| Re: Probs w/ keyword_set [message #2509] |
Wed, 03 August 1994 07:09  |
thompson
Messages: 584
Registered: August 1991
|
Senior Member |
|
|
tai@rockola.larc.nasa.gov (Alan Tai) writes:
> This is probably old news, but in IDL 3.6 (and I think in 3.51), I
> realized early on that you run into problems by setting a keyword
> parameter to 0 and running it through keyword_set. I.e. keyword_set
> believes the parameter is unset:
> pro test, Key = Key
> if (keyword_set(Key)) then print, 'Set' $
> else print, 'Unset'
> end
> test, Key = 0
> Solution: use n_elements instead of keyword_set. I found this out
> from reading through some procedures from the JHUAPL library.
This is the way it's supposed to work. Why do you think it's a problem? It's
standard for 0 to represent false and 1 to represent true. The beauty of the
KEYWORD_SET function is that if the variable is undefined, then it assumes
false by default.
It's possible you're trying to use KEYWORD_SET for keywords that aren't
true/false switches. If a keyword can take other values than true/false, and
you're trying to decide whether or not that keyword has been passed, then
N_ELEMENTS is the correct way to test for this, as you point out.
For example, in the AXIS routine, the keyword XAXIS can take the values 1 or 0.
However, in this case the values do not represent true or false. Instead,
setting XAXIS=0 means draw an axis under the plot window, and XAXIS=1 means
draw it over the plot window. Leaving it undefined means to not draw either.
In a situation like this, using KEYWORD_SET to decide whether or not XAXIS was
passed would be a mistake.
However, in the same routine, it would be correct to use KEYWORD_SET to
determine the state of the /SAVE keyword.
I've noticed that sometimes people will use the /KEYWORD form to set a keyword
that can take a variety of values to 1. This is bad programming practice. It
confuses the distinction between keywords that make true/false distinctions,
and keywords that take more traditional values.
Hope this clears up the confusion,
Bill Thompson
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
