Chris Pelkie wrote:
There is a history to this mess, probably found by searching the archives. Roughly, Greg tried to clean up the code resulting in the 4.3.0 situation. That immediately broke something I had developed years back that assumed that invalids would 'pass through' without side-effects in certain cases. So Greg and/or David fixed and/or rolled-back the patch to make 4.3.2. In my case, I was creating a bit mask and AND/ing or OR/ing it with another to create a joint 'invalids' array. But the 4.2.0 patch was disabling my nice new array before it exited the Compute or the Replace (I forget which), so I could never get my resultant field back the way it had worked prior (and now works again).
Thanks! I've followed the thread http://opendx.npaci.edu/mail/opendx-users/2003.08/msg00040.html and looked at Import and Export documentation. However, I was not able were we document that an unreferenced position is invalid, and I still think (perhaps G.D. Abram at the end of the thread) that this is odd, and that the user should be able to choose the desired behavior. Anyway, let's live with it if it is not easy to change and/or would break some nets. Thanks again, Marco
