| Re: Pass-By-Reference question. [message #38494 is a reply to message #38492] |
Wed, 10 March 2004 15:55   |
Mark Hadfield
Messages: 783
Registered: May 1995
|
Senior Member |
|
|
Craig Markwardt wrote:
> Mark Hadfield <m.hadfield@niwa.co.nz> writes:
>
>> Not so. As Tim said, keywords passed in via inheritance override
>> keywords supplied explicitly. This is a very useful feature and
>> virtually all of my code relies on it.
>>
>
>
> Mark, are you relying on a documented feature of IDL, or are you just
> assuming that because it works, it's right? I've never assumed that
> keyword inheritance works by overriding explicit values. In fact, I
> would have assumed the exact opposite, namely that the explicit
> overrides the generic.
This behaviour *is* documented for value inheritance, eg under the topic
"Keyword Inheritance", heading "Writing a wrapper routine", sub-heading
"By Value" the IDL 6.0 help file says the following"
Note that keywords passed into a routine via _EXTRA override previous
settings of that keyword. For example, the call:
PLOT, a, b, COLOR = color, _EXTRA = {COLOR: 12}
specifies a color index of 12 to PLOT.
Under the sub-heading "By Reference" it says something very similar, but
I must admit I don't understand why structures get introduced here:
These inherited keywords are then passed from TEST to the PLOT routine
using the _EXTRA keyword. Note that keywords passed into a routine via
_EXTRA override previous settings of that keyword. For example, the
call:
PLOT, a, b, COLOR = color, _EXTRA = {COLOR: 12}
specifies a color index of 12 to PLOT. Also note that we are passing a
structure (the by value format used by _EXTRA) as the value of the
extra keyword to a routine that uses the by reference keyword
inheritance mechanism (_REF_EXTRA). There is no problem in doing
this, because each routine establishes its own inheritance mechanism
independent of any other routines that may be calling it. However,
any keyword values that are changed within PLOT will fail to be
returned to the caller due to the use of the by-value mechanism.
Anyway, back in Aug 2000, during the beta test period of IDL 5.4, there
was a long thread on this newsgroup with the title "Keyword precedence",
involving me & JD mainly, during which I noted that there was an
inconsistency between the keyword precedence rules for value and
reference inheritance, and I argued that they should both behave the way
described above (inherited keywords override explicit keywords). I also
raised this with RSI and persuaded them to change the behaviour for
reference inheritance by the time IDL 5.4 final was released.
If you review that thread (though I doubt you'll have the stamina to
read the whole thing) you may note that I began by saying that I had
until recently assumed the contrary (explicit should override
inherited). My first message in that thread explains why I changed my
mind: it allows one tow write cleaner wrapper routines.
--
Mark Hadfield "Ka puwaha te tai nei, Hoea tatou"
m.hadfield@niwa.co.nz
National Institute for Water and Atmospheric Research (NIWA)
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
