| Re: Unsupported keyword on older IDL version [message #31631] |
Mon, 29 July 2002 20:30 |
Paul Hick
Messages: 9
Registered: November 1999
|
Junior Member |
|
|
William Thompson wrote:
> Paul Hick <pphick@ucsd.edu> writes:
>
>
>> This kind of problem must have come up in this group before, but here it
>> is anyways.
>>
>
>> If I call the IDL systime function with the utc keyword:
>> print, systime(/utc)
>>
>
>> on IDL versions older than 5.4 I get the expected 'keyword not
>> supported' error message, since, well, it's not supported, right? The
>> error message pops up when the function containing the systime call is
>> compiled, so testing e.g. !version.release before executing the systime
>> call doesn't work.
>>
>
>> The only solution I have come up with so far is to put the systime call
>> inside a call_function call:
>> print, call_function( 'systime', /utc)
>> This will compile (I will still need to test for !version.release before
>> actually executing the statement of course).
>>
>
>> This to me looks like a somewhat illegitimate use of call_function. Not
>> that I mind bending the rules a bit, but are there any other (better)
>> techniques of dealing with keywords that don't work in all IDL versions.
>>
>
>
> Why do you call this "illegitimate"? To me this seems like just the sort of
> thing that call_function() was designed for. I've always done it this way in
> the past. In fact, I've usually used execute() in the past, but
> call_function() is more elegant.
>
> Bill Thompson
>
My main use for call_function has been as the IDL version of the Fortran
'external' mechanism: passing function names as arguments to
procedures that would then call the function using call_function. That's
a lot more powerful than the current keyword problem. But of course you
are right: whatever works ....
Paul
|
|
|
|
| Re: Unsupported keyword on older IDL version [message #31632 is a reply to message #31631] |
Mon, 29 July 2002 20:21  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Paul Hick (pphick@ucsd.edu) writes:
> The _extra mechanism is a fairly recent addition to IDL, isn't it?
Yes, I think it has only been added in the past 10 or 12 years.
Cheers,
David
P.S. Let's just say the response would probably be overwhelming
if we started an IDL 4.0 branch of the IDL Expert Programmer's
Association. :-)
--
David W. Fanning, Ph.D.
Fanning Software Consulting, Inc.
Phone: 970-221-0438, E-mail: david@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
| Re: Unsupported keyword on older IDL version [message #31633 is a reply to message #31632] |
Mon, 29 July 2002 19:41 |
Paul Hick
Messages: 9
Registered: November 1999
|
Junior Member |
|
|
Craig Markwardt wrote:
> Paul Hick <pphick@ucsd.edu> writes:
>
>
>> This kind of problem must have come up in this group before, but here it
>> is anyways.
>>
>> If I call the IDL systime function with the utc keyword:
>> print, systime(/utc)
>>
>> on IDL versions older than 5.4 I get the expected 'keyword not
>> supported' error message, since, well, it's not supported, right? The
>> error message pops up when the function containing the systime call is
>> compiled, so testing e.g. !version.release before executing the systime
>> call doesn't work.
>>
>> The only solution I have come up with so far is to put the systime call
>> inside a call_function call:
>> print, call_function( 'systime', /utc)
>> This will compile (I will still need to test for !version.release before
>> actually executing the statement of course).
>>
>
> Another possibility is to use the _EXTRA mechanism:
>
> ;; Assume EXTRA is undefined
> if double(!version.release) GE 5.4 then $
> extra = {utc: 1}
> value = systime(_EXTRA=extra)
>
> Anyway you cut it, you will need an IDL version check.
>
> Craig
>
>
The _extra mechanism is a fairly recent addition to IDL, isn't it? So
that would just move the problem to a different arena.
The systime function is 'special' in the sense that it is a 'built-in'
IDL function (there is no systime.pro file floating around anywhere that
I can see). That suggests another solution. The systime(/utc) call could
be hidden in a separate user-written function systime_for_idl5_4.
Calling this after checking !version.release would also work, since the
user-written routine would never get compiled on pre5.4 versions.
Paul
|
|
|
|
| Re: Unsupported keyword on older IDL version [message #31636 is a reply to message #31633] |
Mon, 29 July 2002 13:17 |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Mon, 29 Jul 2002 08:14:12 -0700, Craig Markwardt wrote:
> Paul Hick <pphick@ucsd.edu> writes:
>
>> This kind of problem must have come up in this group before, but here
>> it is anyways.
>>
>> If I call the IDL systime function with the utc keyword:
>> print, systime(/utc)
>>
>> on IDL versions older than 5.4 I get the expected 'keyword not
>> supported' error message, since, well, it's not supported, right? The
>> error message pops up when the function containing the systime call is
>> compiled, so testing e.g. !version.release before executing the systime
>> call doesn't work.
>>
>> The only solution I have come up with so far is to put the systime call
>> inside a call_function call:
>> print, call_function( 'systime', /utc)
>> This will compile (I will still need to test for !version.release
>> before actually executing the statement of course).
>
> Another possibility is to use the _EXTRA mechanism:
>
> ;; Assume EXTRA is undefined
> if double(!version.release) GE 5.4 then $
> extra = {utc: 1}
> value = systime(_EXTRA=extra)
>
> Anyway you cut it, you will need an IDL version check.
Unless you use _STRICT_EXTRA, the inheritance mechanism doesn't care if
you pass it bogus keywords... in fact, that deficiency is what prompted
the creation of _STRICT_EXTRA. Here you can use it to your advantage.
IDL> print,systime(/UTCABC)
% Keyword UTCABC not allowed in call to: SYSTIME % Execution halted at:
$MAIN$
IDL> print,systime(_EXTRA={UTCABC:1})
Mon Jul 29 13:08:09 2002
IDL> print,systime(_STRICT_EXTRA={UTCABC:1}) % Keyword UTCABC not allowed
in call to: SYSTIME % Execution halted at: $MAIN$
Unfortunately, for this case, it doesn't work, because SYSTIME went from
having none to some keywords, so in old versions of IDL (prior to the
introduction of keywords to SYSTIME), you get:
IDL> print,systime(_EXTRA={UTC:1})
print,systime(_EXTRA={UTC:1})
^
% Keyword parameters not allowed in call.
For routines with new keywords, which always had keywords, this should
work.
JD
|
|
|
|
| Re: Unsupported keyword on older IDL version [message #31637 is a reply to message #31636] |
Mon, 29 July 2002 13:35 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
I wrote earlier today:
> On the other hand, this does have the advantage of
> running in a run-time version of IDL, whereas the
> CALL_FUNCTION approach would fail.
I was apparently wrong about this. Dick Jackson points
out to me in a private e-mail that he can include
EXECUTE and CALL_FUNCTION in run-time programs and they
work perfectly OK.
I'm embarrassed to say I didn't try this myself
when I wrote this advice, but based it on a recent
posting to this newsgroup from the usually reliable
Reimar Bauer, who wrote:
"there is a small amount of rules you should know.
[about run-time IDL]
* !PATH does not exist in runtime
* the filename of the routine have to be the same as
the calling procedure.
* EXECUTE won't work (e.g. read_ascii uses EXECUTE)
* nothing could be compiled but other compiled routines
could be loaded."
Here is a test program. Runs fine on Windows 2000 with
IDL 5.5 in run-time. Does anyone else have problems with
it?
**********************************************************
PRO RuntimeFeaturesTest
READ_JPEG, FILEPATH('rose.jpg', SUBDIR=['examples', 'data']), image
IF image[0] NE -1 THEN TV, /True, image
Wait, 1
Call_Procedure, 'Plot', RandomU(seed, 10)
Wait, 1
pixels = TVRd()
count = Call_Function('Total', pixels)
XYOutS, /Device, 100, 100, 'Total of pixels: '+StrTrim(count, 2)
Wait, 1
ok = Execute('Erase')
Wait, 1
END
**********************************************************
IDL> runtimefeaturestest
IDL> Resolve_All
IDL> Save, File='runtimefeaturestest.sav', /Routines
What do you make of this, Reimar?
Cheers,
David
--
David W. Fanning, Ph.D.
Fanning Software Consulting, Inc.
Phone: 970-221-0438, E-mail: david@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
| Re: Unsupported keyword on older IDL version [message #31646 is a reply to message #31636] |
Mon, 29 July 2002 08:34 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Craig Markwardt (craigmnet@cow.physics.wisc.edu) writes:
> Another possibility is to use the _EXTRA mechanism:
>
> ;; Assume EXTRA is undefined
> if double(!version.release) GE 5.4 then $
> extra = {utc: 1}
> value = systime(_EXTRA=extra)
>
> Anyway you cut it, you will need an IDL version check.
I think this might depend on how old a version of IDL
you were calling. I don't think _Extra always failed
gracefully in the early days, when routines were passed
a keyword they didn't understand.
On the other hand, this does have the advantage of
running in a run-time version of IDL, whereas the
CALL_FUNCTION approach would fail.
Cheers,
David
--
David W. Fanning, Ph.D.
Fanning Software Consulting, Inc.
Phone: 970-221-0438, E-mail: david@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
| Re: Unsupported keyword on older IDL version [message #31647 is a reply to message #31646] |
Mon, 29 July 2002 08:14 |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
Paul Hick <pphick@ucsd.edu> writes:
> This kind of problem must have come up in this group before, but here it
> is anyways.
>
> If I call the IDL systime function with the utc keyword:
> print, systime(/utc)
>
> on IDL versions older than 5.4 I get the expected 'keyword not
> supported' error message, since, well, it's not supported, right? The
> error message pops up when the function containing the systime call is
> compiled, so testing e.g. !version.release before executing the systime
> call doesn't work.
>
> The only solution I have come up with so far is to put the systime call
> inside a call_function call:
> print, call_function( 'systime', /utc)
> This will compile (I will still need to test for !version.release before
> actually executing the statement of course).
Another possibility is to use the _EXTRA mechanism:
;; Assume EXTRA is undefined
if double(!version.release) GE 5.4 then $
extra = {utc: 1}
value = systime(_EXTRA=extra)
Anyway you cut it, you will need an IDL version check.
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@cow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
| Re: Unsupported keyword on older IDL version [message #31649 is a reply to message #31647] |
Mon, 29 July 2002 07:40 |
thompson
Messages: 584
Registered: August 1991
|
Senior Member |
|
|
Paul Hick <pphick@ucsd.edu> writes:
> This kind of problem must have come up in this group before, but here it
> is anyways.
> If I call the IDL systime function with the utc keyword:
> print, systime(/utc)
> on IDL versions older than 5.4 I get the expected 'keyword not
> supported' error message, since, well, it's not supported, right? The
> error message pops up when the function containing the systime call is
> compiled, so testing e.g. !version.release before executing the systime
> call doesn't work.
> The only solution I have come up with so far is to put the systime call
> inside a call_function call:
> print, call_function( 'systime', /utc)
> This will compile (I will still need to test for !version.release before
> actually executing the statement of course).
> This to me looks like a somewhat illegitimate use of call_function. Not
> that I mind bending the rules a bit, but are there any other (better)
> techniques of dealing with keywords that don't work in all IDL versions.
Why do you call this "illegitimate"? To me this seems like just the sort of
thing that call_function() was designed for. I've always done it this way in
the past. In fact, I've usually used execute() in the past, but
call_function() is more elegant.
Bill Thompson
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
