`get-data-table-from-csv-list` appears to "sample" by truncating rows instead of selecting `sample` rows

[atgreen]

I realize that this fn in undocumented, but

                    (if sample
			(map-fn (subseq x 0 sample))
                        (map-fn x)))

is not what I'm expecting it to do when sample is provided.

[sebasmonia]

Similarly, consider these two implementations:

(defun first-five-rows (file-path &key (skip-first t) (separator #\,))
  (let ((counter 0)
        rows)
    (cl-csv:do-csv (row file-path :skip-first-p skip-first
                                  :newline #\Newline
                                  :separator separator)
      (push row rows)
      (incf counter)
      (when (> counter 4)
        (return-from first-five-rows (nreverse rows))))))

(defun first-five-rows (file-path &key (skip-first t) (separator #\,))
  (cl-csv:read-csv file-path :skip-first-p skip-first :sample 5 :separator separator))

The first version is much faster than the second one. When I found out that there was a built-in :sample parameter, I thought it would be faster than my naive approach, but that's not the case at all (for a 35 MB CSV).

[bobbysmith007]

The sample in cl-csv is a random sample from the entire dataset, rather than the first 5 rows of the dataset. As you have to read the entire dataset to randomly sample from it, I would expect that the sample clause would be much slower than reading the first 5 rows.

I have not worked on CL-CSV or any common lisp in a long time. I am happy for someone to branch this project, or take over as maintainer, but its difficult for me to even open the projects at this point.

---

As to the original comment... I agree that the sample in get-data-table-from-csv-list does not match the word sample as used in the rest of the library, nor does it seem to behave the way I would expect, that should probably be removed as an argument

[sebasmonia]

It seems I completely misunderstood/misread the :sample parameter, thanks for clarifying!

And regarding the second part, I don't think I am much qualified to take over as maintainer, but thanks for making cl-csv a thing in the first place :)