When I use the Cell Splitter node in a loop something strange happens.
In practice, some cells are empty. My delimiter is “;” so there are situations like “1;;” for example. I then expect 3 new columns but only 2 are created so the loop gives an error.
If I manually set the Cell Splitter noda to 3 columns then it works fine.
another little question . . . the Chunk Loop Start requires the type “string” Therefore I use the String Manipulation (Multi column) node to convert to the type string.
However, this node / process makes the flow verry, very slow. How can you fasten that?
Thats not the case for the above mentioned flow because its very small. On my real live actual dataset it takes 15-30min
Why don’t you transform the string after the loop? this will runs at once and can bring some speedy for you. When you are inside the loop, just make the split as you need, less process inside the loop will let more memory to make others stuffs.
you’re fully right . . . however, it was my understanding that to run the loop the type string was required. So, that’s why included the conversion to type string within the loop.
But when I set “Allow changing table specifications” (see above cq @HansS his tip) the type string isn’t needed anymore. And this brings down the processing time by 60%