Having 'undo' in REPORT is great, and would be just as great in VIEW! Removing and adding channels from a 2D axis would be my favorite use of an "undo" function.
If you want a basic version of Undo for a 2D Axis System using a User command looking at the event "View.Events.OnCursorChanged" and recording and stores the XBegin and XEnd of the 2D Axis System.
It would be nice if we could more View.Events to work with in the future. One i personally want is something along the lines of a OnRangeChanged which would look at XBegin and XEnd so an event would happen each time the 2D Axis Systems x range is shifted or focused or unzoomed.
I'm just getting started with DIAdem, and am honestly *floored* that Ctrl-Z (undo) isn't available for VIEW and ANALYSIS. I figure there must be a reason, but can't fathom why...
It seems like this seriously inhibits exploratory analysis of data sets....
Yeah, it is very surprising when you first run into it. The core issue is that DIAdem is frequently used to manipulate extremely large data sets. Allowing a user undo actions that change the values in Data Portal channels (like copying out a flagged region in VIEW or running virtually any ANALYSIS dialog) would require keeping multiple versions of the Data Portal. In REPORT there are 10 steps of undo. If that were the case for the Data Portal, we'd have to keep 10 separate versions of all the channels in the Data Portal or track a change tree that could include any subset of those Data Portal channels.
I agree with Julia that configuration changes in VIEW would not be unduly expensive to track and would add a lot of value. It would be awkward to explain why parts of VIEW have undo and other parts don't, but I still think it would be worth it.
Alternatively, I could imagine having a VIEW and/or ANALYSIS undo functionality that could be turned on or off by user, like the auto-calculate state in Excel. But then which state would be the default installed by this future version of DIAdem, the state that means nobody notices (like unit management, turned OFF) or the state that means DIAdem runs out of memory as soon as you load a huge data state and make 10 changes to it (turned ON)?
We've discussed this periodically, but it keeps being more complicated and less bounded than R&D likes.
