[Dirvish] The future of Dirvish

Keith Lofstrom keithl at kl-ic.com
Thu Sep 25 18:52:31 UTC 2008

On Thu, Sep 25, 2008 at 12:48:17PM -0500, Richard wrote:
> Yes, I know....I hiijacked the topic...   BAD RICHARD!
> Dave Howorth wrote:
> > Loren M. Lang wrote:
> >   
> >> <snip> It looks like dirvish-expire does not expire images in the right order when all
> >> images have expired.
> >>
> >> # dirvish-expire; dirvish-runall
> >> cannot expire server:default:20080423 No unexpired good images
> >> server:default 20080424 2008-04-24 02:04  +28 days == 2008-05-22 02:03
> >> <snip>
> > Looking at the code, I tend to agree. It appears to sort in increasing
> > order of date - as confirmed by Loren's output - but it treats the first
> > one as special - again as confirmed by Loren's output. Especially
> > interesting are the *two* warnings about the sort order!
> >
> > I haven't done any testing but at first glance this does look like a bug
> > to me.
> >
> >   
> What is the status of dirvish version 1.3.1?  The website still lists it 
> as 'experimental' and not ready for production.  I tried to look for a 
> changelog for 1.3.1 and I seem to be going in circles on the website.    
> OH!!!  ok.. I downloaded the 1.3.1 tar.gz and looked at the changelog in 
> there.    Hmm... 1.3.1 seems to be approaching an age of three years old.

Keith responds:

1.3.1 was intended as a vehicle for making improvements to the layout
of the code.  There is a lot of duplicate structure that I attempted
to combine, and I wanted to move away from the shell-script install
kludge and move towards a standard Perl installer like Module::Build.

What I learned was that I am not much of a programmer, and I do not
have time to learn enough to do it right.  I still hope to provide
a home for dirvish, maintain the shared server,  clean out the wikispam,
fix the problems that are simple to fix, etc., but we really do need
programmers that (1) use dirvish (2) respect the needs of other users
and (3) have the time to contribute.  That hasn't happened yet.

In my ideal world, some young programmer would take this on as a
volunteer activity and move it forward, drawing in others to help,
eventually taking administrative control of the website.  I will
keep dirvish alive until that person appears.

BTW, I consider 1.3.1 an experimental release, and after it reaches
some degree of maturity, it becomes the 1.4.x production release.
We obviously aren't there yet.    1.5.x would probably involve an
object-oriented and test-oriented rewrite, with translators from
old config files to whatever works best in the new format. 2.0.x
would be the production release of that.

> What about setting up a community or member edited FAQ that can get 
> updated with questions from the list.   Maybe a faq-o-matic ?

That can go on the wiki, cut and paste from the mailing list.  If you
want to lead that, I will do what I can to help.   Note: I am not too
happy with Kwiki wiki;  it gets less support than dirvish.  I would
like to move the wiki to MoinMoin, but Mr. Clock keeps saying 1AM
when I finish the day's work, and I haven't had time to plan the 
move.  BTW, there are a lot of Wikis written in PHP, but I am already
supporting Perl and Python on the server, and without more help I
don't want to support a third server language.

> "You haven't seen much from me, I've spent the last 6 weeks on a 
> death-march-from-hell consulting project. Paid very well, and in about 3 
> years, everyone here will own a copy of the product we are designing. 
> I'll tell more when permitted"

Good catch.  That company is dead, so I can say more - I think.   This
is off topic to dirvish:

As many of you know, my day jobs are as a semiconductor design
consultant ( www.kl-ic.com ), and peddling a silicon identification
technology ( www.siidtech.com ).  With volunteer activities
( www.eeconsult.org ) and ( www.pdxlinux.org ) and ( www.freegeek.org )
and tech support for my doctor wife.  Free time, what's that?

That KLIC job was working on chip design for a projection display system
for Steridian of Vancouver, Washington.  The details are still under
nondisclosure, but the much of the idea is public (business articles,
patents).  The basic idea is that there is a tiny Liquid Crystal
on Silicon (LCOS) display chip called a "write valve" that makes an
image in pulsed UV (or IR) light.  That image is tranferred to a much
larger piece of magic called a "read valve" that acts as an optical
amplifier - the low power UV image changes its visible optical
transmissivity, allowing the very small (5 micron pixel array) write
valve to control 10's of square centimeters of inexpensive imager, and
a few milliwatts of UV image to control 10's of watts of visible light.  
There are 3 read valves, RGB, and they can be electrically sensitized
or desensitized to the controlling image from the single write valve. 
What that all means is that a $5 write valve, some cheap plastic
optics, 3 large $5 read valves, and a cheap large-spot light source
(such as a projector bulb) can project to a bright two meter or larger
display.  It would be possible to construct a one-meter rear-projection
display box, perhaps 20 centimeters deep, with 2Kx1.5K pixel resolution,
for under $100 manufacturing cost.  Beautiful high-contrast low-latency
wide-spectrum images, more energy efficient, no moving parts, no exotic
materials in land fills, etc.  Most of the usage would be for displaying
HDTV and broadcast television, and while I think that is a vast waste of
time, I would rather that time waste be done while wasting less energy 
nd creating less toxic garbage.  This would be a "CRT killer";  we would
no longer be putting a few pounds of lead glass into land fills every
time a picture tube burns out or a television is discarded.

Sadly, high defect density in the read valves (a few dozen spots per
image) killed the project.  The venture capitalists pulled the plug
before the process could be improved.  However, I expect somebody
somewhere is working on stuff like this.  

Since then, there have been many more death-march-from-hell projects;
it is the nature of the consulting business.  Siidtech takes up most
of my time now. 


Keith Lofstrom          keithl at keithl.com         Voice (503)-520-1993
KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon"
Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs

More information about the Dirvish mailing list