Deleting rows from linked data shouldn't be this hard! (SOLVED)

[Craig Carlson]

I have a table created from linking in numerous CSV's. When I mark data in a scatterplot by selecting individual points and filter out, it won't let me create a filter transformation because the filter is of type "mask". I tried changing the mask column name to something else and that didn't help. The only solution I can find for saving all my work of deleting individual points is to switch to embedded. However this kills all the value of the data canvas. I can't go replace any of the source CSV's. It shouldn't be this difficult or counterintuitive to be able to just drag and delete some points and have the choice of it being remembered on next load. I am OK with the idea of this sort of filtering being a block in the data canvas, but I can't figure out how to do that except with an equation.

[Craig Carlson]

My TIBCO support ticket didn't give me a useableanswer, so I finally solved this on my own.

 

Data -->Data Table Properties -->Key columns for linked data -->Edit

 

Limit available columns to: -->All columns (if needed)

Select your unique columns (more on this below).

OK-->apply -->OK

 

 

View -->Tags

 

New Tag Collection (icon for plus sign over multiple files)

Create Tag Collection Points to Exclude, and new tag Excluded Points

 

 

Filter

 

Mark some points (note that ALT changes box selection to lasso!).

Right click on the Excluded Points tag -->Attach Tag to Marked Rows.

File -->Save.

 

 

The tag Points to Exclude is now a persistently available filter, and you can continually add to it as you link new data that needs manual filtering.

 

Don't blame me that the cut and paste override my logical outline structure above: (1.a) with (1.1).

Regarding #1. I am not sure why Spotfire didn't consider a "time" formatted column to be valid, but once I overrode it (not considering it valid), it seemed to work. Also, even though I knew it to be horribly inefficient, at one point I selected every column in the table. It allowed me to do this without warning, but then when I saved and reopened it didn't save my tags. When I selected fewer key columns it did work.

I find it really disappointing that something as simple as selecting points you want to persistently filter out would take this many steps, and pretty unforgivable that if you do a lot of marking and filtering, it gives you zero warning that you will lose your work due to the nature of the temporary "mask" column that is created. Similarly, when you tag data, if you haven't selected the key columns for your linked data, you will lose those upon closing, again without warning.