[geeks] Sun to adopt newest Intel Xeon chips for upcoming servers (link)

[Joshua Boyd]

On Sat, Jan 27, 2007 at 01:39:46AM +0000, Charles Shannon Hendrix wrote:
> Joshua Boyd wrote:
> 
> > There are a lot of strong points for PC gaming.  I just found that for
> > me it wasn't worth the hassle of keeping a windows computer, or even a
> > reasonably current computer around for that purpose.
> 
> Oh, I don't keep the computer updated for gaming.
> 
> Games are secondary issues.
> 
> The main thing is I have limited space, so I want one machine to do as 
> much as possible.  I'm the same way with servers and utility boxes.  The 
> more they can do, the less machines I have to take care of.

I don't know what you space is like, but my space is pretty limited as
well.  I have maybe 9x4 to store all of this stuff in (a table and a 4
foot rack next to the table).  Thus, to save space I have no windows
machine.  I suppose I could try to play games on Macs or Linux, but a
gamecube was cheap, and it fits nicely on the desk immediately in front
of the Mac, and runs through a video->VGA adapter before hitting the KVM
for switching between the Mac, Linux machine, or DVD player/GC.
 
> >> Do you use OpenGL for that, or are you doing Windows programming?
> > 
> > OpenGL on Linux.
> 
> None of my OpenGL references cover shaders.  Any online documentation 
> for that?
> 
> I should get newer books, but I want to first learn enough to justify it.

I'm not sure you need to replace your books so much as you need a single
new book.  I would think that OpenGL(R) Shading Language (2nd Edition)
would be the perfect compliment to even an older edition of OpenGL(R)
Programming Guide: The Official Guide to Learning OpenGL(R).  I have the
later, but so far I've only looked at the former in a store.

So far, I've made do with scatterings of online stuff, like:
http://85.152.40.5/index.php?option=com_content&task=view&id=24&Itemid=26

It did take me awhile to find some sample linux code that was ready to
be compiled, but once I did, I've been using that as the starting point
for following other guides.

Also, I've only played with pixel shaders (fragment shaders in OpenGL
terms).  
 
> I probably should forget it until I get the basics mastered, but its 
> hard to not at least take a peek.

It seems to me that you might as well tackle both at once.  It sounds
like the modern approach for games at least is to push as much as
possible into shaders anyway.

> I've been playing with the KDE program Digikam, and it is nice, but not 
> very polished.  Most open source stuff is like that.  However, it has a 
> lot of potential because it lets you tag images and once you've gone 
> through the trouble of that, finding things is fairly easy.

I keep meaning to run some photos through F-Stop, a GNOME program.
Running some through DigiKam might be worthwhile as well.
 
> However, I'm a bit of a purist. I like GUIs just fine, but I hate losing 
> access to metadata when I'm on the command line.

That's why I haven't settled for anything else yet.  Also, I want
whatever I pick to at least work with Linux and OS X, if not Irix as well.
 
> Also, Digikam doesn't automatically add tags for metadata like camera, 
> camera settings, etc.  That would be convienient.  I suppose I could add 
> it myself, but that's a lot of work and I'd then have to track Digikam 
> sources to stay compatible.

Although if you did that work, maybe someone else would take up
maintainance, and maybe you could figure out how to build a CLI client
to the database.

> I thought a long, long time about that, and in the end I didn't like the 
> camera's format completely.  Plus, its original filename is stored in 
> the metadata, so I could go back if I really wanted to.

Previously I would have said either way is find.  However, recently I
was go through the database and found pictures where their exif
timestamp was wildly wrong.  If I hadn't had the original file name, it
would have been significantly harder to fix, since the filename at least
let me frame the date as being between two other known dates based on
image sequence numbe.  However, if the original file name is in the
exif, I guess it wouldn't matter so much, so I just didn't feel like
bothering to change file names.

> >  - I can't find the picture unless I remember the date, or I go through
> >    everything in the date range.
> 
> You need a program to let you add tags and comments.

I know.  I should probably just whip up a little program to do this
backed by postgres, but I can't seem to get started on that next stage
of the project.
 
> >  - Deb is interested in what has been printed already.
> 
> You could use tags for that, if you are disciplined enough.

I know.
 
> >  - It would be handy to have a quick means to take only a subset of
> >    pictures into a new location for generating web albums.
> 
> Digikam does that, and of course most Windows and Mac programs will do that.

Yeah, now I just need to drive that feature via a web interface.

Actually, something like an open source flickr clone for home
installation would probably be 90% of what I want, especially if it
offered a SOAP or XML-RPC api.
 
> Beware though: several Windows programs are brain dead stupid about not 
> leaving your originals alone.

iPhoto didn't seem to like to leave my originals alone either.
 
> No, I've not done all of mine.  I keep running out of patience.

Maybe a add a feature that shows you a random untagged picture, so you
can flip through tagging things when bored.
 
> I did have mine tagged in a Windows program 2 years ago, but decided 
> against using the program because I wanted my photos stored on my Linux 
> box instead, so I lost all of that hard work.

I wonder if there would be any money to be made in selling NSLU2Gs
reloaded with an image management system...
 
> Yeah, me too... and it will have a GUI and a command line interface, 
> automatic tagging, print tracking, online ordering... like I said, I'll 
> let you know when I'm done.

Ooh, I never thought of online ordering.  I might have to look into that
someday. 
 
> Seriously though, every photo program I try, I find it just doesn't do 
> what I want.  Some come very close, but then have one annoying problem I 
> can't work around.

I think Aperture could almost be there if they would make a few
seemingly minor (but probably not) changes.

First, open the data format, or make an API for it (although, it at
least sounds like a lot of work is being done to reverse engineer it,
and that Apple didn't obfuscate it).  Second, make it work over NFS.
Third may it multi-user safe.  Fourth, make it able to simultaneously
use multiple data stores (although that might be less of an issue with
NFS support and it stored on a ZFS).  I've heard professional complain
about all of those because they either want add-ons, or they want to
have assistants be able to work at the same time that they are working,
on the same data store. 

Actually, Aperture already uses a SQL database for all data storage, so
why doesn't it offer the option of using a SQL server instead of sqlite3
for workgroup usage?
 
> I have ideas, but I know a project like this would take a lot of time, 
> so I never even bother to start.

I've sorted started, but I really need to get tagging and comments
going.

I think a good system would have the potential to be a commercial
product.  Getting into the home market would be a killer, but it seems
to me like there are large voids in several professional arenas.
 
> Even simple projects I've started quickly ballooned out of control and I 
> had to give up.

It is easy to get bogged down.

-- 
Joshua D. Boyd
jdboyd at jdboyd.net
http://www.jdboyd.net/
http://www.joshuaboyd.org/