| Re: Inheritance query [message #17629] |
Fri, 05 November 1999 00:00 |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
Struan Gray (struan.gray@sljus.lu.se) writes:
> I'm confused. I got the point that if you don't have any INIT or
> CLEANUP method when the first instance of an object is created, it is
> impossible to add one later in the same session, but your article
> seems to imply that it is impossible to modify those methods either,
> even if they successfully compile and run the first time round. I do
> the latter all the time; in fact, it's one of the joys of non-blocking
> widgets. The class *structure* can't be edited, but the lifecyle
> methods can (IDL 5.2.1 on MacOS if it matters).
Yes. I ran out of time yesterday and put that article up
before I had tested everything one more time. I was writing
it from sketchy notes I had made in recent IDL courses. I'll
do more careful editing today.
But I think you are right that the INIT and CLEANUP
methods can be edited once they have been properly
associated with the object. Frankly, in classes,
once we get the damn things to work we don't touch
them anymore. :-)
> If the INIT or CLEANUP methods crash so that I get bounced back to
> the command line with the offending routine opened for editing, IDL
> seems to be in a very odd state, but typing RETALL and HEAP_GC several
> times (at least twice) gets me back to the point where I can
> sucessfully edit and recompile.
Ah, of course. Now I see why this feature is undocumented. :-)
I think I'll stick with my suggestion to exit IDL.
At least I can come up with a believable story
for *that* recommendation.
> A final tip: it is worth including an INIT and CLEANUP routine for
> all top-level classes (ie, those that don't inherit anything). If you
> don't, IDL rummages around in your currently-defined search path
> looking for one every time you create or destroy an instance of that
> class, which can severely degrade performance.
Yes, I meant to include this caveat and just completely forgot.
Thanks for reminding me. I'll do it now.
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438 E-Mail: davidf@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
| Re: Inheritance query [message #17631 is a reply to message #17629] |
Fri, 05 November 1999 00:00  |
Struan Gray
Messages: 178
Registered: December 1995
|
Senior Member |
|
|
David Fanning, davidf@dfanning.com writes:
> I'm just about done with an article explaining
> this phenomemon in more detail. It should be up
> on my web page later today.
I'm confused. I got the point that if you don't have any INIT or
CLEANUP method when the first instance of an object is created, it is
impossible to add one later in the same session, but your article
seems to imply that it is impossible to modify those methods either,
even if they successfully compile and run the first time round. I do
the latter all the time; in fact, it's one of the joys of non-blocking
widgets. The class *structure* can't be edited, but the lifecyle
methods can (IDL 5.2.1 on MacOS if it matters).
If the INIT or CLEANUP methods crash so that I get bounced back to
the command line with the offending routine opened for editing, IDL
seems to be in a very odd state, but typing RETALL and HEAP_GC several
times (at least twice) gets me back to the point where I can
sucessfully edit and recompile.
A final tip: it is worth including an INIT and CLEANUP routine for
all top-level classes (ie, those that don't inherit anything). If you
don't, IDL rummages around in your currently-defined search path
looking for one every time you create or destroy an instance of that
class, which can severely degrade performance.
Struan
|
|
|
|
| Re: Inheritance query [message #17639 is a reply to message #17629] |
Thu, 04 November 1999 00:00 |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
Bernard Puc (bpuc@va.aetc.com) writes:
> For the object programming gurus: I'm creating a class called data.
> I'm then creating subclasses of data called type1, type2, etc. The
> type1 class inherits the data class attributes. Now, is it possible to
> inherit, lets say, the data::INIT method and somehow add to it? Or, do
> I have to write an entirely new INIT method for type1 class which
> incorporates the statements in the data::INIT method?
If you don't define an INIT method for, say, your
TYPE1 object, IDL will look for an INIT method in
a subclass object and use that. If you decide
that you would like to add to an INIT method
for the TYPE1 object, and still use the INIT
method of a superclass, you can call the
superclass INIT method from within the TYPE1
INIT method. (The only time you can call
a lifecycle method yourself is from within
a subclass lifecycle method.) It might look
something like this:
FUNCTION TYPE1::INIT, _Extra=extra
ok = self->DATA::INIT(_Extra=extra)
IF NOT ok THEN RETURN, 0
Note this really strange behavior: If you don't
create an INIT method for your TYPE1 object
and then create a TYPE1 object, it will, of
course, use the INIT method of the inherited
DATA object. But if you now decide to write
a TYPE1::INIT method, you will never to able
to attach it to a TYPE1 object in that IDL
session! You will have to exit IDL and restart
for the correct INIT method to be associated
with the TYPE1 object.
I'm just about done with an article explaining
this phenomemon in more detail. It should be up
on my web page later today.
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438 E-Mail: davidf@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
