|
|
|
|
| Re: Automatic truncation of trailing dimension..... [message #28251 is a reply to message #28250] |
Tue, 27 November 2001 17:48  |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
tam <tam@lheapop.gsfc.nasa.gov> writes:
> Paul van Delst wrote:
>>
>> .....of an array when the dimension size is 1 is a real pain in the ass. Given:
...
>
> Under our previous president I guess I would have said
>
> "I feel your pain"
>
> but now it's
>
> "This is evil."
>
> You might find:
>
> < http://groups.google.com/groups?q=dimension+trailing&hl= en&group=comp.lang.idl-pvwave&rnum=8&selm=onlngh jin1.fsf%40cow.physics.wisc.edu>
>
> to be helpful.
What he said, errr, I mean, what I said.
Actually, that particular thread was a little more specialized than
Paul's complaint. Paul, I totally agree with you that dropping the
final unit-dimension is bogus, but the quoted post was premised upon
it happening, and then showing that there were *still* surprises after
that.
For example, the statement "A = DBLARR(N1, N2, N3)" can surprisingly
produce a two- or one-dimensional array!
Or, REFORM() does not work on scalars.
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@cow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
| Re: Automatic truncation of trailing dimension..... [message #28255 is a reply to message #28251] |
Tue, 27 November 2001 17:05 |
R.Bauer
Messages: 1424
Registered: November 1998
|
Senior Member |
|
|
"Pavel A. Romashkin" wrote:
>
> Oh, of course should be
>
> s = size(x)
> help, reform(x[*,*,1], s[1], 1)
>
> Duh!
>
> "Pavel A. Romashkin" wrote:
>>
>> Of course,
>>
>> s = size(x)
>> help, reform(x, s[1], 1)
Dear Pavel,
I have the same problems as you.
Therefore I have written a set routine this is attached.
If RSI won't change the state of removing last dimension 1 I like to
have a
compile option to extend these rule.
Everytime I have defined a three dimensional data set e.g. O3 in LAT and
LON and 1 time
the time is missing and this gives problems if I write HDF or netCDF
datafiles or
if I like to concatinate on the last dimension.
This was one of the reasons I have written dref and set.
It is so terrible if I have once changed my variable by reform and then
all is lost
if I initialize another variable like a=b then a is wrong while b is ok.
regards
Reimar
--
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 Juelich
http://www.fz-juelich.de/icg/icg1/idl_icglib/idl_lib_intro.h tml
http://www.fz-juelich.de/zb/text/publikation/juel3786.html
============================================================ ======
read something about linux / windows
http://www.suse.de/de/news/hotnews/MS.html
; Copyright (c) 2001, Forschungszentrum Juelich GmbH ICG-1
; All rights reserved.
; Unauthorized reproduction prohibited.
; This software may be used, copied, or redistributed as long as it is not
; sold and this copyright notice is reproduced on each copy made. This
; routine is provided as is without any express or implied warranties
; whatsoever.
;
;+
; NAME:
; set
;
; PURPOSE:
; This functions derefences pointers. If last dimension is 1 this is returned and not the
; standard IDL Array without last dimension 1.
; If value is a array of pointer the values are concatinated by concatinate_arrays at the last dimension
; If value isn't a pointer this value is returned but with the right dimensions.
;
;
; CATEGORY:
; PROG_TOOLS
;
; CALLING SEQUENCE:
; result=set(value,[/free])
;
; INPUTS:
; value: the pointer to exchange to it's value
;
;
; KEYWORD PARAMETERS:
; free: if set the pointer is purged from memory
;
; PROCEDURE:
; This routine should be used if it could be possible that's a value is a value or a pointer
; of the value. In difference to the idl dereference operator * this routine returns the
; right dimensions back if last dimension are 1.
; This function returns concatinated array values if a vector of pointer is submitted.
; Therefore the dimensions should be conform to do this.
;
; EXAMPLE:
; a=10
; help,set(a) & print,set(a,/free)
; A INT = 10
; 10
;
;
; a=ptr_new(10)
; help,set(a) & print,set(a,/free)
; A INT = 10
; 10
;
;
;
; a=reform(indgen(10),10,1)
; X=set(a)
; y=a
; help,set(a) & print,set(a,/free) & help, x & help,y
; A INT = Array[10, 1]
; 0 1 2 3 4 5 6 7 8 9
; X INT = Array[10,1]
; Y INT = Array[10]
;
;
; a=ptr_new(reform(indgen(10),10,1))
; X=set(a)
; Y=a
; help,set(a) & print,set(a,/free) & help, x & help,y
; A INT = Array[10, 1]
; 0 1 2 3 4 5 6 7 8 9
; X INT = Array[10, 1]
; Y POINTER = <PtrHeapVar1388>
;
;
;
;
;
; MODIFICATION HISTORY:
; Written by: R.Bauer (ICG-1), 2000-Jul-09
; 2001-11-18 : feature with last dimension eq 1 added
; : feature concatenate_arrays added: always on last dimension concatination is done
; : feature added : if value is no pointer this value is returned with the right dimensions
;-
FUNCTION set,value,free=free,_extra=p_key
IF SIZE(value[0],/TName) EQ 'POINTER' THEN BEGIN
n_ptr=SIZE(value,/N_ELEMENTS)
sz=SIZE(*value,/struct)
IF sz.n_dimensions LE 1 THEN BEGIN
IF N_ELEMENTS(result) EQ 0 THEN result=(*value)
ENDIF ELSE BEGIN
IF sz.dimensions[sz.n_dimensions-1] EQ 1 THEN BEGIN
IF N_ELEMENTS(result) EQ 0 THEN result=REFORM((*value),sz.dimensions[0:sz.n_dimensions-1])
ENDIF ELSE IF N_ELEMENTS(result) EQ 0 THEN result=(*value)
ENDELSE
IF KEYWORD_SET(free) THEN PTR_FREE,value
RETURN,result
ENDIF ELSE BEGIN
sz=SIZE(value,/struct)
IF sz.n_dimensions GE 2 THEN IF sz.dimensions[sz.n_dimensions-1] EQ 1 THEN BEGIN
RETURN,REFORM(value,sz.dimensions[0:sz.n_dimensions-1])
ENDIF ELSE RETURN,value
RETURN,value
ENDELSE
END
-
Attachment: set.pro
(Size: 3.38KB, Downloaded 87 times)
|
|
|
|
| Re: Automatic truncation of trailing dimension..... [message #28260 is a reply to message #28255] |
Tue, 27 November 2001 14:06 |
Paul van Delst
Messages: 364
Registered: March 1997
|
Senior Member |
|
|
tam wrote:
>
> Paul van Delst wrote:
>>
>> .....of an array when the dimension size is 1 is a real pain in the ass. Given:
>>
>> IDL> x=fltarr(100,1,15)
>> IDL> help, x
>> X FLOAT = Array[100, 1, 15]
>>
>> Is there anyway to prevent:
>>
>> IDL> help, x[*,*,1]
>> <Expression> FLOAT = Array[100]
>> IDL>
>>
>> i.e. to give:
>> <Expression> FLOAT = Array[100,1]
>>
>> Argh wot a pain.
>>
>> --
>
> Under our previous president I guess I would have said
>
> "I feel your pain"
>
> but now it's
>
> "This is evil."
Those darn evil evildoers.
>
> You might find:
>
<Link to Craig Markwardt's tirade against IDL mucking about and changing his data arrays
without at least asking first....snipped>
> to be helpful.
Oh, I've read Craig's, uh, comments about this, er, feature in the past.
> Everyone seems to be bitten by this at one time or another...
Well, usually those large chunks missing from my rear-end are my fault. Nuts. The chances of
this being FIXED by RSI are probably small since there is more than likely a lot more code that
would break if the auto-array-reformat wasn't done. Sigh.
paulv
--
Paul van Delst Religious and cultural
CIMSS @ NOAA/NCEP purity is a fundamentalist
Ph: (301)763-8000 x7274 fantasy
Fax:(301)763-8545 V.S.Naipaul
|
|
|
|
| Re: Automatic truncation of trailing dimension..... [message #28261 is a reply to message #28260] |
Tue, 27 November 2001 13:53 |
Paul van Delst
Messages: 364
Registered: March 1997
|
Senior Member |
|
|
"Liam E. Gumley" wrote:
>
> Paul van Delst wrote:
>>
>> .....of an array when the dimension size is 1 is a real pain in the ass. Given:
>>
>> IDL> x=fltarr(100,1,15)
>> IDL> help, x
>> X FLOAT = Array[100, 1, 15]
>>
>> Is there anyway to prevent:
>>
>> IDL> help, x[*,*,1]
>> <Expression> FLOAT = Array[100]
>> IDL>
>>
>> i.e. to give:
>> <Expression> FLOAT = Array[100,1]
>>
>> Argh wot a pain.
>
> I take it you mean
>
> <Expression> FLOAT = Array[100, 1, 1]
No, I mean
<Expression> FLOAT = Array[100,1]
i.e. an rank-2 array where one of the dimensions just happens to be 1. IDL reforms these by
default to a vector. This is a pain. Doesn't do it for arrays dimensioned as [1,100], so I
don't see why it should do it for the other.
> Does your code absolutely require it?
Not usually - except when I run test cases and I process one instance of something rather than
"X" hundred (atmospheric profiles).
> If you must maintain the dimensions:
>
> IDL> dims = size(x, /dimensions)
> IDL> index = 1
> IDL> help, reform(x[*, *, index], dims[0], dims[1], 1)
> <Expression> FLOAT = Array[100, 1, 1]
This is good to know but a pain in the rear to back-implement. To get things working. I just
set my minimum allowed dimension to 2 rather than 1. A work-around rather than a fix. I'm not
going to bother fixing something that shouldn't be broken - and besides, using IDL is supposed
to *enhance* my productivity, not decrease it. :o)
I find this auto-truncate behaviour for arrays with unity dimension in one particular order but
not the other quite ridiculous.
paulv
--
Paul van Delst Religious and cultural
CIMSS @ NOAA/NCEP purity is a fundamentalist
Ph: (301)763-8000 x7274 fantasy
Fax:(301)763-8545 V.S.Naipaul
|
|
|
|
| Re: Automatic truncation of trailing dimension..... [message #28262 is a reply to message #28261] |
Tue, 27 November 2001 13:45 |
tam
Messages: 48
Registered: February 2000
|
Member |
|
|
Paul van Delst wrote:
>
> .....of an array when the dimension size is 1 is a real pain in the ass. Given:
>
> IDL> x=fltarr(100,1,15)
> IDL> help, x
> X FLOAT = Array[100, 1, 15]
>
> Is there anyway to prevent:
>
> IDL> help, x[*,*,1]
> <Expression> FLOAT = Array[100]
> IDL>
>
> i.e. to give:
> <Expression> FLOAT = Array[100,1]
>
> Argh wot a pain.
>
> --
Under our previous president I guess I would have said
"I feel your pain"
but now it's
"This is evil."
You might find:
< http://groups.google.com/groups?q=dimension+trailing&hl= en&group=comp.lang.idl-pvwave&rnum=8&selm=onlngh jin1.fsf%40cow.physics.wisc.edu>
to be helpful.
I suppose you could do something like:
sz = size(x)
nvar = reform(x[*,*,1],sz[1],sz[2],1)
help,nvar
Everyone seems to be bitten by this at one time or another...
Regards,
Tom McGlynn
|
|
|
|
| Re: Automatic truncation of trailing dimension..... [message #28263 is a reply to message #28262] |
Tue, 27 November 2001 13:32 |
Liam E. Gumley
Messages: 378
Registered: January 2000
|
Senior Member |
|
|
Paul van Delst wrote:
>
> .....of an array when the dimension size is 1 is a real pain in the ass. Given:
>
> IDL> x=fltarr(100,1,15)
> IDL> help, x
> X FLOAT = Array[100, 1, 15]
>
> Is there anyway to prevent:
>
> IDL> help, x[*,*,1]
> <Expression> FLOAT = Array[100]
> IDL>
>
> i.e. to give:
> <Expression> FLOAT = Array[100,1]
>
> Argh wot a pain.
I take it you mean
<Expression> FLOAT = Array[100, 1, 1]
Does your code absolutely require it? In many cases, you can use the two
interchangeably.
If you must maintain the dimensions:
IDL> dims = size(x, /dimensions)
IDL> index = 1
IDL> help, reform(x[*, *, index], dims[0], dims[1], 1)
<Expression> FLOAT = Array[100, 1, 1]
Cheers,
Liam.
Practical IDL Programming
http://www.gumley.com/
|
|
|
|
| Re: Automatic truncation of trailing dimension..... [message #28264 is a reply to message #28263] |
Tue, 27 November 2001 13:31 |
Liam E. Gumley
Messages: 378
Registered: January 2000
|
Senior Member |
|
|
Paul van Delst wrote:
>
> .....of an array when the dimension size is 1 is a real pain in the ass. Given:
>
> IDL> x=fltarr(100,1,15)
> IDL> help, x
> X FLOAT = Array[100, 1, 15]
>
> Is there anyway to prevent:
>
> IDL> help, x[*,*,1]
> <Expression> FLOAT = Array[100]
> IDL>
>
> i.e. to give:
> <Expression> FLOAT = Array[100,1]
>
> Argh wot a pain.
I take it you mean
<Expression> FLOAT = Array[100, 1, 1]
Does your code absolutely require it? In many cases, you can use the two
interchangeably.
If you must maintain the dimensions:
IDL> dims = size(x, /dimensions)
IDL> index = 1
IDL> help, reform(x[*, *, index], dims[0], dims[1], 1)
<Expression> FLOAT = Array[100, 1, 1]
Cheers,
Liam.
Practical IDL Programming
http://www.gumley.com/
|
|
|
|
| Re: Automatic truncation of trailing dimension..... [message #28265 is a reply to message #28264] |
Tue, 27 November 2001 13:36 |
Pavel A. Romashkin
Messages: 531
Registered: November 2000
|
Senior Member |
|
|
Paul van Delst wrote:
>
> IDL> help, x[*,*,1]
> <Expression> FLOAT = Array[100]
> IDL>
>
> i.e. to give:
> <Expression> FLOAT = Array[100,1]
Now this is almost asking for magic. It would be more reasonable to ask for
<Expression> FLOAT = Array[100,1,1] :)
Of course,
s = size(x)
help, reform(x, s[1], 1)
But I think this was brought up so many times that by now it should be
on the request to RSI list. I mean, not truncating the empty dimension.
Those wanting truncation easily could use Reform, or a compile option
could be introduced to toggle this behavior.
Pavel
|
|
|
|
| Re: Automatic truncation of trailing dimension..... [message #28266 is a reply to message #28265] |
Tue, 27 November 2001 13:38 |
Pavel A. Romashkin
Messages: 531
Registered: November 2000
|
Senior Member |
|
|
Oh, of course should be
s = size(x)
help, reform(x[*,*,1], s[1], 1)
Duh!
"Pavel A. Romashkin" wrote:
>
> Of course,
>
> s = size(x)
> help, reform(x, s[1], 1)
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
