this niche.  In terms
of choice and pricing, this will benefit neither Linux users nor
Mac users.  Rather, what it does is re-animate the long dormant
internecine Unix Wars of the days of old.  We've been through
that, and we know how it turns out.  

For those who missed it the first time, it doesn't benefit
Apple, nor users of any other Unix-like operating system.

-- 
D. Joe

----- End forwarded message -----

-- 
-- 
D. Joe Anderson          http://www.etrumeus.com/~deejoe
". . . we could end up with a situation where it's legal to own
hardware that automatically fires bullets really fast but
illegal to have software that automatically fires bits really
fast. -- Jenny Reiswig"





More information about the linux-elitists mailing list