| Re: Structure assign question [message #28123] |
Mon, 26 November 2001 09:20  |
btt
Messages: 345
Registered: December 2000
|
Senior Member |
|
|
Hi Paul,
One approach you might consider is making each of your PDprofiles data structures into objects and
your PDdata structure an IDL container. I find this approach easier than wrestling with arrays of
pointers to structures with pointers that have other pointers... well, you know what I mean.
Also, you don't have to know how many profiles must be stored ahead of time. I prefer to base all
of my structures on Martin Schultz's MGS_BaseObject and I only use his MGS_Container in which he
has added a great many useful features to the basic IDL_Container class.
PRO PDdata__define
struct = {PDData, $
INHERITS MGS_Container}
END
; -- Loop over profiles
FOR m = 0L, n_profiles - 1L DO BEGIN
READU, file_id, k, j
profile = PDprofile_create( k, j ) ;<--- this would have to return an object
;<---- with all the bells and whistles
;<-----like Init, GetProperty, SetProperty
and Cleanup
self->Add, profile
n_layers[ m ] = k
n_absorbers[ m ] = j
ENDFOR
Ben
> Anyway, I have the following object/structure definition procedure:
>
> PRO PDdata__define
> PDcommon_definition
> COMMON PDcommon
> data_structure = { PDdata, n_profiles : IO_INT_TYPE, $
> profile : PTR_NEW() }
> END ; PRO PDdata__define
>
> nice and simple.
>
> In my PDdata::Read method, which fills the above data structure/object, I have the following:
>
> ; -- Create the top level PDdata structure
> self.n_profiles = n_profiles
> self.profile = PTR_NEW( REPLICATE( { PDprofile }, n_profiles ) )
>
> ; -----------------------
> ; Each profile dimension
> ; -----------------------
>
> ; -- Initialise arrays
> n_layers = LONARR( n_profiles ) & k = 0L
> n_absorbers = LONARR( n_profiles ) & j = 0L
>
> ; -- Loop over profiles
> FOR m = 0L, n_profiles - 1L DO BEGIN
>
> READU, file_id, k, j
>
> profile = PDprofile_create( k, j )
> (*self.profile)[ m ] = profile
>
> n_layers[ m ] = k
> n_absorbers[ m ] = j
>
> ENDFOR
>
> My question relates to the PDprofile_create() function. It returns a named structure (name is
> "PDprofile"). This structure is replete with various other (named) structures and pointers etc.
> hence the separate function. I then assign this structure to the required element of the data
> object, self.profile :
>
> (*self.profile)[ m ] = profile
>
> Here is where the question may or may not be considered brain-dead. Is this sort of (named)
> structure assign o.k., i.e. does the entire heirarchy of pointers and structures within the
> "profile" structure variable get assigned (it appears so at this point but I'm still testing).
> And am I creating a separate instance of the PDprofile structure or am I confusing the pointer
> with what the pointer is pointing at (get it?).
>
--
Ben Tupper
Bigelow Laboratory for Ocean Science
180 McKown Point Road
West Boothbay Harbor, ME 04575
www.bigelow.org
btupper@bigelow.org
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
