| Re: Curious Keywords [message #8393 is a reply to message #8390] |
Thu, 27 February 1997 00:00   |
J.D. Smith
Messages: 214
Registered: August 1996
|
Senior Member |
|
|
Kevin Spencer wrote:
> The reason for this behavior is that the default value of UNDEF in both
> cases is 0. For the POLAR keyword, supplying a 0 means don't do anything,
> so it doesn't. But supplying 0 to THICK generates an error because the
> line has to have *some* thickness to it.
>
> Dig?
O Contraire!
Please try:
plot, findgen(100), findgen(100)/99.*2*!PI, THICK=0
Works pretty nicely.
The question is really about the mechanism of Keyword Checking for built-ins
vs. non-built-ins. That is, if *I* wrote a routine that took the "THICK"
keyword, a user could pass an undefined value through THICK and I would never
know that he did. I could not use n_elements() or keyword_set() or any other
mechanism I know of to discern in any way that he has used "THICK". And so,
consequently, I would not be able to issue an error message in the case he
passed an undefined variable. It is as if he never passed it at all! Not so
for Plot and other built-in routines (for *some* of their keywords). They
somehow "know" that I used e.g., "THICK", even when I pass them an undefined
variable.
My point was that this would be a useful feature to have (although it would
possibly result in some subtle and harmful programmatic issues). For
instance, if I pass a variable through a keyword into which I'd like to put
the result of some calculation, I have to give that variable a value before
passing it, in order to test whether to go through the bother of doing the
calculation at all. There are, of course, other ways to do this (e.g.
optional *parameters*), but I've found myself wishing for this particular
mechanism on some occassions, when the other techniques had limitations.
JD
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
