[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ascii_template

Mark Hadfield wrote:
> "Paul van Delst" <pvandelst@ncep.noaa.gov> wrote in message
> news:3989BC70.16B7BA85@ncep.noaa.gov...
> > David Fanning wrote:
> > >
> > >    FOR j=0,zillion-1 DO BEGIN
> >
> > Should be
> >
> >      FOR j=0L,zillion-1 DO BEGIN
> For a zillion you need at least a 64-bit integer.
> But seriously, don't you use "compile_opt IDL2"?

I didn't even know what that was. After checking the on-line help, I see
it is a command that is equivalent to the -d8 or -i4 on some Fortran
compilers that automatically elevate certain default type definitions.

Now, this is my Fortran half talking, I *never* like to use these
features. If in some IDL code I type:

FOR i = 0, n-1 DO BEGIN....

when I look at code surrounding that statement I know that "i" is a
short (read 2-byte) integer and thus the numbers will never exceed the
values allowed by this data type. I guess it's just the years and years
of poring over old Fortran code full of default type definitions and
default type elevations causing havoc that induce this bitterness.
That's why all my F90 code starts IMPLICIT NONE.

Then again, it's not like I'll ever deliver a software product written
in IDL to a general audience (other than mates and colleagues who want a
robust planck function procedure :o). Fortran-90/95/200x yes. IDL no. I
don't know if majority is a better word, but a lot of people I supply
code to (for free) don't use IDL for a number of reasons (too expensive,
too slow for operational use, etc.) It is noteworthy that the number
crunching hardware/software used by most atmospheric science type bods
comes bundled with all manner of compilers - the cost of which is
transparent. What good is an SGI Origin without a C or Fortran

At a satellite data processing conference I was at last year IDL was
rejected outright as a tool for data product visualisation as it cost
extra on top of what people pay for hardware. For those that can afford
fancy stuff like IDL great, but that doesn't help the weather
forecasters in countries that have very limited budgets and rely,
partially, on software "donated" by the user community.

Sorry for blathering totally off topic. Please feel free to comment -
seeing the "big picture" is not one of my strengths.


p.s. You can replace all instances of "IDL" in the above with "Matlab"
or similar products and the conclusions differ not.

Paul van Delst           Ph:  (301) 763-8000 x7274
CIMSS @ NOAA/NCEP        Fax: (301) 763-8545
Rm.202, 5200 Auth Rd.    Email: pvandelst@ncep.noaa.gov
Camp Springs MD 20746