lats4d

Brian Doty doty at COLA.IGES.ORG
Wed Jan 30 16:53:25 EST 2008


Arlindo, with version 2 we are taking a long term view of GrADS and
not focusing on short term issues.  It is our judgement that over the
next 5 years and beyond it will be less work overall for us to re-
implement the data output capability of GrADS rather than to go
forward with the prior implementation.  This decision is based on a
number of factors, not all of them technical.  While it is desirable
to maintain upward compatibility, we of course view an entirely new
version of GrADS as an opportunity to put GrADS on the right track
for the future.  We must work towards an extremely stable code base.
We cannot keep patching and putting out fires and still have time to
implement the new features that COLA and others need -- and COLA
needs rapid improvements to GrADS on a number of fronts.  The
ensemble dimension is just the first major item.  The many new data
formats already released or in the pipeline have already been
discussed and these are external forcings we have no control over.
We have warned you a number of times over the last couple years of
our intention to come out with a new version that will change many
things including lats, and this code will continue to have a rapid
rate of improvement for another year or two and perhaps longer.
GrADS is first and foremost a tool for COLA and the research
community to meet strategic research goals going forward.  If the
situation were different and we had more staff and resources perhaps
we would have been able to support more legacy code.  But we must
face the reality of what we have and fit our project into those limits.

And as already stated, we have not yet had time to re-implement the
data output feature, since we have released v2 somewhat earlier than
we had intended due to NCEP's grib2 timeline.  We have ben working
very hard ever since NCEP announced their changeover date to get v2
to the best state possible for that deadline... Brian

On Jan 30, 2008, at 11:30 AM, Arlindo da Silva wrote:

> On Jan 30, 2008 9:37 AM, Jennifer Adams <jma at cola.iges.org> wrote:
> We know that the loss of the LATS interface severely impacts many
> users. But our intention is to *replace* it with something better,
> not to eliminate the functionality completely. Consider this:
>
> set x; set y; set z; set t; set e
> set gxout fwrite
> set fwrite -nc file.nc
> d var
> disable fwrite
>
> Actually, if the "fwrite -nc"  feature is implemented you could
> still use lats4d for avoid having to write a script. It is
> important to understand that while lats4d started as a wrapper
> around LATS, it has evolved beyond LATS. For example, if you enter:
>
> lats4d -i myfile.ctl -o example -format sequential
>
> it would write the f77 binary file "example.bin". Other options are
> "stream" for plain binary. The options "-be" and "-le" have also
> been added specify if you want your binary file to be little_endian
> or big_endian.  And of course, you can use all the subsetting, -
> func, etc to customize what gets written. The sequential/stream
> formats were written around the current "fwrite" feature, so any
> enhancement of fwrite could be easily incorporated. One feature
> that is missing and that could be easily incorporated is for lats4d
> to write a companion .ctl for the binary files it writes.
>
> Recently I've be using lats4d as a generic iterator over a dataset.
> For example,
>
> lats4d -i myfile.ctl -format myscript.gs
>
> will call my script for each time/level/variable, so you can use
> the lats4d command line options for space/time/variable subsetting,
> and then run your little script on each chunk. A variant of this is
>
> lats4d -i myfile.ctl -format myscript.gsf
>
> where "myscript.gsf" is treated as a script library, therefore
> having access to the lats4d.gs namespace.
>
> And finally, a new feature that people may not be aware of is
>
> lats4d -i myfile.ctl -format stats
>
> which prints an ASCII table with min/max/ave/etc for each variable/
> level/time. This is a quick way to examine what is on a dataset.
>
> I am attaching my latest lats4d.gs to save people the trouble of go
> looking for it. It should be in the 1.9.0-rc1 build as well.
>
>      Arlindo
>
>
>
>
>
>
>
>
>
>
>
>
> --
> Arlindo da Silva
> dasilva at alum.mit.edu<lats4d.gs><lats4d><user.gs><user.gsf>



More information about the gradsusr mailing list