| Re: Using _extra for all routines? [message #5176] |
Sat, 14 October 1995 00:00 |
meron
Messages: 51
Registered: July 1995
|
Member |
|
|
In article <jackel.47.0D621ECF@canlon.physics.uwo.ca>, jackel@canlon.physics.uwo.ca (Brian Jackel) writes:
> In article <45ltaq$8q7@llnews.ll.mit.edu> knight@ll.mit.edu writes:
>
>> Does anybody see any drawback to including _extra as a keyword in ALL
>> procedures and functions? My motivation is to allow keywords passed through to
>> multiple routines inside others. Here's an example:
>
>> Is this ok? One drawback might be ambiguity, two keywords whose names
>> conflict. Does anybody see any other drawbacks?
>
> One thing to watch out for is that any mis-spelled keywords will be absorbed
> by _EXTRA. No big deal if you're using KEYWORD_SET() to check on all legal
> keywords, and using lots of messages if they're not provided. However, if
> un-supplied keywords are being quietly given default values the user may never
> realize that they're using an improper keyword.
The biggest problem is that when a single routine calls few other, all
the stuff in _extra will be passed down to the first one. If this one
is a "bottom routine" and it doesn't recognize some of the keywords,
it'll object. If it is not bottom, and is also defined with _extra,
it'll pass the unidentified ones farther on, etc. Eventually chances
are that you'll hit a "bottom" that doesn't recognize some of the
keywords and it'll bomb. Of course you can deal with by pre and post
processing (removing pieces from the structure defined by _extra
before a call and returning them there after the call, but that's
tedious and you have to know in advance what you may want to remove.
A case like the one described by the previous poster (misspelled
keyword) will slip right through.
_extra is a powerful stuff, but I recommend using it with care.
Mati Meron | "When you argue with a fool,
meron@cars3.uchicago.edu | chances are he is doing just the same"
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
