lidavidm opened a new issue, #33999:
URL: https://github.com/apache/arrow/issues/33999
### Describe the bug, including details regarding any error messages,
version, and platform.
String.Value slices a backing string:
```go
func (a *String) Value(i int) string {
i = i + a.array.data.offset
return a.values[a.offsets[i]:a.offsets[i+1]]
}
```
Which points directly into the buffer data:
```go
if vdata := data.buffers[2]; vdata != nil {
b := vdata.Bytes()
a.values = *(*string)(unsafe.Pointer(&b))
}
```
This seems OK when the buffer is Go-allocated since the GC will take care of
it (regardless of Release/Retain). But when the buffer is C-allocated (e.g. C
Data Interface, mallocator), if the string outlives the array, we have a
dangling pointer!
I believe we should copy the string (and frankly, get rid of the unsafe
here) to be safe (which also has the side effect that getting a single value
out of an array will no longer keep the entire backing array alive). If the
extra garbage is an issue, we should consider some sort of string view type.
### Component(s)
Go
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
