Last note on the impossible PDI pivot

Here is a blog post by Matt Casters, the founder of Kettle (which was rebranded as Pentata Data Integration). In it he is un-pivoting data. That is to say, doing the opposite of what I did in the last two posts. I solved the problem of knowing how many columns to create by hard coding 60. He uses a more elegant solution of analyzing the data in one transformation and then using the results of the analysis in a second one and using the metadata injection step. I suspect that following his lead I could also use the metadata injection step and a separate analysis transformation to dynamically set the number of columns. But first I have to study the metadata injection step to see how it works.