| Re: Structures: [message #23836] |
Fri, 23 February 2001 11:37 |
Chris Bull
Messages: 7
Registered: January 2001
|
Junior Member |
|
|
Thanks all, some useful Ideas here to ponder on!
Cheers :*)
"Chris Bull" <cjbull@another.com> wrote in message
news:owUk6.10166$5n4.162849@news6-win.server.ntlworld.com...
> Hi all,
>
> A question regarding IDL structures, I have an application that I am
> writing that I am using a very complex
> structure as a Uvalue to hold my data in, as this application evolves I am
> wondering whether it
> is possible to add 'branches' to a structure after it is defined (and set
as
> a uvalue), from within a seperate subroutine?
>
> I'm Thinking along the lines of describing an application that can hold
many
> data sets, of different types in memory
> and load them up seperatly (much like a imaging program or multiple
> documents in word) except the data sets are
> very complex and of differing structures
>
> Any Ideas?
>
> Sorry I dont have the code here! as this is my home PC not work
>
> Regards
>
>
> Chris Bull
>
>
>
>
>
|
|
|
|
| Re: Structures: [message #23868 is a reply to message #23836] |
Wed, 21 February 2001 23:24  |
R.Bauer
Messages: 1424
Registered: November 1998
|
Senior Member |
|
|
Chris Bull wrote:
>
> Hi all,
>
> A question regarding IDL structures, I have an application that I am
> writing that I am using a very complex
> structure as a Uvalue to hold my data in, as this application evolves I am
> wondering whether it
> is possible to add 'branches' to a structure after it is defined (and set as
> a uvalue), from within a seperate subroutine?
>
> I'm Thinking along the lines of describing an application that can hold many
> data sets, of different types in memory
> and load them up seperatly (much like a imaging program or multiple
> documents in word) except the data sets are
> very complex and of differing structures
>
> Any Ideas?
Yes it's possible if you have anonymous structure.
the trick is to store all structure levels to a string vector
and all tagvalues to a vector of pointers.
Then the new values in both are inserted and the structure is
new generated.
This is quite fast.
For example add_tag
; d = {A: 1, B: {B1: 0, B2: 1}, C: {B1: 0, B2: 1}}
;
; help,d,/str
; ** Structure <1331608>, 3 tags, length=10, refs=1
; A INT 1
; B STRUCT -> <Anonymous> Array[1]
; C STRUCT -> <Anonymous> Array[1]
; result = add_tag(d, 'ASA', 11, sub = 'B')
; HELP, result.b, /STR
; ** Structure <133ce88>, 3 tags, length=6, refs=1
; B1 INT 0
; B2 INT 1
; ASA INT 11
; result = add_tag(d, 'BSB', {C11: INDGEN(3)}, sub = 'B.B3')
; HELP, result.b, /STR
; ** Structure <1357758>, 3 tags, length=10, refs=2:
; B1 INT 0
; B2 INT 1
; B3 STRUCT -> <Anonymous> Array[1]
Please have a look at the following routines.
http://www.fz-juelich.de/icg/icg1/idl_icglib/idl_source/idl_ html/dbase/download/replace_tagvalue.tar.gz
http://www.fz-juelich.de/icg/icg1/idl_icglib/idl_source/idl_ html/dbase/download/add_tag.tar.gz
http://www.fz-juelich.de/icg/icg1/idl_icglib/idl_source/idl_ html/dbase/download/rename_tag.tar.gz
http://www.fz-juelich.de/icg/icg1/idl_icglib/idl_source/idl_ html/dbase/download/delete_tag.tar.gz
For further routines and licensing please look at
http://www.fz-juelich.de/icg/icg1/idl_icglib/idl_lib_intro.h tml
--
Reimar Bauer
Institut fuer Stratosphaerische Chemie (ICG-1)
Forschungszentrum Juelich
email: R.Bauer@fz-juelich.de
http://www.fz-juelich.de/icg/icg1/
=============================================
a IDL library at ForschungsZentrum J�lich
http://www.fz-juelich.de/icg/icg1/idl_icglib/idl_lib_intro.h tml
http://www.fz-juelich.de/zb/text/publikation/juel3786.html
|
|
|
|
| Re: Structures: [message #23870 is a reply to message #23868] |
Wed, 21 February 2001 13:58 |
Jason P. Meyers
Messages: 24
Registered: September 2000
|
Junior Member |
|
|
Try using anonymous structures. Here is a quick and dirty demo that
illustrates it can be done. Note, you can even store structures within
anonymous structures!
I hope this helps.
Example:
pro test_struct, data
If N_Elements(data) gt 0 then print, data
data = {a:3, b:'Hello World', c:indgen(3,3)}
print, ''
help, data
print, 'Data is now:'
print, data
data = {a:4, b:'Hello World', c:indgen(4,4), d:indgen(3,3)}
print, ''
help, data
print, 'Data is now:'
print, data
end
--
Jason Meyers
Ph.D. Student, Center for Imaging Science
Rochester Institute of Technology
jpm7934@rit.edu
|
|
|
|
| Re: Structures: [message #23872 is a reply to message #23870] |
Wed, 21 February 2001 12:52 |
Mark Hadfield
Messages: 783
Registered: May 1995
|
Senior Member |
|
|
"Chris Bull" <cjbull@another.com> wrote in message
news:owUk6.10166$5n4.162849@news6-win.server.ntlworld.com...
> A question regarding IDL structures, I have an application that I am
> writing that I am using a very complex
> structure as a Uvalue to hold my data in, as this application evolves I am
> wondering whether it
> is possible to add 'branches' to a structure after it is defined (and set
as
> a uvalue), from within a seperate subroutine?
Well, yes and no. The IDL built-in CREATE_STRUCT allows you to add tags to a
structure, but what it actually does is erase the original value and
generate a new one, i.e. a full copy of the structure data in memory. This
is fine for small structures, but inefficient on large ones. It is a
consequence of the decisions IDL made when they designed the structure data
type: keep it contiguous (except for padding) in memory, simple and fast.
> I'm Thinking along the lines of describing an application that can hold
many
> data sets, of different types in memory
> and load them up seperatly (much like a imaging program or multiple
> documents in word) except the data sets are
> very complex and of differing structures
You could use a data container base on pointers, but accessing and modifying
this will get complicated. Me, I get confused by all those dereferncing
symbols, e.g. *(*c(*a.b)))
So my reccommendation is to use an object-based data container. There are no
adequate ones supplied with IDL (more's the pity) so you will have to write
your own or adapt someone else's. For a start you could look at a few on my
WWW page: go to http://katipo.niwa.cri.nz/~hadfield/gust/software/idl/ and
look at:
mgh_vector__define
mgh_queue__define
mgh_stack__define
The vector is the one that might be of use. It is a 1-D vector that can hold
data elements of any type, indexed by number. The size is adjusted
dynamically & automatically as elements are added.
A few months ago I tried writing a similar "dictionary" object, i.e. one
that holds a series of name:value pairs (rather like an IDL structure.) I
found that accessing the data elements was too slow. Such data structures
are common and provide reasonable performance in other languages (e.g.
Python, Matlab) so it should be possible in principle to create one in IDL,
but you've got to get clever in the way you look up the names.
--
---
Mark Hadfield
m.hadfield@niwa.cri.nz http://katipo.niwa.cri.nz/~hadfield
National Institute for Water and Atmospheric Research
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
