| Re: IDL 8.1 is a Piece of Work [message #78038 is a reply to message #78036] |
Thu, 20 October 2011 01:43   |
lecacheux.alain
Messages: 325
Registered: January 2008
|
Senior Member |
|
|
On 20 oct, 00:43, David Fanning <n...@dfanning.com> wrote:
> Beaker writes:
>> So my question is should we move to 8.1 or stay put, given the
>> problems people are reporting? We don't do anything too fancy
>> visualisation wise, just static 2d plots and maps, so we don't need to
>> sit and the bleeding edge of fancy interactive graphics, if that is
>> what causes the most problems.
>
> Broken graphics routines is about half the problem.
> The other half of the problem seems to be the Workbench.
> If you use the command line to do non-graphical things,
> IDL 8.1 might please you.
>
> Certainly lists and hashes are useful. The object
> syntax is fabulous. Lists and hashes have been
> implemented in the IDL language and exist for
> earlier versions of the language. (I use my LinkedList
> object for many "listy" things, and Craig Markwardt
> has a very useful hash object.) As far as I know,
> there are no "memory" issues with either. They
> understand how to clean themselves up properly.
>
> I won't be able to use lists and hashes for another
> couple of years or so, just because of the kind of
> work I do, but if you are not keen to share your
> software and/or everyone you know is using a recent
> version of IDL, then I say go for it.
>
> ITTVIS claims the graphics system is "95% working",
> but I seem to run into that other 5% every time I
> try to do something useful. I have made a line plot
> (sans legend, of course) that works properly. For
> my purposes, I would say the graphics system is
> 95% broken. I guess your mileage may vary, because
> I run into people who say they love the graphics
> routines. They probably aren't as anal about
> how things look as I am. :-)
>
> If you do decide to upgrade to IDL 8.1, be sure
> to purchase the three year maintenance support
> contract at the same time. I have a feeling you
> are *really* going to need it. ;-)
>
> Cheers,
>
> David
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming:http://www.idlcoyote.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
As a constant user of IDL since IDL v2, I would like to both reinforce
and a little bit attenuate what David is saying.
IDL is an extremely powerful, efficient and elegant tool for doing
calculation, data processing or data visualization in science (maybe
the most powerful among those presently available). Partly because of
the clever properties of the IDL language (continuously improved),
partly due to the robustness of the IDL engine even in case of massive
data processing (I never seen any crash of IDL before v7, except when
there was some strong programming fault made by the user). Therefore,
the future of IDL should be guaranteed.
Now, from v7 and especially with v8, we encounter two surprising
weaknesses:
1) the new IDE, - the so called Workbench -, which relies on the
Eclipse freeware, and replaces (for MS-Windows users only) a native,
perfectly working IDE (available since v3). The Workbench clearly
suffers from several internal bugs, mainly regarding memory
allocation, which preclude, at the moment, any "professional" use of
it. The IDL engine itself does not seem to be in fault; the problem
likely lies in the implementation within Eclipse or in Eclipse itself.
The(se) bug(s) hopefully will be corrected by ITTVIS in the next IDL
versions.
2) the introduction of a new way for doing interactive graphics, which
is only at "95% working" (ITTVIS saying) and certainly at 0.001%
documented ! I am pretty sure that NG is a clever step towards an
easier (interactive) use of IDL graphic objects, following other
unsuccessfull tentatives like LiveTools, ITools, etc... But NG was
delivered by ITTVIS with a number of remaining and particularly
frustrating inconsistencies or bugs. For example the fact, described
by David, that resizing a window may desorganize this window content !
Or that labels go everywhere when axis scale are set as logarithmic...
Many correcting recipes (brightly revealed by Mark Piper in his
webseminars) are impossible to be conceived by a common user alone,
because of the absence of documentation. In other words, I think that
NG was introduced by ITTVIS too early, in a finalization state which,
at the moment, rules out its use in any real programming application.
But, once more, except in these two particular points, IDL 8.1 is
actually a very valuable tool, with many impovements with respect to
earlier versions !
alain.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
