| Re: How Object-oriented? [message #9033 is a reply to message #9018] |
Thu, 22 May 1997 00:00  |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
David Ritscher <david.ritscher@zibmt.uni-ulm.de> writes:
> Is anyone ready yet to comment on how object- oriented IDL 5.0 is?
I am *not* an expert on OOP. My only qualification here is that
reticence is not usually associated with my name. Please keep that
in mind.
And my remarks are going to be colored by the fact that
I was up until the wee hours of the morning trying to do the very
simplest thing in IDL 5.0: display a surface with some axes on it.
I have spent the better part of three days on this, so far without
success, so I am feeling peevish about OOP just this moment. So
please keep that in mind.
And finally, to give RSI every benefit of the doubt, I am still working
with the beta version of the software, which everyone expects to
be buggy. So *please* keep that in mind. :-)
> OO has become a big buzz word, so it has become important to at least
> give lip service to this concept. Can anyone comment on IDL 5.0?
> Is it more on the lip-service level, or are the changes significant
> enough that one can write real OO software?
I believe that RSI's commitment to OOP is more than lip service.
The pointer implementation is just outstanding and I like very
much the way they have implemented objects. I think the object
structures are straightforward. I like the way they can be
automatically initialized. I especially appreciate the almost
complete flexibility you have in writing methods for your objects.
And the elegance and simplicity of the Data Miner object
implementation may be the best thing RSI has ever done
with respect to building easy-to-use software.
I think, in short, that yes, the changes are significant enough
that you will be able to begin to write real OO software.
You only have to work with objects for a few minutes to start
getting all kinds of ideas for powerful programs that could
be written with them. I think, eventually, that objects will have
as big an impact on IDL programs as widgets did when they
were first introduced.
But I can remember when widgets were first introduced and
how frustrated I got at times. The documentation was sketchy
at best, often wrong or misleading. The software was buggy.
Things changed from one version of IDL to another, usually
without warning, etc. (To give RSI credit, they usually
changed for the better. I just sometimes didn't want to hear
it after I had spent many days modifying my programs to work
around some bug.)
I am in that frustrated mode right now after trying to
write what I consider to be the very simplest graphics displays
with the new object graphics routines. How in the world can
something touted to be the very latest in programming innovation
be so infernally hard to program!
I want to do to two simple things: (1) Display a surface with
axes, and (2) Change the size of the axes labeling to something
other than the grotesquely large default size. (I was actually
embarrassed for RSI when I first saw the plots that came
up in the Insight demo on my Windows NT machine.)
I expected to have to do more low-level programming with
the object graphics system, especially with the first
release of the software. But I did not expect to have to
do so much low-level programming without adequate
documentation. Last night I was reduced to the kind of
programming I hate: making almost random changes in
my programs in the futile hope that I might see a pattern
in the results that gave me a clue as to what in the world
was going on.
Just to give an example. I cannot figure out how to put
axes on my surface so that it looks like a normal surface
plot. (Yes, I know there are examples. I've run them. They
work fine when the data is DIST(40), but they don't work
at all if I try to display *real* data!) There is a LOCATION
keyword in the Axis object that would seems to be what
I need. Here is the explanation of that keyword:
"Set this keyword to a two- or three-element vector of the
form [x, y] or [x, y, z] to specify the coordinate through which
the axis should pass. The default is [0, 0, 0]."
"The coordinate through which the axis should pass!?" What
can this possibly mean? I can imagine an X axis having to
pass through a point in the YZ plane, but beyond this I am
stumped. And in any case, two hours worth of plugging all
kinds of values into this keyword led to no insights. (A pun.)
And although I can now change the size of an axis title,
I have still discovered no way to change the size of the
axes annotation, beyond specifying each individual
annotation separately, which cannot possibly be what
RSI has in mind.
So, bottom line? I think RSI is moving in the right direction
with their OOP ideas. I think eventually it will revolutionize
the way IDL programs are written. But the current implementation
of object graphics (in my beta version, please remember) is
*not* ready for prime time.
Let's just say I'm glad I'm not an RSI technical support
engineer right now!
Cheers,
David
----------------------------------------------------------
David Fanning, Ph.D.
Fanning Software Consulting
Customizable IDL Programming Courses
Phone: 970-221-0438 E-Mail: davidf@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
