| Re: Pre-processor-like Capability? [message #6405] |
Sat, 15 June 1996 00:00 |
dave
Messages: 31
Registered: February 1994
|
Member |
|
|
From: dave@msrad71.med.uth.tmc.edu (David Fenyes)
Subject: Re: Pre-processor-like Capability?
Newsgroups: comp.lang.idl-pvwave
Date: 13 Jun 1996 13:36:50 -0500
Organization: UTHSC-Houston, Dept of Radiology
Path: news.uth.tmc.edu!usenet
Lines: 44
Message-ID: <x6afy78sal.fsf@msrad71.med.uth.tmc.edu>
References: <31BD4280.351B@hsonline.net>
NNTP-Posting-Host: msrad71.med.uth.tmc.edu
Mime-Version: 1.0
Content-Type: text/plain;charset=US-ASCII
Content-Transfer-Encoding: 7bit
In-reply-to: "Bruce E. Thomason"'s message of Tue, 11 Jun 1996 09:55:12 +0000
X-Newsreader: Gnus v5.1
Some typos:
> FUNCTION magic_number
> return 'boing'
> END
> PRO whatever, x, y, z
forward_function magic_number
> . . .
> IF a NE magic_number THEN BEGIN
^ magic_number()
> print, "Bad Magic"
> return, 0
> ENDIF
> . . .
> END
David.--
David Fenyes University of Texas Medical School
dave@msrad72.med.uth.tmc.edu Dept. of Radiology
|
|
|
|
| Re: Pre-processor-like Capability? [message #6416 is a reply to message #6405] |
Thu, 13 June 1996 00:00  |
dave
Messages: 31
Registered: February 1994
|
Member |
|
|
>>>> > "Bruce" == Bruce E Thomason <brucet@hsonline.net> writes:
Bruce> Does anyone know of a IDL/PV-Wave capability that is
Bruce> analogous to C's preprocessor?
Bruce> I am particularly interested in efficiently establishing
Bruce> symbolic references (in the coding process) to constant
Bruce> values for use across many routines (as #define does w/ the
Bruce> C pre-processor), but to have these values efficiently
Bruce> accessed at run-time.
I create small functions at the beginning of the source that return the
desired values:
FUNCTION magic_number
return 'boing'
END
PRO whatever, x, y, z
. . .
IF a NE magic_number THEN BEGIN
print, "Bad Magic"
return, 0
ENDIF
. . .
END
This is portable, can be included in one source file, and used in
another, and is almost as simple as a #define. It's a good idea to
use a FORWARD_FUNCTION declaration to declare all such functions at
the beginning of a procedure, in case the definition is in another
file.
Performace won't suffer unless your code is highly looped, in which
case you chould probably be linking in a C for FORTRAN module anyway.
Good luck,
David.
--
David Fenyes University of Texas Medical School
dave@msrad72.med.uth.tmc.edu Dept. of Radiology
|
|
|
|
| Re: Pre-processor-like Capability? [message #6431 is a reply to message #6416] |
Tue, 11 June 1996 00:00 |
thompson
Messages: 584
Registered: August 1991
|
Senior Member |
|
|
"Bruce E. Thomason" <brucet@hsonline.net> writes:
> Does anyone know of a IDL/PV-Wave capability that is analogous to
> C's preprocessor?
> I am particularly interested in efficiently establishing symbolic
> references (in the coding process) to constant values for use across
> many routines (as #define does w/ the C pre-processor), but to have
> these values efficiently accessed at run-time.
> The two alternative's I've come up with are:
> 1. Declaring variables via an "include" (@xxx) file. This has the down
> side of potentially including declarations to variables that are not
> used in a routine, as well as accessing variables in what (I expect) is
> a less efficient manner than accessing a literal value.
> 2. Using system variables, which I also expect is less efficient that
> accessing a literal value.
> I would prefer a language intrinsic approach, but might have to settle
> for writing/using a (language independent) preprocessor. Does anyone
> have a recommendation in this regard?
I suggest you follow your second option above. Use system parameters, and set
the actual values in the IDL startup file.
Considering the way that IDL works, I don't think that there is any significant
performance difference in using a literal value versus storing the value in a
system variable. In some sense, IDL uses a "preprocessor". The first time
that you call a procedure, it is read and converted into a format that IDL uses
to actually run it. I'm not sure in detail what happens to literal values
embedded in the code, but expect that they're converted to temporary parameters
which are accessed in a very similar way as other parameters.
Bill Thompson
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
