You're right, plot cares only about validity of positions. In
fact there is no concept in dx of "invalid data"; only of
"invalid positions" and "invalid connections". The data
implicitly derives its validity from that of the component on
which it is dependant.

The way to do what you want is to pass the invalidated field
through Compute with the expression "invalid(a)". This
will make an array of 1's and 0's corresponding to the validity
of the corresponding position. Then you can "and" or "or" the result
of the two arrays to do what you want. Make sure that you
set the component to have a "dep" attribute of "positions" in
the final field.

_________________________________
Donna L. Gresh, Ph.D.
Visual Analysis Group
IBM T.J. Watson Research Center
(914) 784-5049
[EMAIL PROTECTED]


"Sharon Cady" <[EMAIL PROTECTED]>@opendx.watson.ibm.com on
01/16/2001 03:18:12 PM

Please respond to [email protected]

Sent by:  [EMAIL PROTECTED]


To:   [email protected]
cc:
Subject:  [opendx-users] invalid positions array type



Anybody know of a way to force the invalid positions array which is
obtained by
Extract to always be given as bytes rather than (which is sometimes the
case)
integer ? I am doing x/y plots of two fields, and want to plot only points
that
are valid in both x and y. I believe that the inherent nature of the Plot
module is to plot points which are valid in x ("positions"), with no regard
to
validity in y ("data").

So, I am tinkering with invalid positions arrays by extracting invalid
positions of each of 2 fields and then (assuming they are both type byte)
doing
a logical or operation to change some "0"s to "1"s. I then Replace the
result
as invalid positions into the field which is then plotted.

Obviously, this will not work if the arrays are of different types and of
different lengths. Also, the form of the type int array that is returned
(the
actual positions of the invalid points, or something like that) seems to be
of
little use in this case. I'm aware that dx chooses which data type to
return
depending on the number of invalids, because of memory issues. But this can
sure make some things unduly painful, no ?

Someone mentioned that it took 10 hrs for her message to be posted.
Although I
haven't really timed it, the last few messages that I've posted (a few
weeks
ago) seemed to take more like 24 hrs to register. Why so slow ? I don't
recall
this being an issue in the past.


Reply via email to