New package: GNU Datamash

lists.gnu.org

13 pointstaeric3 comments

Looks nice, but I foresee usability issues with that -H flag.

The last two examples on http://www.gnu.org/software/datamash/examples/ seem to indicate that -H adds output headers, but it also requires input headers, and there is no way for the program to discriminate a header line from a sample. So, if your FILE.TXT doesn't have headers, the tool will silently ignore the first sample in the input file.

I had to think about this for a bit. I'm not entirely sure how you could fix this. Best I can do is the system could test if the first row has "numeric" data and give a warning if so. Are there smarter ways?

Excellent looking tool. Hope to test it out with some large datasets and see how well it does.

Reply on news.ycombinator.com