I regularly use the CDK connectivity node in order to split molecule mixture (e.g. salts ions, etc.). Generally this node works well and it's very useful. Nevertheless I found a problem while I was using it in a large database. There are some molecular items (52) for which, even if they're composed by distinct connected molecules, the connectivity node gives as output just one molecule (see the attached workflow).
Do you know why this happens? Can it be considered a bug?
thanks for your post. You are right, that was a bug.
The CDK connectivity node split the molecules correctly but then collected them as set. Because the molecular entities in your example are identifical in each set (ignoring stereochemistry), 'duplicate' entries were removed. That's why just one molecule was added to the final output.
I have corrected that behavior. The molecules are now collected in a list and not dereplicated. I get the correct output with your example workflow.
Please update your nightly version for this change.