| no backwards compatibility in IDL 5.6 [message #34266] |
Thu, 27 February 2003 04:52  |
Alexander Rauscher
Messages: 9
Registered: August 2001
|
Junior Member |
|
|
sorry for posting the same thing twice under different subjects, but i
think this is important...
many of idl programs have to be adapted due do the non existing
backwards compatibility of atan (and probably many other functions) .
one wouldn't expect a change in such a fundamental function. so now
atan(z, /phase) gives the same result as atan(z) in older versions did,
where z is (re,im)... this is worse than stupid. this is dangerous.
or does anybody know away to circumvent the new "features" of IDL 5.6?
alex
|
|
|
|
| Re: no backwards compatibility in IDL 5.6 [message #34276 is a reply to message #34266] |
Mon, 03 March 2003 09:12  |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Fri, 28 Feb 2003 13:02:15 -0700, JD Smith wrote:
> On Fri, 28 Feb 2003 12:42:42 -0700, David Fanning wrote:
>
>> Pavel Romashkin (pavel_romashkin@hotmail.com) writes:
>>
>>> Why is EXECUTE used in this program? Why can't the value just be
>>> returned from each CASE? Execute will slow it down and as far as I can
>>> tell, does nothing special. There is no code that follows the CASE to
>>> prevent you from returning at any point. Will it not compile in 5.4
>>> with the extra keyword? I thought keyword mismatches are runtime
>>> errors. Am I missing something?
>>
>> Alright, here is why I am using EXECUTE. If I change the code to this:
>>
>> returnValue = 0.0
>> version = Float(!VERSION.Release)
>> IF (version LE 5.5) THEN returnValue = ATAN(imgpart, realpart) $
>> ELSE returnValue = ATAN(complexNum, /Phase)
>>
>> Then the code won't compile in IDL 5.4, complaining about the PHASE
>> keyword not being defined. :-(
>>
>>
>> P.S. The code *does* compile in IDL 5.5, by the way, even though the
>> PHASE keyword is not defined there, either.
>
> Which is when _STRICT_EXTRA was first introduced. Coincidence?
>
> My consipiracy theory: RSI switched from checking built-in system
> routine keywords at compile to run-time with v5.5. This isn't the
> first time they've done this type of thing: around v5.3, they switched
> from checking the validity of system variables at compile time to run
> time (which was rather convenient, actually).
>
> JD
Just to clarify, for the sake of RSI's hard working engineers: this
conspiracy theory, like many others, is entertaining, but entirely
untrue. IDL checks at compile time whether *any* keywords are allowed
at all. Since ATAN went from having none to having one keyword, this
explains the difference.
JD
|
|
|
|
| Re: no backwards compatibility in IDL 5.6 [message #34300 is a reply to message #34266] |
Fri, 28 February 2003 20:40 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Craig Markwardt (craigmnet@cow.physics.wisc.edu) writes:
> And even, the last EXECUTE is not required, if one uses CALL_FUNCTION
> instead, which should be a bit faster, not as clunky looking to me.
Yeah, I already changed that when I woke up from my nap. :-)
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: no backwards compatibility in IDL 5.6 [message #34301 is a reply to message #34266] |
Fri, 28 February 2003 18:20 |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
thompson@orpheus.nascom.nasa.gov (William Thompson) writes:
> Pavel Romashkin <pavel_romashkin@hotmail.com> writes:
>
>> Why is EXECUTE used in this program? Why can't the value just be
>> returned from each CASE? Execute will slow it down and as far as I can
>> tell, does nothing special. There is no code that follows the CASE to
>> prevent you from returning at any point. Will it not compile in 5.4 with
>> the extra keyword? I thought keyword mismatches are runtime errors. Am I
>> missing something?
>> Pavel
>
>
> Yes, without the execute statement, it will not compile in versions earlier
> than 5.4. You get the error message
>
> IDL> .run atan_complex_wrapper
>
> (version GE 5.6): returnValue = ATAN(complexNum, /Phase)
> ^
> % Keyword parameters not allowed in call.
> At: /disk1/thompson/atan_complex_wrapper.pro, Line 12
> % 1 Compilation errors in module ATAN_COMPLEX_WRAPPER.
>
> However, only the last execute statement is actually required. The first two,
> without the new keyword, can be direct statements.
And even, the last EXECUTE is not required, if one uses CALL_FUNCTION
instead, which should be a bit faster, not as clunky looking to me.
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@cow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
|
|
| Re: no backwards compatibility in IDL 5.6 [message #34306 is a reply to message #34266] |
Fri, 28 February 2003 15:32 |
Pavel Romashkin
Messages: 166
Registered: April 1999
|
Senior Member |
|
|
And exactly, who of you two folks report to whom? I am getting confused.
It must be those drinks at that banquet from which you, David, got the
headac... I mean, the cold :-)
Pavel
David Fanning wrote:
>
> Alright, another case closed by the ace detectives
> at the IEPA agency. I'll write a report, sir.
|
|
|
|
| Re: no backwards compatibility in IDL 5.6 [message #34307 is a reply to message #34266] |
Fri, 28 February 2003 12:35 |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Fri, 28 Feb 2003 13:12:17 -0700, William Thompson wrote:
> Pavel Romashkin <pavel_romashkin@hotmail.com> writes:
>
>> Why is EXECUTE used in this program? Why can't the value just be
>> returned from each CASE? Execute will slow it down and as far as I can
>> tell, does nothing special. There is no code that follows the CASE to
>> prevent you from returning at any point. Will it not compile in 5.4 with
>> the extra keyword? I thought keyword mismatches are runtime errors. Am I
>> missing something?
>> Pavel
>
>
> Yes, without the execute statement, it will not compile in versions
> earlier than 5.4. You get the error message
>
> IDL> .run atan_complex_wrapper
>
> (version GE 5.6): returnValue = ATAN(complexNum, /Phase)
> ^
> % Keyword parameters not allowed in call.
> At: /disk1/thompson/atan_complex_wrapper.pro, Line 12
> % 1 Compilation errors in module ATAN_COMPLEX_WRAPPER.
>
> However, only the last execute statement is actually required. The
> first two, without the new keyword, can be direct statements.
>
Aha! Another conspiracy laid to rest. IDL apparently checks whether
zero, or more than zero keywords are allowed.
Try compiling:
pro foo
empty,/FOO
end
and you'll see what I mean. Since ATAN went from having zero to more
than zero keywords, this explains it. The compile vs. runtime
handling of keywords (perverse as it is) has remained unchanged.
JD
|
|
|
|
| Re: no backwards compatibility in IDL 5.6 [message #34308 is a reply to message #34266] |
Fri, 28 February 2003 12:28 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
JD Smith (jdsmith@as.arizona.edu) writes:
> Which is when _STRICT_EXTRA was first introduced. Coincidence?
>
> My consipiracy theory: RSI switched from checking built-in system routine
> keywords at compile to run-time with v5.5. This isn't the first time
> they've done this type of thing: around v5.3, they switched from checking
> the validity of system variables at compile time to run time (which was
> rather convenient, actually).
Alright, another case closed by the ace detectives
at the IEPA agency. I'll write a report, sir.
Cheers,
David
P.S. There is a new wrapper program on my web page
that uses only a singe EXECUTE statement. :-)
--
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: no backwards compatibility in IDL 5.6 [message #34309 is a reply to message #34266] |
Fri, 28 February 2003 12:02 |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Fri, 28 Feb 2003 12:42:42 -0700, David Fanning wrote:
> Pavel Romashkin (pavel_romashkin@hotmail.com) writes:
>
>> Why is EXECUTE used in this program? Why can't the value just be
>> returned from each CASE? Execute will slow it down and as far as I can
>> tell, does nothing special. There is no code that follows the CASE to
>> prevent you from returning at any point. Will it not compile in 5.4
>> with the extra keyword? I thought keyword mismatches are runtime
>> errors. Am I missing something?
>
> Alright, here is why I am using EXECUTE. If I change the code to this:
>
> returnValue = 0.0
> version = Float(!VERSION.Release)
> IF (version LE 5.5) THEN returnValue = ATAN(imgpart, realpart) $
> ELSE returnValue = ATAN(complexNum, /Phase)
>
> Then the code won't compile in IDL 5.4, complaining about the PHASE
> keyword not being defined. :-(
>
>
> P.S. The code *does* compile in IDL 5.5, by the way, even though the
> PHASE keyword is not defined there, either.
Which is when _STRICT_EXTRA was first introduced. Coincidence?
My consipiracy theory: RSI switched from checking built-in system routine
keywords at compile to run-time with v5.5. This isn't the first time
they've done this type of thing: around v5.3, they switched from checking
the validity of system variables at compile time to run time (which was
rather convenient, actually).
JD
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
