Omega359 commented on PR #20604: URL: https://github.com/apache/datafusion/pull/20604#issuecomment-3980125922
> Maybe I'm missing something, but what is the actual advantage of returning a `Utf8View` from a UDF like reverse? > > For functions like `trim` or `substr` it makes sense to return a `Utf8View` regardless of the input type, as it allows for reuse of the underlying string buffer, given that each output value is just a slice of existing string data. But for a function like `reverse` that creates brand new string values, surely returning `Utf8View` just adds extra overhead with no benefit? In a word: consistency. What you outline is true if you look at the function in isolation. For my use case where I use DataFusion in an ETL environment it's rarely just one function being applied to a column, rather it's a chain of them. I've been attempting to enabled Utf8View in my app for the better part of the last month and I'm continuing to encounter issues where df calls are emitting utf8view in one case and utf8 in another. Works fine till you try and merge the schemas and then .. boom. Finding all of the instances where something is consuming utf8view and emitting utf8 and throwing a cast after it is tiresome and expensive both in my time and actual test runs. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
