Mirko Vukovic wrote:
>
> J.D. Smith wrote:
>>
>> Another in a long line of complaints about keyword inheritance. It
>> seems that now (though I don't recall this always having been a
>> problem), a wrong keyword is not flagged.
>>
>> E.g.
>>
>> pro testex2,b=b
>> print,'test2'
>> return
>> end
>> pro testex, a=a,_EXTRA=e
>> print,'test1'
>> help,/st,e
>> testex2, _EXTRA=e
>> return
>> end
>>
>> IDL> testex,b=1
>> test1
>> ** Structure <82441b4>, 1 tags, length=2, refs=1:
>> B INT 1
>> test2
>>
>> IDL> testex,a=1,b=1
>> test1
>> ** Structure <81ad99c>, 1 tags, length=2, refs=1:
>> B INT 1
>> test2
>>
>> IDL> testex,a=1,b=1,c=1
>> test1
>> ** Structure <8234884>, 2 tags, length=4, refs=1:
>> B INT 1
>> C INT 1
>> test2
>>
>> This last one should have been an error in testex2! C is not a valid
>> keyword to testex2, but it was silently ignored. I don't think this has
>> always been the case. It's a real problem if you mistype a keyword, but
>> aren't alerted.. you may not realize that your program didn't recieve a
>> keyword argument. Has anyone else noticed this?
>>
>> JD
> i don't see a problem. if c had made it into testex2, yes, but e in
> testex cannot know what routines it will be passed on to.
> --
I don't expect e to know what is inside of it, but when the choice is
made as to where e's keywords will be directed, I recall in prior
versions that the keywords were checked. I even went back and ran IDLv4
to verify. To make my point a little more clear, consider this
different pair of procedures:
pro testex2,thisreallycoolkeyword=b
if keyword_set(b) then print,'The magic answer is 42' else $
print,'Cupiditas est radix omnia malorum.'
end
pro testex, a=a,_EXTRA=e
if keyword_set(a) then print, 'A is set'
help,e,/st
testex2, _EXTRA=e
return
end
when run:
IDLv4:
IDL4> testex,/a,/thisreallycoolkeyword
A is set
** Structure <225e38>, 1 tags, length=2, refs=1:
THISREALLYCOOLKEYWORDINT 1
The magic answer is 42
IDLv5:
IDL5> testex,/a,/thisreallycoolkeyword
A is set
** Structure <81c3124>, 1 tags, length=2, refs=1:
THISREALLYCOOLKEYWORD
INT = 1
The magic answer is 42
All is as it is supposed to be there. But suppose I accidentally
mistype the keyword, as:
IDLv4
IDL4> testex,/a,/thisrealycoolkeyword
A is set
** Structure <225e78>, 1 tags, length=2, refs=1:
THISREALYCOOLKEYWORDINT 1
% Keyword THISREALYCOOLKEYWORD not allowed in call to: TESTEX2
% Execution halted at: TESTEX 9 test2.pro
% TESTEX 9 test2.pro
% $MAIN$
IDLv5
IDL5> testex,/a,/thisrealycoolkeyword
A is set
** Structure <81c3124>, 1 tags, length=2, refs=1:
THISREALYCOOLKEYWORD
INT = 1
Cupiditas est radix omnia malorum.
As you can see, in v4 my incorrect keyword was flagged, and an error
resulted. In v5, things run without interruption, as if I had not
provided any extra keywords at all. Of course in my testex2, this is
apparent from which message is printed, but for a general routine, the
user may receive no indication that his keyword was mistyped, and that,
consequently, the program isn't doing what he thinks it is.
On the other hand, this "feature" is eminently useful when inherited
keywords are going to be passed to several different routines. You can
just give _EXTRA=e to all of them, and only the relevant ones will be
used. It's not exactly that straightforward, though. If two keywords
are the same, or even begin the same, in these two or more routines,
then you'll get no warning, and both will be called... an example of
this is:
pro testex3,key3=c
if keyword_set(c) then $
print,'Ein spatz in der Hand is besser als eine Taube auf dem Dach'
else $
print,'I like candy.'
end
pro testex2,key2=b
if keyword_set(b) then print,'The magic answer is 42' else $
print,'Cupiditas est radix omnia malorum.'
end
pro testex, _EXTRA=e
help,e,/st
testex2, _EXTRA=e
testex3, _EXTRA=e
return
end
And now (v5):
IDL> testex
E UNDEFINED = <Undefined>
Cupiditas est radix omnia malorum.
I like candy.
IDL> testex,/key2
** Structure <81c35ec>, 1 tags, length=2, refs=1:
KEY2 INT 1
The magic answer is 42
I like candy.
IDL> testex,/key3
** Structure <81c360c>, 1 tags, length=2, refs=1:
KEY3 INT 1
Cupiditas est radix omnia malorum.
Ein spatz in der Hand is besser als eine Taube auf dem Dach
IDL> testex,/key2,/key3
** Structure <81c360c>, 2 tags, length=4, refs=1:
KEY2 INT 1
KEY3 INT 1
The magic answer is 42
Ein spatz in der Hand is besser als eine Taube auf dem Dach
all as they should be. But what happens with:
IDL> testex,/key
** Structure <81c3634>, 1 tags, length=2, refs=1:
KEY INT 1
The magic answer is 42
Ein spatz in der Hand is besser als eine Taube auf dem Dach
Egad! As you can see, the root of the keywords being the same in both
routines, both recieved it! This is actually the same behavior in V4,
but no less a cause for being careful.
What would be quite useful, although likely equally impossible, is a
system by which any leftover inherited keywords, that is keywords which
*don't* get through to any internal routine calls, are flagged as
errors. One can only dream.
JD
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
