I came across some bizarre behavior yesterday that I'd never noticed
before. I'd be interested if anyone tries it and confirms it. This is
on the SGI 4.1.1. build but I doubt that has any significance.
I feed a field into Compute. The field has normal positions,
connections, data, colors, and invalid positions (byte array style).
Since upstream I have allowed the user a choice of "data" and the
ranges are quite different for different data types, and I want them
all to come in Glyph with the same size, the Compute statement is
simply "1", i.e. force all data to take the scalar float value "1"
(no quotes of course). This works as expected.
Lo and behold, the invalid positions disappear as a result of
performing this statement! WHA??
This leads to the display of bogus data that had been masked off by
these invalids.
So after fuming, checking docs, finding nothing, I change the statement to:
"a/a"
Now, no problem: invalids pass through unchanged as expected and
naturally, the data is all "1".
I think that's weird. What do you think? I did try adding a "b" input
and feeding a "1" into b, field into a, and making the statement "b".
Invalids are stripped there as well. I just can't fathom a rationale
for having an exception statement in Compute that performs such a
profound and sneaky restructuring. Yuck!
Chris Pelkie
Vice President/Scientific Visualization Producer
Conceptual Reality Presentations, Inc.
30 West Meadow Drive
Ithaca, NY 14850
[EMAIL PROTECTED]
