20000719: Lyndon State model data display Tcl/Tk GUI (cont.)

>From: Bill Fingerhut <address@hidden>
>Organization: Lyndon State
>Keywords: 200007131522.e6DFMAT21195 McIDAS-X Tcl/Tk GRID interface

Bill,

>I use a BATCH file because I know BATCH much better and 
>longer than Tcl/Tk.

OK.  I figured as much.

>I found something in the BATCH file that I did not like
>and cleaned it up. Would you like me to forward it?

Yes.

>From address@hidden Wed Jul 19 12:45:39 2000

>If I request an ADDE dataset, then a minute later request
>the same dataset, the data are requested again as thought it
>was a first request?

Do you mean that you ask for data from a particular dataset and then
request it again?

>It would be much more efficient if data
>hung around for a while. Any comments?

The caching mechanism of Unix/Linux pretty much insures this.  If the
server is not swapping memory continuously, then a second request to a
dataserver should be faster than the first.  As far as implementing
other types of data caching, I could point out that there could be a
LOT of requests to the same server for different pieces of the same or
different datasets.  Trying to cache all of this would be wasteful of
system resources (and hard to implement).

On a totally different subject, I have noticed that Lyndon has not
downloaded the new ldm-mcidas distribution.  The manner in which the
images in the Unidata-Wisconsin datastream (MCIDAS data type in the
LDM) is slated to change at the end of the month.  I gave sites a
month's notice to get install and play with the new code so that they
could be ready for the change, but the _great_ majority (like Lyndon)
have not taken the opportunity of installing the new decoder and
testing it on the CIMSS images that were added to the UW datastream at
the end of June.  I figured that sites could get used to using the new
decoder's capabilities with the CIMSS images and be prepared when the
rest of the images get changed.  You might want to alert Mark Tucker
about this!

Tom