diff showing different results for the same input

[caeu]

Having two files f1.csv and f2.csv. This seems to happens when the first file has duplicates in the key column.

❯  cat f1.csv
c1,c2
a,1
a,2
b,3
c,4
b,5

❯  cat f2.csv
c1,c2
a,1
b,3
c,4

❯ qsv diff f1.csv f2.csv
diffresult,c1,c2

❯ qsv diff f1.csv f2.csv
diffresult,c1,c2

❯ qsv diff f1.csv f2.csv
diffresult,c1,c2

❯ qsv diff f1.csv f2.csv
diffresult,c1,c2
-,a,2
+,a,1
-,b,5
+,b,3

❯ qsv diff f1.csv f2.csv
diffresult,c1,c2

❯ qsv diff f1.csv f2.csv
diffresult,c1,c2
-,a,2

❯ qsv diff f1.csv f2.csv
diffresult,c1,c2
-,a,2
-,b,5
+,b,3

❯ qsv --version
qsv 2.2.1-mimalloc-apply;fetch;foreach;geocode;Luau 0.653;to;polars-0.46.0:release;self_update-14-14;28.80 GiB-1.03 GiB-0 B-36.00 GiB (aarch64-apple-darwin compiled with Rust 1.84) prebuilt

Platform: Macbook M3
OS: 15.3

[datatraveller1]

Hi @caeu,

Yes, the primary keys values must be unique per csv file, see comment here:
#2443 (comment)

[datatraveller1]

@janriemer
Maybe it is interesting how the new Kani model checker behaves with the example above.

[caeu]

@datatraveller1 Thanks!

I only searched under open issues for similar issue.

Then perhaps this should be documented in the help of diff

Closing as duplicate of #2443 (comment)

[datatraveller1]

Then perhaps this should be documented in the help of diff

Yes, and it would be nice if diff could output something like "duplicate primary keys found" and exit with error.