GroupByLoopStart

Groups BioInformatics
1

 

Hi

Thanks a lot for the very useful [GroupByLoopStart] node . To my eyes, this node could be included in the standard knime distribution. It  is working fine for me in most of the time. But sometime I encounter this error message that stop my process:

Unable to merge flow object stacks: Stack can't be merged: Conflicting FlowObjects:org.knime.core.node.workflow.FlowLoopContext@18225f9 vs. org.knime.core.node.workflow.FlowLoopContext@18b0b84

The problem comes from the fact that when you use a lot of [Java IF] node without [End IF] node the GroupByLoopStart node gets confused.

I create a surrogate workflow that solve this issue in my case, but I find your solution more elegant and simple. Do you mind to have a look at this problem?

In advance thanks for you help in finding a fix.

Frederico

2

Hi,

sorry I was very sick the last week and didn't pay much attention to this group. I will have a look at your problem on Monday...

Best,

Bernd

3

Hi Frederico,

I had a look at the workflow you sent and have to admit that I don't quite understand your problem.

I haven't been using any if/else nodes without the corresponding end nodes. I believe that KNIME internally keeps track of which loops are open and has a stack that it works with. So if you have a node (e.g. if/else) after the loop start then you probably need to first close the if/else before being able to close the loop. This should be independant of the loop node you are using.

Can you reproduce the behavior you are describing only with the groupbyloopstart node or with other loop starter nodes as well? 

Thanks for clarifying,

 

Bernd

4

 

Hi Bernd,

Thanks for your replay. In fact I still encounter this issue, only with the GroupByLoopStart node: If I use Chunk Loop Start I do not have this error message:

WARN  WorkflowManager Unable to merge flow object stacks: Stack can't be merged: Conflicting FlowObjects:org.knime.core.node.workflow.FlowLoopContext@567513 vs. org.knime.core.node.workflow.FlowLoopContext@203bcc

Infact, the same workflow can produce the above error and still be able to carry-on the execution when [execute all nodes’] is pressed or can just run perfectly without producing any errors.

Please find here by the smallest workflow I could create that reproduce the error. I hope that could help you further.

To reproduce the error you may need to reset table creator and then press Shift+F7 (run all nodes). Also if you expend the metanode the error disappear???

Again thanks for you help

Cheers

Frederico

 

 

 

 
5

Hi Frederico,

just to let you know that the problem has been escalated to the KNIME people and they can reproduce the problem, too.

B

6

This has been fixed and will be released with our next bug fix release in early March. It was really a problem of a loops - when loading workflows saved with all data and executing only the loop start before re-execution the loop-objects were not identified correctly.

For now making sure you reset _all_ nodes in the loop before executing them should help. Or don't save executed workflows...

Thanks for reporting this and submitting such clean test cases!

Michael