| Re: V8.2 compilation error, works in V8.1 [message #80390] |
Thu, 07 June 2012 10:19 |
wlandsman
Messages: 743
Registered: June 2000
|
Senior Member |
|
|
For completeness, I note that the differences between V8.2 and V8.1 also occur (not unexpectedly) with lists.
IDL> b = list('header',0)
IDL> n = [1]
IDL> t = b.get(position=n)
IDL> help,t
In V8.1
T LONG = 0
But in V8.2
T LIST <ID=6 NELEMENTS=1>
Lajos pointed out that this change made the behavior consistent with subscripting an array with a one element array vs. with a scalar. But whereas it is often doesn't matter whether one has a scalar or a one element array, it will always make a difference when code that used to return a scalar integer now returns a list data type.
On Friday, June 1, 2012 1:57:08 PM UTC-4, wlandsman wrote:
> Lajos, Mike
>
> Thanks for the IDL_Object suggestion -- that seems like the simplest fix.
>
> Here's another IDL 8.2 change that is giving us compilation errors:
>
> IDL> print,!version
> { x86_64 linux unix linux 8.1 Mar 9 2011 64 64}
> IDL> x = HASH('key1',1,'key2',2)
> IDL> y = ['key1']
> IDL> help,x[y]
> <Expression> LONG = 1
>
> IDL> print,!version
> { x86_64 linux unix linux 8.2 Apr 10 2012 64 64}
> IDL> x = HASH('key1',1,'key2',2)
> IDL> y = ['key1']
> IDL> help,x[y]
> <Expression> HASH <ID=8 NELEMENTS=1>
>
> In this case, I think the V8.1 interpretation makes more sense. --Wayne
|
|
|
|
| Re: V8.2 compilation error, works in V8.1 [message #80432 is a reply to message #80390] |
Fri, 01 June 2012 11:54  |
b_gom
Messages: 105
Registered: April 2003
|
Senior Member |
|
|
I ran into the same problem with some older object code that
mysteriously stopped working. Instead of spending a few hours tracking
the problem down to the missing INHERITS IDL_OBJECT line, I should
have just waited for the group to solve it for me!
|
|
|
|
| Re: V8.2 compilation error, works in V8.1 [message #80433 is a reply to message #80432] |
Fri, 01 June 2012 11:18 |
Lajos Foldy
Messages: 176
Registered: December 2011
|
Senior Member |
|
|
On Friday, June 1, 2012 7:57:08 PM UTC+2, wlandsman wrote:
> Lajos, Mike
>
> Thanks for the IDL_Object suggestion -- that seems like the simplest fix.
>
> Here's another IDL 8.2 change that is giving us compilation errors:
>
> IDL> print,!version
> { x86_64 linux unix linux 8.1 Mar 9 2011 64 64}
> IDL> x = HASH('key1',1,'key2',2)
> IDL> y = ['key1']
> IDL> help,x[y]
> <Expression> LONG = 1
>
> IDL> print,!version
> { x86_64 linux unix linux 8.2 Apr 10 2012 64 64}
> IDL> x = HASH('key1',1,'key2',2)
> IDL> y = ['key1']
> IDL> help,x[y]
> <Expression> HASH <ID=8 NELEMENTS=1>
>
> In this case, I think the V8.1 interpretation makes more sense. --Wayne
The IDL 8.2 behaviour is consistent with the array subscripting version:
IDL> x=[1,2,3] & y=[1] & help, x[y]
<Expression> INT = Array[1]
I think this was the motivation for the change.
regards,
Lajos
|
|
|
|
| Re: V8.2 compilation error, works in V8.1 [message #80434 is a reply to message #80433] |
Fri, 01 June 2012 10:57 |
wlandsman
Messages: 743
Registered: June 2000
|
Senior Member |
|
|
Lajos, Mike
Thanks for the IDL_Object suggestion -- that seems like the simplest fix.
Here's another IDL 8.2 change that is giving us compilation errors:
IDL> print,!version
{ x86_64 linux unix linux 8.1 Mar 9 2011 64 64}
IDL> x = HASH('key1',1,'key2',2)
IDL> y = ['key1']
IDL> help,x[y]
<Expression> LONG = 1
IDL> print,!version
{ x86_64 linux unix linux 8.2 Apr 10 2012 64 64}
IDL> x = HASH('key1',1,'key2',2)
IDL> y = ['key1']
IDL> help,x[y]
<Expression> HASH <ID=8 NELEMENTS=1>
In this case, I think the V8.1 interpretation makes more sense. --Wayne
|
|
|
|
| Re: V8.2 compilation error, works in V8.1 [message #80435 is a reply to message #80434] |
Fri, 01 June 2012 10:23 |
Lajos Foldy
Messages: 176
Registered: December 2011
|
Senior Member |
|
|
On Friday, June 1, 2012 5:51:31 PM UTC+2, wlandsman wrote:
> We are getting many compilation errors in V8.2 for code that worked in V8.1. Apparently, there was a loophole in V8.1 that allowed one to directly access an object's variables in certain case, but now a line like
>
> self.obj2.index
>
> gives an error
> % Expression must be a structure in this context: <OBJREF (<ObjHeapVar2(OBJ2)>)>
>
> One can always access the variable with a GETPROPERTY, and I suppose this might have been a "bug fix". Is there any place where one can get a list of the V8.2 bug fixes?
>
> Below I give an example with two simple objects.
>
> IDL> t = obj_new('obj1')
> IDL> t.printvalue
>
> gives an error in V8.2 but not V8.1
>
> ;obj1__define.pro
> pro obj1::printvalue,index=index
> print,self.obj2.index
> end
> pro obj1::getproperty,obj2=obj2
> if arg_present(obj2) then obj2 = self.obj2
> end
> function obj1::init,obj2=obj2
> if ~isa(obj2,'obj2') then self.obj2 = obj_new('obj2')
> return,1
> end
> pro obj1__define
> compile_opt idl2
> obj1 = {obj1, $
> obj2:obj_new()}
> end
>
> ;obj2__define.pro
> pro obj2::getproperty,index=index
> if arg_present(index) then index = self.index
> end
> pro obj2__define
> compile_opt idl2
> obj2 = {obj2, $
> index:0L}
> end
After replacing 'print,self.obj2.index' with 'print,(self.obj2).index', it works. So the loophole is still there :-)
regards,
Lajos
|
|
|
|
| Re: V8.2 compilation error, works in V8.1 [message #80436 is a reply to message #80435] |
Fri, 01 June 2012 09:57 |
Lajos Foldy
Messages: 176
Registered: December 2011
|
Senior Member |
|
|
On Friday, June 1, 2012 5:51:31 PM UTC+2, wlandsman wrote:
> We are getting many compilation errors in V8.2 for code that worked in V8.1. Apparently, there was a loophole in V8.1 that allowed one to directly access an object's variables in certain case, but now a line like
>
> self.obj2.index
>
> gives an error
> % Expression must be a structure in this context: <OBJREF (<ObjHeapVar2(OBJ2)>)>
>
> One can always access the variable with a GETPROPERTY, and I suppose this might have been a "bug fix". Is there any place where one can get a list of the V8.2 bug fixes?
>
> Below I give an example with two simple objects.
>
> IDL> t = obj_new('obj1')
> IDL> t.printvalue
>
> gives an error in V8.2 but not V8.1
>
> ;obj1__define.pro
> pro obj1::printvalue,index=index
> print,self.obj2.index
> end
> pro obj1::getproperty,obj2=obj2
> if arg_present(obj2) then obj2 = self.obj2
> end
> function obj1::init,obj2=obj2
> if ~isa(obj2,'obj2') then self.obj2 = obj_new('obj2')
> return,1
> end
> pro obj1__define
> compile_opt idl2
> obj1 = {obj1, $
> obj2:obj_new()}
> end
>
> ;obj2__define.pro
> pro obj2::getproperty,index=index
> if arg_present(index) then index = self.index
> end
> pro obj2__define
> compile_opt idl2
> obj2 = {obj2, $
> index:0L}
> end
Add 'inherits IDL_Object' to your definitions.
("If your object inherits from the IDL_Object class, you can retrieve property values by calling the property directly with the dot operator.")
regards,
Lajos
|
|
|
|
| Re: V8.2 compilation error, works in V8.1 [message #80437 is a reply to message #80436] |
Fri, 01 June 2012 09:43 |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
On 6/1/12 9:51 AM, wlandsman wrote:
> We are getting many compilation errors in V8.2 for code that worked
> in V8.1. Apparently, there was a loophole in V8.1 that allowed one to
> directly access an object's variables in certain case, but now a line like
>
> self.obj2.index
>
> gives an error % Expression must be a structure in this
> context:<OBJREF
(<ObjHeapVar2(OBJ2)>)>
>
> One can always access the variable with a GETPROPERTY, and I suppose
> this might have been a "bug fix". Is there any place where one can get a
> list of the V8.2 bug fixes?
Yes, I confirm your code seems to indicate that the loophole allowing
member variable access of an object from another object's method has
been eliminated. I would count that as a bug fix, but I wish it had been
mentioned in the release notes. I think there is going to be some broken
code that will need to be updated for IDL 8.2.
Also, note that by making obj2 inherit from IDL_Object will make the
code work again on IDL 8.2, i.e., update obj2__define to:
pro obj2__define
compile_opt idl2
obj2 = { obj2, inherits IDL_Object, index:0L }
end
Mike
--
Michael Galloy
www.michaelgalloy.com
Modern IDL: A Guide to IDL Programming (http://modernidl.idldev.com)
Research Mathematician
Tech-X Corporation
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
