You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think we probably want to get some broader thoughts on this, whilst I agree unsigned is likely more appropriate in the Rust ecosystem, lots of systems, arrow included, prefer signed quantities for compatibility reasons (see StringArray, ListArray, etc...)
Either way, this isn't a bug and would be a breaking change
Describe the bug
The string array functions
length()
andbit_length()
should not returnlength
as signed integer.To Reproduce
See
arrow-rs/arrow-string/src/length.rs
Line 47 in 5ad621f
arrow-rs/arrow-string/src/length.rs
Line 111 in 5ad621f
Expected behavior
Should return unsigned integer, since there are no negative length in our context.
Additional context
Thanks for @alamb bringing this up in the discussion on #6671 (comment)
Note: The actual Arrow spec uses integer types.
The text was updated successfully, but these errors were encountered: