feat: Exported CSV rows are ordered newest to oldest. Can we add an oldest to newest option?

[thanoskn]

Problem

When you export your results to CSV, rows follow this order: newest to oldest. This causes some issues.

• It breaks ML models that need chronological ordering (e.g. Linear Regression).
• It's relatively unconventional.

Proposed fixes

1. Change the default order on the CSV file to: oldest to newest. This is the quickest fix but it may break systems that rely on the current ordering.

2. Add a switch or an option in the settings where the user chooses the row order. This fix ensures backwards compatibility while still fixing the issue.

Conclusion

I suggest implementing option 2. Although it requires more efford I think it's worth it. Let me know what you think so that I can start working on the pull request.

[fehmer]

docker run --rm -v "$PWD":/data alpine sh -c '
awk '\''NR==1{print; next} {a[NR]=$0} END{for(i=NR;i>1;i--) print a[i]}'\'' /data/results.csv > /data/results-reversed.csv
'

[thanoskn]

Thanks for the repsonse. I appreciate your help and actually I am already doing something similar. But my original question refers on how should we implement a fix for the problem. Specifically whether we should choose option 1 or 2. From your reply it seems like your suggesting we don't change anything in monkeytype and instead leave it to the users to handle it themselves. Is that what you meant? Just want to make sure I understand your position before moving forward.

[fehmer]

Hi,
As you mentioned, option 1 potentially breaks other people’s workflows.

Option 2 would clutter the UI for a very specific use case. In my opinion, the CSV download already serves as a way to get data out of the site for custom processing. I assume nobody is reading CSV files in a text editor. Users will either open them in a spreadsheet tool or process them programmatically, and in both cases they can handle sorting themselves.

If we were to add only a flag for the sort direction of the timestamp field, the next question would be: why limit sorting to the timestamp at all? Someone might want to sort by WPM, acc, or con instead. Or they might want to choose which columns to export and in which order. This quickly leads to a complex export settings popup, followed by the need to permanently store all those export preferences.

In summary, I don’t think adding just a sort order for the timestamp is necessary, and a full-blown export settings feature would be too convoluted and require significant development effort. It also doesn’t contribute to the core functionality of the site.