| IDL vs. PV-WAVE [message #275] |
Tue, 03 March 1992 10:56  |
toma
Messages: 3
Registered: February 1992
|
Junior Member |
|
|
Awhile back I posted a reply to someone who asked about the differences
between IDL and PV-WAVE.
My comments were to the effect that they were basically equivalent,
except that PV-WAVE had much more employees, and cost a lot more.
This information was based on my experience with the two packages
as of Oct.1990. I am now given to understand the the two
have since gone their separate ways and hence are no longer equivalent.
In particular IDL now has many features PV-WAVE apparently does not.
My own experience for the last 18 months has been exclusively with
IDL; it is more than sufficient for our needs.
Anyone desiring more information about the differences may email me.
I apologize for any misunderstanding this may have caused.
--
Tom Atwater
|
|
|
|
| Re: IDL vs. PV-WAVE [message #9165 is a reply to message #275] |
Tue, 10 June 1997 00:00  |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
Grant W. Petty writes:
> This has got to be a novice FAQ, but I can't find a FAQ file that
> addresses it:
>
> What is the difference between IDL and PV-WAVE? I had understood they
> were two competing packages that did similar things, but the name of
> this newsgroup (comp.lang.idl-pvwave), plus the fact that EVERYONE
> here is talking only about IDL (at least today), seems to suggest
> otherwise.
Five years ago IDL and PV-Wave were the identical thing.
Then Precision Visuals (now Visual Numerics) bought a copy
of the IDL source code and the products have diverged from
there. The differences are mostly a matter of implementation,
not content, as both companies look over their shoulders
(quite literally at times in Boulder) at the other. There
is a "keep up with the Jones" mentality that at times makes
you weary.
> Anyway, I am looking at purchasing an IDL-like package for my research
> group and, assuming that PV-WAVE and IDL really are distinct
> creatures, would like to know what the crucial differences are. My
> applications will include the interactive graphic display and
> statistical analysis of scientific data sets and the preparation of
> publication quality figures on HP workstations. Particularly useful
> features would be the ability to project geographically referenced
> data as polygons, raster images, labeled contours, etc. on an
> arbitrary geographical map. Also, the ability to interactively label
> and edit plots and then produce PostScript output would be nice.
> Finally, the ability to read directly binary data in various formats
> directly from tape, rather than having to stage it to disk first, and
> to directly manipulate tape drives, would be nice. I have no idea how
> well either IDL or PV-WAVE matches these requirements.
I would say either IDL or PV-Wave would match these requirements
in spades. ENVI, a product that is written in IDL and runs on
top of IDL, can probably do all that you ask here without you having
to do any programming at all.
> After looking around this ng, I have to admit to being put off by a
> couple of seemingly recurring complaints about IDL: (1) persistent and
> difficult-to-work-around bugs in basic plotting routines -- e.g.,
> missing grid lines, etc., and (2) the apparently steep learning curve,
> based on people's remarks about the need for expensive IDL training
> courses to learn widgets, etc.
Don't be put off. IDL just released a new version of the
software and everyone is trying to figure it out. As usual,
reality does not jibe with marketing hype, so there is some
settling of expectations. The same thing happens when a new
release of PV-Wave is announced. If you want cheap training,
have a look at my web page.
> After years of struggling with similar
> problems with the NCAR Graphics package routines, I would really love
> to get my hands on a package that works dependably and which is easy
> for new users to figure out how to use from the hardcopy docs.
Yeah, you and everyone else. But I don't think it's going to
happen anytime soon, no matter what you buy. :-)
Cheers,
David
P.S. I'd flip a coin and pick the least expensive option.
P.S.S. This is just a hunch, but I would say you see more
IDL in atmospheric sciences than PV-Wave. Ask around and
get what your colleagues are using. It helps not to have
to write everything from scratch.
------------------------------------------------------------ ----------
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
IDL 5 Reports: http://www.dfanning.com/documents/anomaly5.html
|
|
|
|
|
|
|
|
| Re: IDL vs. PV-WAVE [message #9274 is a reply to message #275] |
Fri, 13 June 1997 00:00 |
sritcey
Messages: 5
Registered: March 1997
|
Junior Member |
|
|
William Connolley wrote:
> 2. Widgets seem to be particularly complex, but you probably don't
> need them.
David Fanning wrote:
> Now hold on there just a minute, William! ....<snip>...
> When they are shown how easy it really is, they often start
> taking the view that widget programs are essential to their research
> and they don't know how they worked without them.
Jonathan Rogness (rogness@NO.sg1.SPAM.cr.usgs.gov) wrote:
: I'd be interested in hearing some of these people speak up, just because
: I'm curious about how exactly they incorporate widgets into their
: research. Personally, I write about 5-10 programs a day, either in IDL
: or C, and only one of them (about a year ago) has ever been a widget
: application.
[snip]
: Perhaps it just depends on the type of research
Perhaps. Most of what I write isn't widget based either, in the
sense that the stuff that takes thought and time is small, modular
routines which I then call from
1. Wave command line
2. Wave scripts
3. Widget interface
I use [1] when I'm developing an analysis concept, [2] to process a large
mass of data when I'm sure I'm doing the right thing, and [3] to explore
a new data set when I've previously chosen which tools to use but want to
quickly apply one, then the other, in an unpredictable way.
Although the learning curve is steep (at least the Wave manuals have lots
of examples), widgets are the quickest way to jump easily from one data
set to another and from one analysis mode to another. I find it
preferable to repeatedly typing command line variations as one explores.
I suppose it really is useless if there is not some degree of sameness to
your analysis from data set to data set or from day to day.
Sorry if this is a bit rambling, but I for one find widget programs useful
for personal use in addition to supplying them to to nonprogrammers.
|
|
|
|
|
|
| Re: IDL vs. PV-WAVE [message #9281 is a reply to message #275] |
Fri, 13 June 1997 00:00 |
Jonathan Rogness
Messages: 7
Registered: April 1997
|
Junior Member |
|
|
David Fanning wrote:
>
> William Connolley writes:
>
>> 2. Widgets seem to be particularly complex, but you probably don't
>> need them. As far as I can tell, they are mostly useful for
>> polished-product programs and not so useful for research-type work.
>
> Now hold on there just a minute, William! ....<snip>...
> When they are shown how easy it really is, they often start
> taking the view that widget programs are essential to their research
> and they don't know how they worked without them.
I'd be interested in hearing some of these people speak up, just because
I'm curious about how exactly they incorporate widgets into their
research. Personally, I write about 5-10 programs a day, either in IDL
or C, and only one of them (about a year ago) has ever been a widget
application. This, of course, puts me in the large group of unfortunate
souls who learned from the manuals. =)
Perhaps it just depends on the type of research; I generally use IDL to
manipulate large arrays and images, and then spend some time TVSCLing
the results in order to decide what needs to be done next. Sometimes
I'll use it to fit models to data, but generally I write programs in C
using the NAG numerical library to do that. But it seems to me that
using widgets would just slow me down, since I can usually accomplish
what I need with a few commands at the prompt or a half page procedure.
Jon
|
|
|
|
| Re: IDL vs. PV-WAVE [message #9283 is a reply to message #275] |
Thu, 12 June 1997 00:00 |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
William Connolley writes:
> 2. Widgets seem to be particularly complex, but you probably don't need them.
> As far as I can tell, they are mostly useful for polished-product programs and
> not so useful for research-type work.
Now hold on there just a minute, William! In my humble, professional
opinion I find that people who hold this view (present company excluded,
of course) don't generally know how to write widget programs.
Most of the time they have tried to learn how to do it from the IDL
manuals. When they are shown how easy it really is, they often start
taking the view that widget programs are essential to their research and
they don't know how they worked without them.
On the other hand, those folks who refuse to jump on the bandwagon
sometimes have the last laugh. Look at the COBOL programmers, for
example, laughing all the way to the bank these days. :-)
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
IDL 5 Reports: http://www.dfanning.com/documents/anomaly5.html
|
|
|
|
| Re: IDL vs. PV-WAVE [message #9292 is a reply to message #275] |
Wed, 11 June 1997 00:00 |
wmc
Messages: 117
Registered: February 1995
|
Senior Member |
|
|
In article c6e@mozo.cc.purdue.edu, gpetty@rain.atms.purdue.edu (Grant W. Petty) writes:
> After looking around this ng, I have to admit to being put off by a
> couple of seemingly recurring complaints about IDL: (1) persistent and
> difficult-to-work-around bugs in basic plotting routines -- e.g.,
> missing grid lines, etc., and (2) the apparently steep learning curve,
> based on people's remarks about the need for expensive IDL training
> courses to learn widgets, etc.
1. There are some irritating bugs in the plotting (especially doing polar
stereos...) but they are few and far between.
2. Widgets seem to be particularly complex, but you probably don't need them.
As far as I can tell, they are mostly useful for polished-product programs and
not so useful for research-type work.
- William
---
William M Connolley | wmc@bas.ac.uk | http://www.nbs.ac.uk/public/icd/wmc/
Climate Modeller, British Antarctic Survey | Disclaimer: I speak for myself
|
|
|
|
|
|
| Re: idl vs. pv-wave [message #9612 is a reply to message #275] |
Thu, 24 July 1997 00:00 |
Franz.Dollinger
Messages: 14
Registered: July 1997
|
Junior Member |
|
|
In article <33D603BD.4FDA@zibmt.uni-ulm.de>, David Ritscher
<david.ritscher@zibmt.uni-ulm.de> writes:
>
> Don't forget to look at MATLAB as another possible choice. They are
> also making strides now, and it looks like they are adding enough new
> language features that one could now program a real system within
> MATLAB (before, there were major deficiencies, such as no 2-D arrays,
> no integer type, etc.).
After using PV-WAVE a few years I changed the company I'm working for and had to
take a look at MATLAB as it was the only available program around for
data analysis.
All I encountered was deficiencies. I was really disappointed by the bad
performance of MATLAB with every single command I entered. Tens of minutes for a
contour plot PV-WAVE needs about half a minute for. When trying to produce a
surface plot MATLAB painted every single point in a way a first semester student
would implement it as quick_and_dirty solution. (Of course screwing up the
information )
In addition MATLAB needed tons of memory and CPU time, causing me to run it on a
machine equipped for doing huge FEM analysis.(I am working on a UNIX workstation
most of the time)
When running an application under NT that needed repeated system calls, MATLAB
made working on the desktop impossible as it created an iconized DOS-window for
every system call deleting every mouse selection or pulldown menu that was
activated at that moment.
A collegue had to pre-format data using a C program to make MATLAB work with
that data. (he is an experienced programmer)
> Particularly interesting with MATLAB are the
> availability of 'compilers' that convert MATLAB code to C or C++ code.
has anyone got experience with the efficiency of the C/C++ code generated
by MATLAB? If the performance is like that of MATLAB programs itself
one better sticks with C/C+ and appropriate toolboxes.
> They have also recently added some object-oriented features. My
> initial impression is that They do not seem to be as strongly
> integrated into the language as do those of IDL. A major advantage of
> MATLAB is the rich collection of toolboxes, often written by
> software has been developed under MATLAB; I can send you references,
> if that's of interest.
Advantage??? You have to buy a toolbox for every single command PV-WAVE
already integrated (at least in PV-WAVE Advantage)!
Its hard for me to believe that so many people use that program. The performance
I experienced was worse than that of a typical MS/Windows program.
And they want real $$$ for that hack. If you buy a license for UNIX the license
has to be renewed (=$$$$) every single year!!
I understand the Windows/PC philosophy to get less quality and less performance
for less $$$ making things more affordable. But the price for MATLAB seems (only
got second hand info) to be even higher than that of PV-WAVE Advantage (!!!!!!)
whith drastically worse performance and much less flexibility. Not talking about
the lack of functionality ( navigator, wz_plots, data structures, IMSL numerics,
...)
bye
Franz
(I never worked with IDL - so I cant comment on that)
|
|
|
|
| Re: idl vs. pv-wave [message #9627 is a reply to message #275] |
Wed, 23 July 1997 00:00 |
David Ritscher
Messages: 30
Registered: August 1995
|
Member |
|
|
William B. Gunter wrote:
> My boss wants a comparison between idl and pv-wave. I know they are
> similar, but what are the main differences between them? Can an idl
> program run under pv-wave with no, minimal or many modifications? What
> are the advantages/disadvantages?
As I am currently using PV-Wave in my cardiology research, I have a
few comments to offer. This post is similar to one I sent in response
to the same question from Steven Jaffe.
This PV-Wave vs. IDL question is, appropriately, a recurring question
here - maybe, if enough good responses come in, the FAQ keepers could
add a detailed section on this theme; or perhaps it belongs in a
separate document (would anyone care to take over this function?).
I am a long-time IDL and PV-Wave user. The comparison between the two
products is currently rather complicated. I try to program so that
most of what I do is compatible with both packages (at least, except
for the widget part of things). The body of the two languages has
remained 90% identical up till the newest release of IDL (5.0). Up
till this point the basics of the languages were the same, with only
trivial differences, except for:
1. A different widget interface
2. A different set of math extension routines. IDL uses Recipes in C,
PV-Wave uses the IMSL routines. After the makers of IMSL and
PV-Wave merged, IDL lost access to the IMSL routines and then
chose to adopt the Recipes in C routines.
3. Each offers their own point-and-click type of interface for those
who don't want to use a command-line programming approach or wish
to supplement this approach with some quick-access tools.
IDL 5.0 introduces some elements of object-oriented programming.
I have developed a software system for high resolution ECG analysis
that is in daily use in our hospital. It contains some 35,000 lines
of code. It is not object oriented, and at this point much time and
energy is lost repairing things in this code, fixing things that get
broken when a new feature is added, etc. From this experience the
need for an object-oriented approach has become clear to me. The
PV-Wave folks say that they have no current intention of changing the
basis of the language so that it would support object-oriented
programming. (They speak about object oriented, but this has to do
with some pre-packaged tools, not with the basis of the language
itself). IDL 5.0 also introduces the ability to deal more directly
with pointers, which can be handy when dealing with large data sets,
such as my ECG measurements.
Whether object oriented is important for your application depends on
what you'll be doing. If the chief goal is to use the software as a
hand calculator with good graphics display capabilities, it is
probably irrelevant. If the goal is to design software components to
be used and extended by a group of people, then I would consider it
crucial.
But speaking of object oriented, the makers of PV-Wave are
demonstrating some strides in this direction, namely JWave and JNL.
It looks like they have given up on the idea of updating their own
language and are instead jumping on the Java bandwagon, and providing
tools for Java that provides advantages normally available within
PV-Wave. There is a writeup on JWAVE on a Netscape DevEdge page:
http://developer.netscape.com/guides/components/
JNL is an extension to Java that provides needed numerical capabilities.
It is a hard time to make a choice between these two branches of the
language, since both companies are working hard to try to
differentiate themselves. There remain problems within the common
core language, and I hope that this competition will lead to solutions
to these problems. As someone here recently mentioned, when one
write, in either language, "a = 3", one has just defined a TWO BYTE
INTEGER variable. This was probably a logical default back when the
language was first conceived. This leads to strange things, such as:
print, 850 * 77
-86
A mechanism for backwards compatibility plus future development needs
to be reached; perhaps a system flag, where, for example,
!LANGUAGE_LEVEL=1 inserted into old routines would provide backwards
compatibility to a specified language version.
Don't forget to look at MATLAB as another possible choice. They are
also making strides now, and it looks like they are adding enough new
language features that one could now program a real system within
MATLAB (before, there were major deficiencies, such as no 2-D arrays,
no integer type, etc.). Particularly interesting with MATLAB are the
availability of 'compilers' that convert MATLAB code to C or C++ code.
They have also recently added some object-oriented features. My
initial impression is that They do not seem to be as strongly
integrated into the language as do those of IDL. A major advantage of
MATLAB is the rich collection of toolboxes, often written by
software has been developed under MATLAB; I can send you references,
if that's of interest.
I would be interested in hearing from current IDL users as to how
sufficient the current object-oriented features are, and what is
missing. Reading through the documentation I see, for example, that
polymorphism and overloading aren't there.
Regards,
David Ritscher
--
David Ritscher
Zentralinstitut fuer Biomedizinische Technik Tel: ++49 (731) 502 5313
Albert-Einstein-Allee 47 Fax: ++49 (731) 502 5315
Universitaet Ulm Internet:
D-89069 ULM
david.ritscher@zibmt.uni-ulm.de
Germany
|
|
|
|
| Re: idl vs. pv-wave [message #9630 is a reply to message #275] |
Wed, 23 July 1997 00:00 |
chitturk
Messages: 1
Registered: July 1997
|
Junior Member |
|
|
William B. Gunter (wbg@crml.uab.edu) wrote:
: My boss wants a comparison between idl and pv-wave. I know they are
: similar, but what are the main differences between them? Can an idl
: program run under pv-wave with no, minimal or many modifications? Waht
: are the advantages/disadvantages?
There _are_ differences and IMO, not trivial. I tried to run an
IDL program in PV Wave .. took me a while, had some widgets that
were different ... once I got past that point, it was OK ...
If you going to _buy_ a new program (IDL or PV Wave) perhaps it
does not matter ... if you have a LOT of programs in one or the
other ... I'd be careful ... Ask IDL / PV WAVE (the companies) as
to how well (or poorly) you may be able to handle it ...
|
|
|
|
| Re: idl vs. pv-wave [message #9632 is a reply to message #275] |
Tue, 22 July 1997 00:00 |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
William B. Gunter writes:
> My boss wants a comparison between idl and pv-wave. I know they are
> similar, but what are the main differences between them? Can an idl
> program run under pv-wave with no, minimal or many modifications? What
> are the advantages/disadvantages?
The main differences between IDL and PV-Wave are all the good
enhancements that have been added to each of them over the past
4-5 years since they diverged as products. This will include
almost all of the good math routines, map projects, ability to
add graphical user interfaces, etc, etc, etc.
There is virtually no chance these days that an IDL program will
run unmodified in PV-Wave and visa versa unless it is a very
simple program, indeed.
At the moment IDL seems more advanced in writing object-oriented
programs, which I think have the chance (if anyone can learn to
write them) of revolutionizing the art of IDL programming as much
as widgets did several years ago. PV-Wave, on the other hand, seems
to be paying a whole lot more attention to JAVA and getting applications
to work on the network.
I think they are both excellent programs with great features.
Many people I know who have been in your shoes have ended up
tossing a coin, or at least that's how it appeared to me. My
favorite reason for choosing PV-Wave, back in the days when
the two products were still identical, was that PV-Wave's
"syntax was easier to use". Sigh...
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
IDL 5 Reports: http://www.dfanning.com/documents/anomaly5.html
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
