On Fri, Nov 03, 2000 at 04:48:54PM -0700, Anthony Walsh wrote:
> In response to developer jobs, I didn't feel it was appropriate to post
> those in here.  Maybe that is an error on my part.  I am more then willing
> to post those positions as well, if the overall group feels that it is
> right.  Do you think that posting a job that is a C++ developer on a Solaris
> system with Oracle on the backend would be appreciate to post out here?  If
> so, I have a ton of those openings too...  :o)

Won't bother me.  But I'm mostly a Java developer... once you have
tried it, C++ seems so clunky by comparison.  :-)  'course I still 
write stuff in C when speed is important, but for rigorous OO
methodology I'd rather use Java these days.  But I suppose Java 
would be a bit more off-topic here unless it's directly linux-related, 
and anyway there is a JUG too.  You can post jobs at www.phxjug.org.

I think it might be fun to do some embedded Java at some point.  I'm getting
tired of writing "business" software.  I thought EJB was going to be
really cool, but it didn't take very long to get disillusioned.

People keep saying Zope is cool, but to me it seems kindof opaque;
not easy to figure out how it works, like they expect the need to 
write new server-side code (other than HTML with special tags) to
be rare.  Maybe it's just a matter of spending more time digging.
If I understood how the so-called "object database" works, I might
think it was cool, but my OO instincts tell me that the thing to 
do when building any system is to design the "model" tier first, as
a set of application-specific objects, and then hope that the OODB
will be able to transparently manage their persistence.  But it 
seems like Zope steers you away from that; I sortof get the impression
maybe they are just pretending that files are "objects".  And if an
OODB is done right, you'd have no need for a relational database; 
so it is suspicious that Zope tries to provide good RDB integration.

EJB doesn't exactly steer you away from the OO modeling approach, but
it does put so many constraints on your model due to its attempt to
map objects to the relational database, that you end up modeling in
many layers; and you are responsible for building too many of them.
All of the hype about how flexible it is, fails to prepare you for
the fact that every degree of freedom that it has, costs you another
whole layer of code.  And then there is the use of XML to "glue"
it all together... but you have to write all of that, too.  There are
no looping constructs, no syntax for doing includes or macros in
this XML code... so it's very verbose.  Several giant XML files each
of which describe some facet of the entire system, for every component
in the whole system.

I've been working on a GUI system similar to NeWS in my little remaining 
spare time (after the urgent projects...); that's in C because speed 
and code size are both really important.  

It would be cool if I could find a company to sponsor that work, 
and the "metawidget" stuff to follow.  Working in a research-lab 
environment would be a lot of fun.  But I don't know where to start, 
or if it would be worth the hassle, and of course I want it to be free 
software.  I have a great fear that doing "heart and soul" kind of 
work for a company usually ends up being a trap in one way or another;
they want to make money from it and it ends up not being quite 
free enough, or they expect you to stick around and support it after
it's built rather than going on and doing something even cooler, etc.  
Of all the ideas I've had, this is one of the most impossible
to turn in to a money-maker.... it really wants to
be free.  (as in the Hacker Manifesto... information wants to be
free, but my personal extension to that principle is that software
wants to be free much more than information does, and basic
infrastructure wants it the most.  This idea falls squarely into
that category.)  So the practical side of me says I should be working
on something that could make money... but it seems like another
basic principle is that the best ways to make money are the most
short-lived projects.  And I prefer to build things to last... if a 
piece of software is "disposable", that usually means it didn't solve
the problem well enough; which usually means that its scope was too
narrow.  But solutions in narrow niches do tend to make a lot of money;
whereas broad-scope solutions have to be free in order to be applied
the most broadly.  All of this makes it hard for me to be totally 
happy in any career that I know of.

So right now, like most people, I just have a job to support 
my habit.  :-)  For that purpose maybe it's best to have it kindof low-
stress and boring, like my current job.

-- 
  _______                   Shawn T. Rutledge / KB7PWD  ecloud@bigfoot.com
 (_  | |_)          http://www.bigfoot.com/~ecloud  kb7pwd@kb7pwd.ampr.org
 __) | | \________________________________________________________________
 Free long distance at http://www.bigredwire.com/me/RefTrack?id=USA063420