Why is the Datafilter in JMP so slow?

[#1]

For a data table with more than 1 Mio rows, the data filter is extremely slow.
It takes several seconds to 
- activate a new column in the data filter
- type text in the filter : a few seconds per character!
- delete a character in the filter

Why?

new table("test", add rows(2000000), new column ("col",Character,  Set each value(Char(random integer(2000000)))))

[#2]

Hi @hogi, Thanks for reporting and taking the time to show the latency issue you are experiencing here.  We are aware of this issue and are working to make performance improvements in this case for tables with larger amounts of rows in future JMP releases.  This will probably be a multi-release effort (performance improvement will take time and testing to ensure we don't have any regressions) so please manage expectations accordingly. Thank you! 

JMP Technical Support.

[#3 · Accepted solution]

Sounds great.

Interesting: It's the number of entries in the Data Filter.
If you have an 'index' column with more than one million entries, you can be sure to encounter speed issues when using this index in the data filter.
There is no issue for a data table with just 20000 entries in the data filter (same number of rows in the table):

new table("test", add rows(2000000), new column ("col",Character,  Set each value(Char(random integer(20000            // was: 2000000
)))))

Interesting² and surprising:
It helps to  use a pre-selection via an auxiliary column. This can be used as a workaround till the issue gets fixed:
use a second column with a shortened version of the original column. 

new table("test", add rows(2000000), new column ("col_long",Character,  Set each value(Char(random integer(2000000)))),
	new column ("col_short",Character, set each value(left(:col_long,3)))
)

It's definitely hard to get used to, but it's much faster than using a single column:

["54321" is typed at the same speed with "conditional" switched on and off]