| Re: Generating keyword parameters from strings [message #67506 is a reply to message #67432] |
Thu, 30 July 2009 08:21  |
JDS
Messages: 94
Registered: March 2009
|
Member |
|
|
On Jul 23, 3:48 pm, Paul van Delst <paul.vande...@noaa.gov> wrote:
> David Fanning wrote:
>> Paul van Delst writes:
>
>>> I really don't understand why, in the SUBCLASS::Get_Property method, I can get away with
>>> using the _EXTRA keyword to the SUPERCLASS::Get_Property method, but it works. Go figure.
>
>> The rule is you use _REF_EXTRA (or _REF_STRICT_EXTRA, etc) on
>> the procedure or function *definition* line, but *all* extra
>> parameters are to be *passed* with _EXTRA.
>
> Ah, o.k. A nice, simple rule to remember.
>
> Of course it would be nicer if I didn't have to remember it at all. Passing
> arguments/keywords by reference or value is an implementation detail the user shouldn't be
> concerned with. Having to consider it in IDL OO programming is mildly ironic.
Nicety is in the eye of the beholder. At the time I proposed
_REF_EXTRA and it was added to IDL 5.1, the IDL programmer who
implemented it was very proud that it required only a single change to
the routine definition line, not all the _EXTRA calls themselves. At
the time, I objected to the duplicate interface. But I came to
understand that it was an absolute requirement due to a simple but
frustrating issue: various people (including those posting to this
thread!) routinely build or augment their own _EXTRA structures.
If you think about it, there is no way (without inventing a new type
of IDL variable) to simultaneously change the semantics of _EXTRA to
allow passing by reference, and *not* break all of those sneaky codes
which are poking around in _EXTRA structure themselves. So we have
only ourselves to blame.
JD
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
