when an algorithm does not response, any gracefully way to stop?

[GoogleCodeExporter]

What steps will reproduce the problem?
1. start a model selection on a big set
2. click run
3.

What is the expected output? What do you see instead?
If there is an issue, say it takes too long, and the user want to stop the 
algorithm.  Anyway to do that without closing the entire application?

Please use labels and text to provide additional information.
For 3rd party algorithms, it may be hard if the algorithm already started.
But what about others? Maybe add an interface method to check if need to stop 
during a long computation, and add the button correspondingly.

Original issue reported on code.google.com by jzho...@gmail.com on 14 Sep 2012 at 1:51

[GoogleCodeExporter]

A more important issue is to add the "stop" button at the "multiple comparison" 
screen.

If many chains have been created, and it takes too long to finish (in user's 
eyes), the use can click the stop button, and s/he will be prompted to save the 
created chains to a file,  and be reminded that it does take a while for some 
algorithm to finish running.  If the user still chooses to stop, just close the 
Frame. (-- The memory should be gced, based on current impl. in event-handling.)

Original comment by jzho...@gmail.com on 14 Feb 2013 at 6:40

• Added labels: Priority-High, Usability
• Removed labels: Priority-Low

[GoogleCodeExporter]

Assigned to Eric

Original comment by jzho...@gmail.com on 14 Feb 2013 at 6:41

[GoogleCodeExporter]

Added a "stop" button in multi-comparison mode. 

Currently if the chain has multiple algorithms, the thread is stopped after an 
algorithm is done. Ideally, it should also stop between cross-validation runs 
-- need to modify Validator

Original comment by jzho...@gmail.com on 4 Apr 2013 at 5:36

• Changed state: Started

[GoogleCodeExporter]

Additionally,  I can add a stop() method to the interafaces for the algorithms, 
so that the algorithm can be stopped in middle.  

For example,  if BPNN is in the middle of training, a stop method should be 
able to stop the training (possibly in a different thread).

Original comment by jzho...@gmail.com on 16 Aug 2013 at 8:08

[GoogleCodeExporter]

Fixed in issue #63.  Currently the approach is the kill the thread that does 
the model selection.

Original comment by jzho...@gmail.com on 11 Nov 2013 at 10:43

• Changed state: Fixed