Am 26.08.24 um 17:18 schrieb Cottrell, Allin: Thank you for your detailed explanation of the current behavior and its rationale. I appreciate the historical reasons behind this design choice. However, I'd like to offer a different perspective on this matter. In my view, empty columns should not be automatically removed, as this responsibility should lie with the user. In my specific use case, I was quite surprised and confused by the differences in column numbers across various datasets. Let me explain my application: Each dataset represents a different variable and contains time series for various countries. I iterate over these datasets to vertically concatenate the time series (using the stack() function) to construct panel series, which are later joined together. At some point, I noticed that the dimensions weren't adding up correctly. I was aware of missing values in the data but was perplexed when the dimensions no longer matched. This automatic removal of empty columns led to unexpected behavior in my workflow and caused confusion. It took me some time to identify the source of the discrepancy, which could have been avoided if the original structure of the data had been preserved. While I understand the intention to save memory by removing empty trailing columns, I believe this approach can lead to unintended consequences and potential data integrity issues for users who rely on consistent column structures across their datasets. Perhaps a compromise could be to keep the current behavior but add an option to preserve all columns, even if empty. This way, users who need the original structure can maintain it, while those who prefer the memory-saving approach can continue to use the current default behavior. What are your thoughts on this suggestion? I'm open to further discussion on this topic. Best, Artur

Am 30.08.24 um 17:17 schrieb Cottrell, Allin: Good idea. Done: https://sourceforge.net/p/gretl/feature-requests/216/ Artur