I often use Reference Row Filter and Reference Row Splitter nodes. And I often miss an option to compare multiple columns. Reference Row Filter & Splitter nodes make it possible to filter a table using a reference table. You can choose a pair of columns and the node filter table rows if the values are equal.
Nice, convenient, helpful. But it could get even better if it was possible to pick some additional columns. What about making it similar to the Joiner node?
I agree with you in having the option to filter rows based on several columns. But you could also use the Rule-based Row Filter (Dictionary) node for this.
You can create rules based on a reference table and filter or split rows based on those rules. Would you like me to provide you with an example workflow?
Yes, you’re right in most cases. But you wouldn’t achieve the option Joiner node gives you. You can’t simulate this way the behaviour “Match any of the following”.
More over. Well, you can usually workaround almost everything. There is Java Snippet node for case you run out of wisdom. But if your workflow is really complex and you work with considerably large amounts of data it is always better avoid a need to prepare data for the logic you employ and revoke your changes after the logic finishes its job.
Rule Based Row Splitter Dictionary.knwf (51.5 KB)
Well, you can achieve the some result as if you would get if Reference Row Splitter or Filter node allowed to search multiple columns.
Two disadvantages:
First, it’s really inconvenient, tricky and easy to make it wrong. Second, the performance is miserable especially if the table containing the rules grows up to thousands of rows. Which is what I usually face and which is why these workarounds often don’t help.